Also:

• The numbers behind The AI Ad Economy: There’s a gap between consumer willingness to use AI for holiday shopping and retailer readiness. Maybe we call it the “HolidAI”gap?

• New! What I’m Reading This Week: I’m trying something new in place of my usual news roundup. This week: Some terrific articles about the risks of AI backlash, Amazon’s sponsored prompt patent and how to figure out your AI search maturity. Like this new format? Hate it? Let me know what you think.

THE MAIN STORY

What Will ChatGPT Advertising Grow Up to Be?

Six months into its ad business, OpenAI is selling ads in 40 markets. It has a growing array of ad performance, targeting and measurement tools (and a growing list of adtech partners). And if the 7,000 people who joined a ChatGPT ads how-to webinar last week is any indication, a lot of businesses want to be there.

But in the same period, the company has made only incremental tweaks to what the ads look like. Big advertisers are still hesitant to commit consistent spending, especially in North America, a crucial advertising market. And building a sustainable ad business is taking on new importance as the company faces extensive criticism ahead of a planned IPO.

The upshot: OpenAI’s ad business is growing up fast, but the company is still figuring out what that business wants to be when it grows up.

Here are four takeaways from OpenAI’s first six months in the advertising business and three things the marketing community should watch in the next six months.

Takeaway 1: Advertising matters a lot more now to OpenAI

In a new cover story for Time about OpenAI, journalist Alex Heath chronicles several existential challenges facing OpenAI, including expensive infrastructure and ballooning costs, competitive pressure from Anthropic and senior executive departures.

Against that backdrop, and with an IPO on the way in 2027 or sooner, a sustainable advertising business is now a top goal for OpenAI. The company needs a scalable revenue stream that can monetize its enormous free consumer audience.

It’s already been widely reported that OpenAI aims to do what no other digital company has ever done: become a $100 billion annual ad business in just five years. But the urgency around building that business has intensified in recent months. It’s building an ad sales team, expanding to new markets, adding back-end tools like self-serve buying, CPC bidding, and expanded reporting, and striking adtech partnerships all at once.

It’s also showing more ads: According to Similarweb’s 2026 Generative AI Landscape Report, 26% of US desktop ChatGPT responses included ads in June, up from 14% in May.

As OpenAI moves on so many advertising fronts, the question becomes: Can the ad business scale and become a bright spot amid the company’s broader challenges?

Takeaway 2: OpenAI knows what’s different about ChatGPT. But it hasn’t figured out how to sell ads against it

OpenAI execs are increasingly discussing ChatGPT ads in the context of the space between discovery and conversion—the middle of the funnel. It’s a stage that “historically has been fragmented, where people are exploring possibilities, they’re comparing alternatives, understanding trade-offs,” OpenAI’s VP of Monetization Benji Shomair said during a recent webinar with StackAdapt, one of OpenAI’s adtech partners. “We believe this creates a fundamentally new channel for marketers.”

I’ve written and spoken a lot about AI’s power to introduce people to products they didn’t know existed or needs they hadn’t recognized. But the challenge with paid ads in AI is twofold: The ad system needs to recognize the right moment to deliver an ad, and the advertiser needs creative suited to that moment.

If someone asks ChatGPT to troubleshoot an error message on their washing machine, the ads system needs to know whether the person simply wants an answer, is considering buying a new machine or needs to make a service request. OpenAI isn’t there yet.

OpenAI recognizes this. “The first time we ran ads in the wild externally, we bootstrapped some of the demand by telling people, ‘Hey, don’t worry about creating new ad formats for us. Just give us some of your search ads, and we’ll form fit it into our new format and we’ll see how they perform. And the truth is, is they didn’t perform that well,” Shomair said during the StackAdapt webinar.

Why? Search advertising matches to expressed intent. If someone is searching for a hotel in Brisbane, Australia, for a family trip, an Expedia ad for hotels in Brisbane is an obvious fit. But a conversational system must recognize and interpret that intent as it develops over the course of a chat. That’s could be a lot more powerful, but it is also harder to execute.

So what kinds of ads do belong in a chat-based conversation? One bright spot is a recent test spotted by OpenAI watcher Juozas Kaziukėnas: ads that launch business agents. Making ads as conversational as ChatGPT itself could be one direction OpenAI tests next: Ads as service, not just a message.

Takeaway 3: Advertisers want to invest in ChatGPT ads, but there’s no consensus about which budgets should fund them

Last week I joined more than 7,000 people for an OpenAI webinar called “The Decision is the Opportunity.” The main draw? Likely the offer of a $500 advertising credit to qualifying businesses that signed up.

Getting thousands of attendees for a single webinar indicates interest, granted. But once those ad credits are used up, it’s not clear which budgets should fund additional spending. So far, advertisers are using search or experimental budgets. But for the kind of growth OpenAI is expecting, they’re going to need a lot more than that.

In the StackAdapt webinar, CEO Vitaly Pecherskiy likened it to the early days of connected TV, when marketers couldn’t decide whether to use broadcast or digital budgets. Eventually, CTV found a home in digital budgets and spending rose. With ChatGPT, he said, “I could see a lot of marketing teams go through similar questions. Do we have this sit in our search, social, display teams? That naturally takes cycles to figure out.”

Finding an organizational owner is one thing; finding incremental budget is another. Budget ownership questions aren’t surprising for a new medium like ChatGPT, but the lack of consensus will slow spending growth and leave OpenAI unsure which teams to sell to.

Leave a comment

Takeaway 4: ChatGPT is the current focus for advertising. It isn’t the end game

ChatGPT is OpenAI’s most popular consumer product and the main focus of its ad business today. But to reach its ad revenue goals, it will need to have other ad surfaces.

After pulling back from standalone initiatives to develop Sora, its video app, and Atlas, its browser, the company is now focused on a future where ChatGPT is part of a larger system, not a standalone platform. That means today’s ChatGPT ads could end up being only a preview of the larger ad opportunity.

As Alex Heath describes in his Time article: “[Sam] Altman imagines a general-purpose AI subscription that dissolves the boundaries between ChatGPT, Codex, and work software. ... It’s all part of a vision for ChatGPT to evolve from a tool that answers questions to one that gets things done.”

Exactly when this agentic future will happen isn’t clear. Speaking about agentic commerce on the StackAdapt webinar, Shomair admitted, “None of us really know exactly how it’s going to develop. We’re all making educated guesses.”

What the marketing community should watch in the next six months

Can OpenAI create demand rather than just harvest it? If it can monetize commercial activity that wasn’t previously monetized—surfacing needs someone didn’t know they had and introducing solutions that generate transactions that might otherwise not have occurred—that could become a linchpin for future revenue growth. That will require new ad formats and new ways to measure the results. Last-click metrics inherited from search may not be the right way to measure the contribution of mid-funnel conversational advertising.

Can OpenAI successfully tap into ad budgets beyond search? If the company can prove that ChatGPT captures mid-funnel intent, it will also need to show advertisers how to plan and buy it, since it doesn’t fit neatly into an existing media category.

Can OpenAI crack the North American market? The region that’s home to some of the largest ad budgets may also be one of the most skeptical. Consumers and marketers in North America are less likely to view chatbot ads positively than counterparts in other parts of the world, a recent study by DoubleVerify found. Expanding into new markets will help, but winning over North American advertisers is crucial.

Final thoughts

Six months in, OpenAI has demonstrated it can build the infrastructure of an ad business quickly. The next six months will tell us whether it can develop ad products that are truly native to AI and unlock the budgets it needs in order to reach its revenue ambitions.

Read Next

Share

THE NUMBERS BEHIND THE AI AD ECONOMY

If there’s a sure sign the holidays are coming it’s Starbucks’ annual pumpkin spice latte launch this week. Another sure sign? Holiday shopping surveys. Narvar, a customer loyalty platform, found that while 65% of US online shoppers plan to use AI in some form this holiday season, only 8% of retailers consider themselves “very confident” in their use of AI.

WHAT I’M READING THIS WEEK

I’m trying something new this week. Instead of a list of news headlines, I’m sharing a few articles that caught my eye, on topics related to The AI Ad Economy. Like it? Let me know.

The Answer Economy

The Great AI Backlash Has Begun—and Brands Should Pay Attention

For three years, the industry has raced to put AI into everything…

Read more

9 days ago · 2 likes · Pete Blackshaw

BRXND Dispatch

The AI Search Maturity Matrix // BRXND Dispatch vol 132

Editor’s note from Mike: Today’s piece is a collab with our friend Tom Critchlow, forward deployed engineer at Alephic and founder of AI Search Leaders. It’s written in the singular first-person because we effectively became one brain on it…

Read more

11 days ago · 12 likes · 1 comment · Tom Critchlow and Mike Mallazzo

Billed for the Ad Before the Ad Exists: Amazon Sponsored Prompts

Retail Media Breakfast Club’s Kiri Masters reports on an Amazon patent that disrupts what it means to be an ad.

Subscribe now

WHAT’S DEBBIE UP TO?

Speaking and Conferences

Here’s where you can find me in September. It’s shaping up to be busy!

Ascendant Network RMN Americas Boot Camp, Sept. 16-17, 2026, Palo Alto CA. I’m conducting two sponsored focus groups with attendees and participating in other sessions. I’ll also be attending Ascendant’s Digital Ascendant on Sept. 15.

Path to Purchase Institute Learning Lab, Sept. 17, 2026 (virtual). I’ll be presenting “The AI-Powered Shopper Journey: How AI is Changing Shopper Engagement and Promotional Strategy.” Path to Purchase Institute members: Register here.

RetailClub AI Festival, Sept 24-26, 2026, Huntington Beach CA. I’ll be discussing “When Chat Becomes the Channel: Advertising in the Age of AI Assistants” with a community that’s focused on moving beyond AI hype to practical implementation and meaningful change across retail. Register to attend here.

Amazon unBoxed, Sept. 28-30, San Francisco. I’m excited to go to my first unBoxed. Got tips? Send them my way. Will you be there too? Let’s connect.

PSA: I have openings in October and November for keynotes, panels and moderating roles. Let’s talk! Here’s a clip from my keynote at Skai’s ShopAble conference.

Watch my speaking reel

Debra Aho Williamson is the Founder and Chief Analyst of Sonata Insights, where she helps advertisers navigate the next major structural shift in media: the rapid rise and monetization of AI platforms. Over nearly two decades as an analyst at EMARKETER, Debra was among the first to forecast the growth of social media advertising and identify how it would fundamentally reshape marketing investment.

Today, she applies that same pattern-recognition lens to AI, analyzing how consumer adoption is accelerating monetization and redefining brand budgets, media strategy and industry power dynamics. Through Sonata Insights, she partners with leading platforms, agencies and technology companies to deliver research, advisory and strategic narratives that bring clarity and credibility to this transition.

Hello from Seattle! Ever since my daughters were young, my dad and I have taken them—individually—on “father-daughter-granddaughter” trips. This week’s issue is all about our first-ever trip to Savannah this summer, and the role AI played (or didn’t) in our planning.

Also:

• The numbers behind The AI Ad Economy: New Pew data shows that even though younger generations are more likely to use AI, they are becoming more concerned about it.

• News: OpenAI expands ads to a bunch more countries, Walmart leans on AI for ad growth, new data shows a drop in Reddit citations in ChatGPT.

THE MAIN STORY

The Most Surprising AI User on Our Savannah Trip

Three generations. One trip. Three completely different ways to use AI.

In July, I took my first-ever trip to Savannah, Georgia, with my father and my daughter. We call these trips “father-daughter-granddaughter” trips; ever since my daughters were young, my dad and I have taken them to places like Portland, Vancouver and even Finland. It’s an incredible way to bond while we explore a new destination.

While Carina, my dad and I were driving to Savannah from my parents’ home in central Florida, we started talking about AI, and I realized something interesting was happening. We all love to research (clearly it runs in the family), but we went about it in very different ways when we were planning the trip.

So, one morning on the rooftop of our hotel, I asked Dad and Carina to tell me where AI fit in and where it didn’t. It turned out to be a revealing look not only at how age and generation correlate to search and AI usage, but also how differing habits and trust systems also play a role.

Share

Recent data from YouGov provides helpful context. Across generations, search engines remain the most common place online searchers look for information or answers. But differences start emerging when you look at other information sources, like video platforms, social media and AI.

But before we get into more data, meet my dad, aka “Grandpa Dave.”

The grandfather who skipped Google

Dad is the kind of person who spends hours researching something before he decides, but he has never used Google much. He prefers to go straight to a website.

When I first introduced him to AI in 2024, he was immediately intrigued. He marveled that when he asked it “Who is David Aho?” it knew that he had been an engineer before he retired to Florida, and that he lived in The Villages. Now he uses AI regularly.

“AI for me is so easy to use. Just simply ask a straightforward question, and it’s like talking to another person.” — my dad

Dad used AI in fairly simple ways while we were planning our trip: “Give me the best hotels by budget.” “Give me parking advice in Savannah.” “Tell me about river cruises.”

“I thought we’d be going by train. We dropped that, and then we decided we were going to drive by car,” he said. “And then once I got past that, I said, ‘What are the kinds of things that I would be interested in? And also my daughter and granddaughter?’ From that point, it came up with a lot of general information, like, ‘You ought to see this, you ought to see this, you ought to see this,’ and I said, ‘Fine.’”

At nearly 90 years old, Dad is a pioneer among his pool-playing buddies in the condo building where he and my mom live.

“I’m always looking at new things to do, and I’m a very active person. I find that the people that I associate with, a lot of them are my same age, but they’re not as interested as I am, so I’m coaching them. Now they’re starting to use AI, and we’re able to converse back and forth about how they’re using it,” Dad told me.

He’s also pretty trusting of what he sees in AI.

“You know, I thought the information was pretty good because I looked at the maps. I had a paper map of the area and it showed distances to different places.”

But he’s also got a few tips for improvement. “I’ll tell you one thing that’s lacking in AI for me is that as I went through AI and the information that was presented, it didn’t give an opportunity to click directly on it and get more information, and that would have been very helpful. For instance, Tybee Island, which is about a half-hour drive from Savannah, I would have liked to know more about it, and I eventually found out more about it. But it was a little more effort than I would expect.”

The Gen X analyst who couldn’t stop prompting

You probably won’t be surprised to learn I’m a heavy user of AI for travel research. Here are a few things it’s helped me do recently:

• Find coffee meeting spots near a specific address in New York

• Plan a winery day tour near San Diego

• Compare packable travel totes from Foldie and Baggalini

It turns out my habits may look more Millennial than Gen X (I chalk that up to the fact that I study AI for my job). Millennials are more likely than any other generation to start a search with AI at least some of the time, YouGov found.

For the Savannah trip, I created two threads, which ChatGPT named “Savannah restaurants” and “Savannah historic district tips.” I kept going back to those conversations while we were in Savannah, adding more questions about specific places or activities.

I used AI as a starting point, but not a final decision-maker. When ChatGPT suggested a few restaurants for a special dinner, I went to the websites to review the menus and came back to it with additional requests.

In some cases, I took its tourist recommendations (stroll through the city’s famous squares, explore the City Market). In others, I went a different direction. I noticed that a restaurant called Chive Seabar & Lounge—which ChatGPT recommended in my initial query but dropped as I went further into the conversation—had scallops on the menu and really good reviews. My dad loves scallops, so I made a reservation. All three of us agreed the seafood we ate was some of the best we’d ever had.

The Gen Z traveler who all but ignored AI

My daughter Carina uses AI for some tasks, but for travel, it’s barely on her radar. Her top tool for trip planning is Google Maps.

“I use Google Maps because I like to know what’s around me that I can walk to easily, or what’s around other things that we’re going to,” she told me. “I like efficiency, so if it’s close to our hotel or close to another thing that we’re doing, that’s best for me. I look at Google Maps reviews pretty extensively, and I’ll use it to look at pictures of the menu or tap the link to the website and see the menu there.”

She’s also a big fan of TikTok, which she relied on when she took a solo trip to Tokyo a few months ago. “TikTok was helpful to see a wide variety of things, because a lot of times people would be making montages. So, if I find someone who has similar tastes to me, I can watch a one-minute video, and they summarize six or seven different spots that they like that I might also like.”

Why not AI? It came down to trust.

“I have a lot of my own methods for searching for things, and a lot of times AI is still wrong.” — Carina

That feeling extends to Google’s AI Overviews too.

“Even when I do a Google search, I might glance at what the AI Overview says, but I don’t trust it 100%,” she told me. “Sometimes I’ll look at the sources and I’ll see they are some random person’s website, like a personal website or Reddit or Facebook, and I’ll be like, ‘I don’t feel like I really want to trust those sources.’ So I’ll scroll past it and see what the top Google search results are to actually find a real answer.”

Carina also isn’t alone in her trust in Google Maps. In YouGov’s study, online searchers were far more likely to trust maps and search engines than AI assistants. Trust in AI increased among those who actually used AI to search but was still well below search engines or maps.

Three mental models for using AI

After hearing what my dad and Carina shared and reviewing the data from YouGov, I believe three different mental models are emerging among people who use AI, and they sometimes (but don’t necessarily always) align to age:

• Trust the answer. That’s my dad. If AI can answer naturally and quickly, it’s useful and doesn’t need much checking.

• Trust but verify. That’s me. AI narrows the search, but it needs to be confirmed with primary sources.

• Show me the evidence. That’s Carina. She needs reviews, menus, maps, ratings, photos and recency. And for her, the evidence is often creator-driven.

Our experience also underlines something else I often see in consumer data.

The way people gather information isn’t linear. It winds across sources and inputs. And even if it’s easy to get a quick answer from AI, many people don’t want to trust it without further checking or verification.

Age is one factor driving our search habits, but longstanding behaviors and trust levels also play an important role—whether it’s my dad’s preference to go directly to a website rather than using Google, my penchant for checking ChatGPT’s work or Carina’s reliance on Google Maps and TikTok videos.

Even though we took different paths to learn about Savannah, we got an amazing payoff: a truly memorable generational trip to one of the most beautiful cities I’ve visited in the US.

Share

THE NUMBERS BEHIND THE AI AD ECONOMY

“For the first time, a majority of adults under 30 (55%) now say they’re more concerned than excited about AI.” — Pew Research Center

For several months, I’ve been watching a growing trend in consumer survey data: Even though younger generations are more likely to use AI, they are becoming more concerned about it. So far, that hasn’t impacted their willingness to use it, but the skepticism is there, and it’s real.

What’s interesting is how this compares to social media.

Pew’s data shows a diverging trend in younger people’s attitudes. When it comes to trust in the information they see in social media, there’s been essentially no change since 2022; as of 2025, 50% of US adults ages 18 to 29 said they had a lot or some trust (compared to 51% trusting national news organizations). But when it comes to social media’s effect on wellbeing, attitudes have worsened in recent years. As of 2024, nearly half of US teens 13-17 said they thought social media had a mostly negative effect on people their age, up from 32% in 2022.

Young people were a key factor driving the early growth of social media and they still spend more time there than other generations do. Will this trend play out differently in AI? I think it’s definitely worth paying close attention to the data.

HEADLINES

OpenAI’s ads business hits Europe at the six month mark (Digiday) — OpenAI will begin serving ChatGPT ads across 31 additional European markets on August 24, using major agency groups as the initial buying channel. Six months after launching ads in the US, OpenAI has expanded rapidly across markets.

As retail stagnates, ad dollars help Walmart stay the course (Reuters) — Walmart is positioning AI as the next growth phase for its advertising business, testing ads in its Sparky shopping assistant while adding AI tools for advertisers.

Rise of AI shopping pushes merchants to protect loyalty, Adyen says (Reuters) — As AI assistants take a larger role in shopping, merchants are increasingly concerned that handing discovery and transactions to third-party agents could weaken direct customer relationships and loyalty, according to payments company Adyen.

CMOs are struggling to link AI visibility with sales (Digiday) — Marketers are investing in tools that measure whether brands appear in ChatGPT, AI Overviews and other AI results, but proving that visibility produces revenue remains difficult.

ChatGPT draws more paid clicks from Google than any other top destination (Search Engine Land) — An analysis of 13.1 billion Google search events found that ChatGPT received a higher proportion of paid clicks from Google than any other major destination studied.

Reddit’s ChatGPT Search citations fell 86% in four days: Report (Search Engine Land) — Reddit’s share of ChatGPT Search citations fell from an average 3.83% to 0.52% within days, according to Promptwatch, although the cause and permanence of the shift remain unclear.

Subscribe now

WHAT’S DEBBIE UP TO?

Speaking and Conferences

Here’s where you can find me in September. It’s shaping up to be busy!

Ascendant Network RMN Americas Boot Camp, Sept. 16-17, 2026, Palo Alto CA. I’m conducting two sponsored focus groups with attendees and participating in other sessions. I’ll also be attending Ascendant’s Digital Ascendant on Sept. 15.

Path to Purchase Institute Learning Lab, Sept. 17, 2026 (virtual). I’ll be presenting “The AI-Powered Shopper Journey: How AI is Changing Shopper Engagement and Promotional Strategy.” Register here.

RetailClub AI Festival, Sept 24-26, 2026, Huntington Beach CA. I’ll be discussing “When Chat Becomes the Channel: Advertising in the Age of AI Assistants” with a community that’s focused on moving beyond AI hype to practical implementation and meaningful change across retail. Register to attend here.

Amazon unBoxed, Sept. 28-30, San Francisco. I’m excited to go to my first unBoxed. Got tips? Send them my way. Will you be there too? Let’s connect.

PSA: I have openings in October and November for keynotes, panels and moderating roles, including in New York during Advertising Week. Want to chat? Reply to this email (if you’re seeing it in your inbox) or DM me (if you’re seeing it on the website or app).

Here’s a clip from my keynote at Skai’s ShopAble conference:

Watch my speaking reel

Debra Aho Williamson is the Founder and Chief Analyst of Sonata Insights, where she helps advertisers navigate the next major structural shift in media: the rapid rise and monetization of AI platforms. Over nearly two decades as an analyst at EMARKETER, Debra was among the first to forecast the growth of social media advertising and identify how it would fundamentally reshape marketing investment.

Today, she applies that same pattern-recognition lens to AI, analyzing how consumer adoption is accelerating monetization and redefining brand budgets, media strategy and industry power dynamics. Through Sonata Insights, she partners with leading platforms, agencies and technology companies to deliver research, advisory and strategic narratives that bring clarity and credibility to this transition.

This week I’m digging into the types of prompts that surface ads on ChatGPT, and a controversial test of ads aimed at AI crawlers from Mobian and Time.

Also:

• The numbers behind The AI Ad Economy: Census Bureau data shows how workers use AI, based on their age.

• News: Apple’s Siri licensing deals, Gemini reaches 1 billion monthly users, the New York Post has a chatbot with a very Broadway sounding name.

THE MAIN STORY

Should Marketers Advertise to AI Agents?

ChatGPT ads aren’t even six months old, and many advertisers are still getting used to the idea of advertising to humans using a chatbot. Now, ads aimed at AI agents are also starting to emerge, in the form of a test between AI company Mobian and publisher Time.

Should you try to advertise to humans using AI or to AI agents? Let’s look at both sides of the AI advertising divide.

Advertising in AI: The Prompts That Trigger ChatGPT Advertising

One of the first academic studies of ChatGPT advertising was published this week, and it offers a lot of insight into the growing business.

Researchers Emma Lurie, Ro Encarnación, Sorelle A. Friedler and Danaé Metaxa created 91 free ChatGPT accounts and tested 335 prompts over the first few weeks ads were available on ChatGPT. They gathered more than 3,600 ads that appeared during the first few weeks that ads were available on ChatGPT in the US.

Among the key findings from the study:

• The median exposed account saw an ad on 24% of studied prompts

• Prompts that were product-oriented or related to certain how-to questions were most likely to have ads.

• After purchasable products, cooking & recipe and health, fitness, beauty & self-care prompts were the second and third most likely to deliver ads.

• Retailers were the most common advertisers (Sensor Tower data I studied a few months ago showed the same). Read more here:

The types of prompts that were most likely to contain ads clearly show that ChatGPT is trying to figure out how to mine problem-solving or advice questions for commercial intent. The ad-containing prompts included:

“How can I replace a broken headlight?” — 19.2%
“My dishwasher won’t drain; what should I try?” — 18.2%
“Is the newest iPhone worth the price?” — 17.2%
“How much are running shoes?” — 16.9%
“What’s the best streaming service for sports?” — 16.0%

A searchable archive of the ads the researchers reviewed is available here: https://emmalurie.github.io/chatgpt-ads-library/ .

The caveat with this study is that ChatGPT advertising is evolving quickly, and this was a very early examination of some of the first ad placements (the authors clearly recognize this; the name of the study is “The Beginning of ChatGPT Ads”). The study also ended abruptly when the flow of ads dropped off; the authors surmised that their activity was flagged as inauthentic behavior by ChatGPT.

Share

Still, the findings help confirm some of my beliefs about the role ChatGPT advertising can play in a media plan.

1) Product-oriented advertising is a key focus for ChatGPT.

2) Problem-solving conversations are a commercial opportunity.

Advertising to AI: Mobian’s Ads to AI Agents

I spend most of my time in this newsletter analyzing the growth of ads in AI-infused platforms and experiences—something I call “AI media.” But a different type of AI advertising is also emerging: ads aimed at the agents themselves.

An experiment between Time and AI advertising company Mobian has gotten a lot of attention for placing ads aimed at AI agents on certain pages of Time’s website. The test—designed to help marketer content get more noticed by the AI platforms—appears to be a novel way to extend AEO/GEO efforts via paid media.

Why is this important? The most common AEO/GEO tactics aimed at influencing AI models include things like creating FAQs on websites, optimizing website content to be machine-readable and placing content in platforms like Reddit or LinkedIn, which AI platforms routinely pull from in their responses.

The problem is, this effort doesn’t always lead to accurate responses. According to Mobian’s internal data, AI models get brand facts wrong about 34% of the time.

Mobian’s ads are created in markdown language and placed on separate web pages that aren’t visible to human eyes. The intent is to catch the attention of crawlers from AI chatbots by sharing extra information that an advertiser wanted to convey to the chatbots.

But this week, Perplexity said it was blocking these ads, calling them “deceptive” in a statement in a Digiday article. Perplexity is also developing ways to block other similar types of ads. It’s not clear whether other AI companies will follow Perplexity’s lead and block these ads.

Mobian cofounder and CEO Jonah Goodhart published a blog post earlier this week defending the practice and reporting early results from its tests. Here’s a snippet:

Leave a comment

Among the results he cited were that AI models were able to answer a brand question correctly after being exposed to Mobian’s “brand facts” unit. Goodhart also described a separate test in which models were asked campaign questions “with no page and no prompting.” He said the question “with the most room to move” had a “statistically significant visibility increase against the pre-campaign baseline.”

It’s not clear how many brands Mobian studied or what AI models were studied, so these vendor-provided results are difficult to parse.

The Decision for Advertisers: Ads in AI or Ads to AI?

I firmly believe that ads aimed at AI agents will eventually be a much bigger phenomenon. The potential I see is the ability to create and target an ad that’s aimed at an agent acting on behalf of a consumer. I laid out my vision for this in an opinion column for WARC back in early 2025. The core idea is still highly relevant. Here’s how I think about it:

Ads that try to influence what AI platforms say about a brand are a different beast. I previously wrote about the fact that AEO and paid AI media should work together, but I didn’t envision the type of integration Mobian is trying.

The AI platforms already face questions about transparency, disclosure and trust in the responses. Anything that even hints that advertisers paid to influence those responses—even if that payment didn’t go to the AI platform—creates a huge risk to the growth of AI usage and consumers’ comfort with advertising in AI.

Key Takeaways

• The academic study on ChatGPT ads confirms that the opportunity to influence humans on AI platforms is real and growing.

• The Mobian test and the subsequent fallout tell me that ads designed to sway what AI platforms say about brands may not be the right choice.

• The opportunity to influence AI agents that operate on behalf of humans will come eventually.

• For now, marketers should focus on testing human-facing AI ad inventory and improving their organic AEO/GEO work.

Share

THE NUMBERS BEHIND THE AI AD ECONOMY

In Does Using Artificial Intelligence Save Time at Work?, the US Census
Bureau finds that more than half of US workers have used AI on the job for at least one of 11 tasks the bureau studied. Older and younger workers are near-equally likely to use AI to search for information, but younger users are much more inclined to turn to it for other tasks.

HEADLINES

Apple in Talks to Pay Publishers to Improve AI-Powered Siri (The Wall Street Journal) — Apple is negotiating multiyear agreements with publishers to license news and informational content for an upgraded AI-powered Siri, with compensation potentially tied to how often publishers’ material is used. Unlike broad, fixed-fee licensing deals, the proposed structure would more directly connect publisher payments to AI consumption.

ChatGPT and Gemini both just passed 1 billion users (The Verge) — Google says Gemini has reached 1 billion monthly users, making it the company’s fastest-growing product ever, while OpenAI says ChatGPT surpassed 1 billion monthly users earlier and reached 1 billion weekly users in July. Although the metrics are not directly comparable, The Verge concludes that ChatGPT remains larger while Gemini appears to be growing faster.

French press body asks competition watchdog to take action over Google AI (Reuters) — A French press association has asked the country’s competition regulator to intervene over Google’s AI-generated search summaries, arguing that they reduce publisher traffic while relying on publishers’ content without adequate compensation. The group cited regulatory data indicating substantial traffic declines and is seeking treatment similar to a previous action requiring Meta to negotiate with publishers over AI content use.

New York Post launches “Hamilton” AI chatbot (Axios) — The New York Post has launched Hamilton, an AI-powered conversational news assistant built with Google Cloud and Gemini, alongside personalized briefings, recommendations and other discovery features. Rather than relying solely on external AI platforms to mediate its content, the Post is creating its own AI interface inside its apps.

OpenAI will bring ChatGPT ads to Brazil and Mexico, and is adding a raft of fresh features (Business Insider) — OpenAI is expanding ChatGPT advertising into Brazil and Mexico while testing product carousels, improving conversion measurement and exploring “sponsored agents” that could allow brands to interact conversationally with users. The additions push ChatGPT advertising beyond simple sponsored placements toward richer formats, performance measurement and potentially AI-native brand experiences.

Retailers tap AI shopping traffic but fight to keep customer data (Reuters) — Walmart, Ulta Beauty, Wayfair and other retailers are optimizing their content to appear in ChatGPT, Gemini and other AI shopping experiences, but many still want transactions completed on their own properties so they retain customer data and loyalty relationships. Adobe Analytics data cited by Reuters indicates that revenue per visit from generative-AI referrals is already outperforming traditional channels.

Subscribe now

WHAT’S DEBBIE UP TO?

Quoted

Stanley adapts its marketing for the AI search era (Digiday)

“Stanley did an incredible job of reintroducing its brand to consumers via human creators and influencers,” said Debra Aho Williamson, founder and chief analyst at Sonata Insights. “Now, brands like Stanley are recognizing that they need to think of AI platforms as influencers of a different type. They are becoming important places where consumers discover new brands, learn about them and then make purchase decisions — the same roles human influencers play.”

Podcasts

Retail Media Executive Brief: I joined Ian Simpson of Sensor Tower, alongside fellow analyst colleagues Sarah Marzano (EMARKETER), Andrew Lipsman (Media, Ads & Commerce) and Kiri Masters (Retail Media Breakfast Club) for a conversation about the growth of retail media and how AI platforms are affecting the flow of intent signals.

Speaking and Conferences

Ascendant Network RMN Americas Boot Camp, Sept. 16-17, 2026, Palo Alto CA: I’m conducting two sponsored focus groups with attendees and participating in other sessions.

RetailClub AI Festival, Sept 24-26, 2026, Huntington Beach CA: I’ll be discussing “When Chat Becomes the Channel: Advertising in the Age of AI Assistants” with a community that’s focused on moving beyond AI hype to practical implementation and meaningful change across retail. Register to attend here.

PSA: I have openings in October and November for keynotes, panels and moderating roles, including in New York during Advertising Week. Want to chat? Reply to this email (if you’re seeing it in your inbox) or DM me (if you’re seeing it on the website or app).

Here’s a clip from my keynote at Skai’s ShopAble conference:

Watch my speaking reel

Debra Aho Williamson is the Founder and Chief Analyst of Sonata Insights, where she helps advertisers navigate the next major structural shift in media: the rapid rise and monetization of AI platforms. Over nearly two decades as an analyst at EMARKETER, Debra was among the first to forecast the growth of social media advertising and identify how it would fundamentally reshape marketing investment.

Today, she applies that same pattern-recognition lens to AI, analyzing how consumer adoption is accelerating monetization and redefining brand budgets, media strategy and industry power dynamics. Through Sonata Insights, she partners with leading platforms, agencies and technology companies to deliver research, advisory and strategic narratives that bring clarity and credibility to this transition.

Here’s what’s inside The AI Ad Economy today:

• What tech companies said about AI and advertising in earnings reports: Two distinct approaches are emerging.

• The numbers behind The AI Ad Economy: Nearly half of shoppers are using AI to compare product features or prices, up from 35% last year.

• News: New data shows bigger brands are investing in ChatGPT; Amazon loses a round in its fight against Perplexity; Mastercard says it’s still relevant in the age of agentic.

THE MAIN STORY

Here’s How Leading Tech Firms View the State of the AI Ad Economy

Every tech company is talking about AI these days, but they don’t agree on what AI means for advertising.

After reading recent earnings reports and conference call transcripts from Alphabet, Amazon, Criteo, Meta and Reddit, I am struck by how much these companies agree that AI is changing discovery and improving ad performance. But they sharply diverge when it comes to discussing whether AI becomes an advertising medium.

Two broad approaches are emerging:

1. Companies that believe AI can become a new ad medium

2. Companies that are more focused on using AI to improve existing products

Notably, OpenAI, which isn’t a public company yet, is straddling both camps.

Here are my top three takeaways from the April-June 2026 quarterly earnings, and what key company execs said.

AI is where consumers are increasingly discovering information, products and content—but not to the exclusion of other surfaces.

Google is no longer talking about AI as a separate consumer product. Instead, it’s increasingly integrating AI-native features across its existing surfaces.

The big takeaway for me is the shift in how Amazon describes Alexa—as agentic, not just conversational.

Meta spent a good portion of its call discussing how AI content will expand both the amount and the types of content available across its platforms.

Only three companies—Google, Amazon and Criteo—are discussing AI as a new advertising surface.

Google was the most explicit about advertising in its AI products, although it shared no specific performance metrics or revenue figures.

Amazon shared a compelling stat about the performance of Sponsored Prompts, but it’s not as remarkable as it seems; it’s a statement about correlation, not causality. Like Google, it also discussed the role of AI in advertising creation.

Subscribe now

Despite having a difficult quarter overall, Criteo clearly sees its OpenAI partnership as a long-term growth opportunity.

Not every company believes AI creates a better media ecosystem.

Every other company I reviewed described AI as an expansionary opportunity. But Reddit took a more negative stance, criticizing the impact of AI Overviews on traffic and emphasizing the importance of human conversation and community.

Key Takeaways for the AI Ad Economy

No new medium has ever developed in a straight line, and AI media appears to be following the same path. These earnings reports demonstrate clear forward progress in some areas and diverging pathways in others.

1) AI discovery is real, not hypothetical.

2) AI advertising remains concentrated in a few tech companies.

3) AI is improving advertising faster than it’s creating new ad surfaces.

4) Multiple models for AI media are beginning to emerge.

Read next:

Share

THE NUMBERS BEHIND THE AI AD ECONOMY

AI’s Role in Shopping Is Getting Bigger

AI usage is growing among adult shoppers in the US and UK, according to new data from Bloomreach. Nearly 42% are using AI tools daily, up from 24.1% in 2025, and more than 60% say AI tools make them more confident in their purchase decisions. There were also notable changes (increases) in how people use AI for shopping. The takeaway: AI’s role in the shopping process is getting more prominent—and more impactful. The genie isn’t going back in the bottle.

HEADLINES

Brands like Home Depot, Intuit, and Booking are betting bigger on ChatGPT ads, new data reveals (Business Insider) — Sensor Tower data shows that ad frequency approximately doubled between April and July on ChatGPT. Major consumer brands including Home Depot, Intuit, Booking Holdings and L.L. Bean have joined the platform, and financial services’ share of estimated spending increased from 2% to 12%. (For more ChatGPT ads data, see my June article.)

Amazon loses US court ban on Perplexity’s AI shopping tools (Reuters) — A federal appeals court overturned an injunction preventing Perplexity’s Comet browser from using AI agents to access Amazon on behalf of shoppers. Amazon argued that the technology improperly accessed customer accounts and bypassed its controls, while Perplexity characterized the dispute as an attempt to prevent shoppers from using AI tools that could circumvent Amazon’s advertising-driven interface.

Can credit cards survive AI shopping? Why Mastercard’s CEO says they’ll ‘prevail’ in this new era. (MarketWatch) — Mastercard CEO Michael Miebach argued that card networks will remain central as AI agents begin making purchases, pointing to the company’s Agent Pay initiative and its work on identity verification, fraud protection and verifiable consumer intent. Mastercard is positioning its existing payment and trust infrastructure as the connective tissue between AI assistants, merchants and consumers.

A debate is brewing over the risks and rewards of serving ads to AI agents (Business Insider) — Adtech startup Mobian and Time have tested text-based “Agent Ads” placed on web pages designed primarily for AI crawlers, with participating advertisers including Ally Bank and the Project Management Institute. The approach attempts to influence the information AI assistants retrieve and use when answering questions, although critics warn that undisclosed commercial content could manipulate recommendations and damage trust.

Shopify shares soar as forecast shows AI is boosting business, not disrupting (Reuters) — Shopify reported stronger-than-expected growth and said partnerships with OpenAI, Google and Microsoft are helping its merchants reach shoppers through emerging AI experiences. Usage of Shopify’s merchant assistant Sidekick also increased sharply, while the company positioned its commerce infrastructure as capable of supporting transactions initiated by both people and AI agents.

AI’s ‘cookie banner’ moment: EU labels come for the bots (Financial Times) — New European Union transparency requirements for AI-generated content took effect on August 2, requiring companies to disclose certain chatbot interactions, synthetic images and AI-generated videos. Retailers, travel companies and advertising agencies are among the businesses adapting their consumer experiences and content-production processes, although companies say portions of the guidance remain unclear.

IAB’s New Advice On How To Measure AI Search Visibility (AdExchanger) — The Interactive Advertising Bureau has published new guidance urging marketers to move beyond traditional search metrics as AI-powered discovery becomes more conversational and recommendation-driven. The framework encourages brands to measure AI visibility using a combination of citations, recommendations, sentiment, referral traffic and business outcomes rather than relying on any single metric, reflecting the industry’s effort to establish common standards for evaluating performance in AI search.

Subscribe now

WHAT’S DEBBIE UP TO?

In the Media

OpenAI Licenses Content From Yelp, Looks To Be Building Ad Network (MediaPost)

Podcasts

The Ripple Effect of AI as a Media Channel: I joined Mediaocean CMO Aaron Goldman for the inaugural episode of The Ripple Effect. We discussed the growth of AI as a media channel and what it means for advertisers and adtech. Catch it on Spotify or Apple Podcasts.

Retail Media Executive Brief: I joined Ian Simpson of Sensor Tower, alongside fellow analyst colleagues Sarah Marzano (EMARKETER), Andrew Lipsman (Media, Ads & Commerce) and Kiri Masters (Retail Media Breakfast Club) for a conversation about the growth of retail media and how AI platforms are affecting the flow of intent signals.

On Stage

RetailClub AI Festival, Sept 24-26, 2026: I’ll be discussing “When Chat Becomes the Channel: Advertising in the Age of AI Assistants” with a community that’s focused on moving beyond AI hype to practical implementation and meaningful change across retail. Register to attend here.

PSA: I have openings in October and November for keynotes, panels and moderating roles, including in New York during Advertising Week. Want to chat? Reply to this email (if you’re seeing it in your inbox) or DM me (if you’re seeing it on the website or app).

Here’s a clip from my keynote at Skai’s ShopAble conference:

Watch my speaking reel

Debra Aho Williamson is the Founder and Chief Analyst of Sonata Insights, where she helps advertisers navigate the next major structural shift in media: the rapid rise and monetization of AI platforms. Over nearly two decades as an analyst at EMARKETER, Debra was among the first to forecast the growth of social media advertising and identify how it would fundamentally reshape marketing investment.

Today, she applies that same pattern-recognition lens to AI, analyzing how consumer adoption is accelerating monetization and redefining brand budgets, media strategy and industry power dynamics. Through Sonata Insights, she partners with leading platforms, agencies and technology companies to deliver research, advisory and strategic narratives that bring clarity and credibility to this transition.

Hello from Seattle! Here’s what’s inside The AI Ad Economy today:

• Did OpenAI just reveal plans for an ad network? A job posting—which was edited after a reporter inquired about it—has some very interesting tea leaves.

• The numbers behind The AI Ad Economy: You probably know which AI platform has the highest awareness, but do you know who’s No. 2?

• This week’s top AI news: Google’s strong earnings, Google launches AI Mode in France, and fresh retail media data from Skai.

Quick programming note: I’m taking a summer break next week; I’ll be back in your inbox on August 6.

THE MAIN STORY

Did OpenAI just reveal plans for an ad network?

OpenAI’s job postings often yield a bounty of information about its advertising business. This week was no exception.

Recently, the company posted a job for a “Support Delivery Lead, Ads.” While the title was ho-hum, the description was anything but.

Among the responsibilities OpenAI listed were:

• “Lead a team responsible for end-to-end support delivery across both sides of the ads ecosystem”

• Work with publishers on “inventory setup, ad serving integrations, reporting and yield-impacting issues”

• “Design and continuously improve support coverage for both ad buyers and publishers, ensuring the team can diagnose buyer-side delivery issues and publisher-side monetization and integration issues with equal rigor”

  ---

While OpenAI didn’t explicitly define what it meant by “publishers,” the fact that it put ad buyers and publishers in the same sentence immediately made me think it wants to play a role in both the buy-side and the sell-side of digital advertising.

But here’s the interesting twist: The job posting no longer references publishers. It was edited shortly after Business Insider’s Lara O’Reilly contacted OpenAI to ask about it. (Her article “An OpenAI job listing described ambitions to build an ad network. Then the company deleted the references” features a quote from me.)

Here’s how the same paragraph appeared as of yesterday in the revised job post. Gone are the words “both sides of” as well as the entire description of the publisher side of the role.

OpenAI has moved quickly to build the infrastructure it needs to run an ad business, including self-serve buying, CPC bidding, adtech integrations, a Conversions API and more. So far, all the attention has been focused on ChatGPT.

I believe it’s worth considering what the original post revealed about OpenAI’s ad growth plans beyond ChatGPT. Here’s what I think it could mean:

• A new direction for OpenAI’s ad business.

• A way for OpenAI to try to placate content publishers that have lost traffic because of AI platforms like ChatGPT.

• Validation—and potentially a huge disruption—to the emerging startups that place advertising across AI apps.

Share

Let’s take those points apart.

A new direction for OpenAI’s fledgling ad business

Thus far, OpenAI has focused on ChatGPT as its primary ad surface. Adding new capabilities for third-party publishers would extend it into a different type of business, not only as a seller of ads on its own properties but also as a broker of ads on others.

In a typical ad network, publishers allocate certain ad spots on their websites and apps to the ad network. Advertisers set budgets and define who they want to target. Then, when a user visits a website or app, the ad network matches them with an eligible ad and serves the ad. The advertiser pays the network, and the network shares a portion with the publisher, while keeping a fee for itself.

OpenAI would generate revenue in its role as a middleman between publishers and advertisers.

An olive branch to publishers

Publishers are stuck between a rock and a hard place when it comes to AI. Bots scrape their content and they’re losing traffic to AI platforms. Some, such as The New York Times, have sued OpenAI.

By explicitly referencing publishers in its job posting, OpenAI could be testing the waters for the idea of helping them generate more ad revenue via a supply-side integration.

The idea of joining an OpenAI ad network would be intriguing for publishers. It could also give them access to new AI ad formats OpenAI might develop.

But there’s also risk: Partnering in this way creates a dependency on a platform that could change the rules at any time. Publishers that became reliant on Facebook for distribution in the 2010s will remember what happened next: Facebook started prioritizing friends and family content over content from publishers and Pages, leaving publishers that made long-term investments in the platform in the lurch.

While the Facebook situation revolved more around content distribution and traffic rather than ad revenue, the comparison is relevant because dependency is still a huge factor.

An 800-pound gorilla in the emerging AI ad network business

Several businesses are already building advertising marketplaces that aggregate inventory across AI-native apps and services. One, Koah, has even called itself “AdSense for AI.”

Another company in the space, Thrad, is evolving beyond the long tail of AI app publishers it started out supporting. In addition to operating a network of independent AI apps, it also offers access to ChatGPT ads as a managed service. When I sat down with CEO Andrea Tortella during Cannes Lions, here’s what he said:

“How we think about Thrad is [that we are] aggregating conversational surfaces wherever they may be across the internet, across the walled gardens, across the long tail.”

The positive spin is that if OpenAI does indeed want to aggregate publisher inventory, it would validate the business models of Koah, Thrad and other companies in the space.

The looming problem: Validation could quickly become displacement. OpenAI already has a self-serve ad platform and could easily add other publishers or apps to fulfill advertiser demand.

That’s a lot like what Google did. It built its AdSense network around its fledgling search business, adding third-party properties to increase inventory. I’m not saying that’s what OpenAI will end up doing, but there’s certainly the potential for history to repeat itself, only with an AI-driven twist.

Final thoughts

One of my core beliefs is that advertising in AI media will be much more than just chatbot advertising. OpenAI is hinting that it’s exploring new ad business areas outside of ChatGPT. Watch what it does—and if you want to read the tea leaves, keep close tabs on the job postings.

Leave a comment

Related reading:

THE NUMBERS BEHIND THE AI AD ECONOMY

AI awareness is high, but when it comes to usage, one platform still stands out

Among US adult AI users, awareness of competing platforms is very high, with ChatGPT and Gemini almost neck and neck and Claude, Copilot and Meta AI not far behind. But when it comes to a primary platform, the percentage of users citing ChatGPT was more than double those choosing Gemini, according to a new study by Profound’s economist Jennifer Zou. Based on a survey conducted in June (so it’s very fresh), the study also found that the experience of using a platform was a stronger indicator of trust than brand reputation. “Trust shows up two ways: some of it rides on brand reputation (the Googles and OpenAIs of the world), but a lot of it only gets built once people actually use a product,” Zou wrote on LinkedIn.

HEADLINES

It’s a Google kind of week…

Google Was a Lifeline for Publishers. Now They’re Thinking of Cutting it Off (The Wall Street Journal) — Reddit, USA Today, Politico, The Economist, People Inc. and Reuters are reassessing their relationships with Google as AI-generated answers reduce referral traffic and weaken the economic value of allowing Google to access their content. Reddit has discussed restricting Google’s AI access as its $60 million-a-year licensing agreement comes up for renewal, while USA Today is considering blocking Google’s crawler altogether.

Google’s Q2 earnings beat Wall Street’s expectations on AI boom (Associated Press) — Alphabet reported strong second-quarter advertising results as AI Overviews and AI Mode approached mass-market scale, with analysts pointing to continued search-ad growth as evidence that Google’s AI products are expanding search activity rather than immediately cannibalizing its core business. Google says AI-generated experiences are encouraging people to conduct more complex searches while preserving strong advertising demand.

Google’s AI Overview, now available in France, raises anticipation and concern (Le Monde) — Google launched AI Overviews and AI Mode in France, bringing generated answers above traditional search links while allowing publishers to opt out of summaries without disappearing from conventional search. Google also said 450 French media outlets would receive compensation when their content appears in AI Overviews, reflecting commitments made after a French competition ruling.

Google’s AI Search Data Is Growing, But The Gaps Remain (Search Engine Journal) — Google has started piloting Merchant Center reporting that shows retailers some of the shopping questions consumers ask in AI Mode and AI Overviews, while adding more AI-related visibility data across its products. However, marketers still lack important details including comprehensive click reporting, individual Search Console queries and clear separation of performance across AI surfaces.

Exclusive data shows AI-powered shopping fuels retail media click growth (EMARKETER) — Skai data showed retail media ad clicks rising sharply in several high-consideration categories during the second quarter, including 64.3% year over year for business and industrial products and 20.3% for computers and consumer electronics. EMARKETER connected the gains to the expansion of AI-powered shopping experiences that help consumers research and compare products before clicking through to retailers.

Subscribe now

WHAT’S DEBBIE UP TO?

In the Media

Q2 2026 Was The Quarter AI Media Started To Scale: AdExchanger recently republished my AI Ad Economy Quarterly article this week. Thank you AdExchanger!

On Stage

After a hectic May-June, I’m taking a pause from speaking to focus on a new slate of presentations for the fall. More soon on those!

In the meantime, here’s a clip from my recent keynote at Skai’s ShopAble conference:

Podcasts

“Future-Proofing Brand Visibility in the Age of AI” with Kevin Lee of the eMarketing Association. We talk about about how AI is reshaping consumer behavior, ad revenue and SEO/GEO and what brands, agencies and publishers must do now to future-proof visibility in an AI-driven world.

Watch my speaking reel

Debra Aho Williamson is the Founder and Chief Analyst of Sonata Insights, where she helps advertisers navigate the next major structural shift in media: the rapid rise and monetization of AI platforms. Over nearly two decades as an analyst at EMARKETER, Debra was among the first to forecast the growth of social media advertising and identify how it would fundamentally reshape marketing investment.

Today, she applies that same pattern-recognition lens to AI, analyzing how consumer adoption is accelerating monetization and redefining brand budgets, media strategy and industry power dynamics. Through Sonata Insights, she partners with leading platforms, agencies and technology companies to deliver research, advisory and strategic narratives that bring clarity and credibility to this transition.

Hello from 35,000 feet (again!). I’m flying back to Seattle after spending a week with my parents in Florida, including a fun side trip with my dad and daughter to Savannah.

Here’s what’s inside The AI Ad Economy today:

• AI disclosure is picking up steam: Why Google and Meta just announced new transparency rules and what they mean for advertisers.

• The numbers behind The AI Ad Economy: 20% of US online shoppers turned to AI to research their most recent purchase.

• This week’s top AI news: New ROAS features from OpenAI, the share of Google AI Mode queries showing ads, and a brewing debate in Germany over the neutrality of AI platforms.

THE MAIN STORY

The Biggest AI Ad Platforms Just Made a Big Bet on Transparency

The world’s two largest digital ad sellers, Google and Meta, have made major changes to their advertising policies, requiring disclosure when AI is used to create or significantly edit ads.

Six months ago, I argued that this kind of transparency would become increasingly important, especially for younger consumers. The moves by the industry’s top two sellers signal that AI ad disclosure is moving beyond a best practice and becoming a standard feature of advertising.

Share

What Google and Meta now require

Within days of each other this month, the companies announced new policies that inform consumers when AI was used to create or significantly edit an ad.

Google now shows a “created or edited with AI” notice within a new “how this ad was made” panel. The panel can be accessed via the three-dot menu or by tapping or clicking on the info icon in ads in search, YouTube and Discover.

The notice appears automatically when advertisers use Google’s AI tools. Advertisers using third-party AI tools can also choose to add the label themselves.

Meta now adds an “AI info” label to ads created with AI features such as Background Generation, Image Generation or Add Animation. Like Google’s implementation, users can see the label if they visit the three-dot menu and review the “About this ad” information.

Meta uses AI detection tools to determine whether AI was used in the ad applies the label automatically. Advertisers can also choose to self-disclose AI usage.

Regulation of AI advertising is increasing around the world

Google and Meta are trying to get ahead of the patchwork of AI disclosure regulation that’s emerging around the world.

Korea: In January, Korea became one of the first countries in the world to impose AI content labeling legislation. The AI Basic Act requires AI business operators (companies that directly provide AI products and services to users) to label AI-generated content. Businesses that use AI for work or creative activities aren’t subject to the requirements.

Europe: A new code of practice, scheduled to take effect next month, creates voluntary rules for both AI platforms and content creators to disclose synthetic content. The rules govern more than just advertising; to comply with the code, companies must label published text and deepfake images, audio or video.

New York: As of June 2026, a statewide law requires any ad that contains an AI-generated person instead of an actor to be labeled “synthetic performer.”

IAB research confirms the need for disclosure

In January, I published a research study I conducted in partnership with IAB that showed—for the second year in a row—that ad execs were deeply out of sync with US Gen Z and Millennial consumers about the benefits of using AI for advertising. The findings have aged remarkably well.

• 82% of ad executives believe Gen Z/Millennial consumers feel very or somewhat positive about AI-generated ads. That’s nearly double the 45% of consumers who actually feel that way.

• Gen Z consumers describe brands that use AI for ads more negatively than Millennial consumers. Thirty percent call AI-using brands “inauthentic” (vs. 13% of Millennials), 26% “disconnected” (vs. 8% of Millennials) and 24% “unethical” (vs. 8% of Millennials).

• Disclosing AI influences purchase consideration. Among Gen Z and Millennial consumers, 73% said that if they knew an ad was created with AI it would either increase or have no difference on their likelihood to purchase the product or service.

What the growing push for AI disclosure means for marketers

Advertisers should focus on getting ahead of regulation, rather than scrambling to keep up. I expect more markets will enact AI advertising regulations in the coming months. Transparency is becoming a way to stay compliant with the growing patchwork of regulation around the world.

Google and Meta’s moves will influence more than just their own platforms. By taking steps to make disclosure a product feature, consumers will be conditioned to expect similar disclosures on other platforms.

Disclosure improves consumer comfort with AI-generated advertising. It can also increase purchase likelihood. That doesn’t mean consumers automatically trust AI-generated advertising, but it does suggest the ad industry increasingly agrees that people deserve to know when AI played a role in creating the ads they see.

For detailed guidance from IAB on when and how to disclose AI usage, read the IAB AI Transparency and Disclosure Framework.

Leave a comment

THE NUMBERS BEHIND THE AI AD ECONOMY

20% of Online Shoppers Used AI for Their Most Recent Purchase

A study from Product.ai finds that AI assistants had a relatively smaller influence on online shoppers’ most recent purchase than reviews, websites or peer/creator recommendations. AI users are also highly likely to verify what they see; 45% reported they always look at another source after seeing recommendations in AI.

HEADLINES

ChatGPT Ads Manager Gains Attributed Sales Value & Sales ROAS & Products (Search Engine Roundtable) — OpenAI has quietly added attributed sales value, sales ROAS and product-level reporting to its advertising dashboard, giving commerce advertisers a clearer view of whether ChatGPT ads generate revenue rather than merely clicks.

AI Mode shows ads on 1 in 3 queries: 50K keywords studied (SE Ranking) — An analysis of more than 50,000 U.S. commercial keywords found text ads appearing in 29.45% of Google AI Mode results, with ad incidence rising above 50% among keywords carrying CPCs of $10 or more.

Meta’s AI advertising dreams have become a nightmare for brands⁠ (Business Insider) — Meta’s push to automate more of the ad creation process is generating frustration among advertisers, who say AI tools are distorting products, altering brand messaging and sometimes enabling creative changes automatically. While Meta argues advertisers remain responsible for reviewing outputs, brands say the added oversight offsets much of AI’s promised efficiency.

ChatGPT Ads New Overview Tab, Expands To Japan & South Korea (Search Engine Roundtable) — OpenAI made ChatGPT ads available in Japan and South Korea while adding an account overview that highlights campaign performance, account health and recommended actions.

Ask YouTube AI search experience expands to U.S. desktop users (Search Engine Land) — YouTube expanded its conversational search feature from a Premium test to signed-in U.S. desktop users ages 13 and older, allowing people to ask natural-language questions and receive AI-generated answers accompanied by videos, Shorts and follow-up prompts.

German media regulator says Google’s AI Overviews subject to German media law (Reuters) — German regulators said Google AI Overviews and Perplexity should be treated as content providers and potentially as media intermediaries, rather than as neutral platforms displaying third-party information. The decision could make AI platforms more accountable for the accuracy, prominence and source selection of generated answers.

Cloudflare’s latest AI rankings expose the web’s biggest free rider (Business Insider) — Cloudflare data showed Anthropic crawling approximately 2,800 web pages for every referral it returned during the first week of July, compared with three crawls per referral from DuckDuckGo; OpenAI and Perplexity also consumed substantially more content than they returned in traffic.

News outlets urge a judge to sanction OpenAI in a high-stakes AI copyright fight (Associated Press) — News organizations including The New York Times and the Daily News asked a federal court to sanction OpenAI over allegations that it withheld or destroyed evidence in their copyright litigation against OpenAI and Microsoft.

Subscribe now

WHAT’S DEBBIE UP TO?

In the Media

Q2 2026 Was The Quarter AI Media Started To Scale: AdExchanger republished my AI Ad Economy Quarterly article this week. Thank you AdExchanger!

On Stage

After a hectic May-June, I’m taking a pause from speaking to focus on a new slate of presentations for the fall. More soon on those!

In the meantime, here’s a clip from my recent keynote at Skai’s ShopAble conference:

Video Interviews

During Cannes Lions, I sat down with Mediaocean’s Ivan Markman and Grant Parker for two “AI Cannesversations” about advertising and media:

• How marketers can drive AI adoption—and the power of NIVO

• Connecting the dots throughout the marketing funnel with agentic advertising

Podcasts

“Future-Proofing Brand Visibility in the Age of AI” with Kevin Lee of the eMarketing Association. We talk about about how AI is reshaping consumer behavior, ad revenue and SEO/GEO and what brands, agencies and publishers must do now to future-proof visibility in an AI-driven world.

Watch my speaking reel

Debra Aho Williamson is the Founder and Chief Analyst of Sonata Insights, where she helps advertisers navigate the next major structural shift in media: the rapid rise and monetization of AI platforms. Over nearly two decades as an analyst at EMARKETER, Debra was among the first to forecast the growth of social media advertising and identify how it would fundamentally reshape marketing investment.

Today, she applies that same pattern-recognition lens to AI, analyzing how consumer adoption is accelerating monetization and redefining brand budgets, media strategy and industry power dynamics. Through Sonata Insights, she partners with leading platforms, agencies and technology companies to deliver research, advisory and strategic narratives that bring clarity and credibility to this transition.

Hello from 35,000 feet; I’m on a plane to visit my parents. This Seattle gal isn’t prepared for Florida in July. Claude said I should have spent the past week “outside during the warmest part of the day, or do exercise sessions that make you sweat (even indoors with less AC) to nudge your body toward heat adaptation.” Oops. Next time?

Here’s what’s inside The AI Ad Economy today:

• Your guide to what happened in AI media in Q2 2026: Three big developments, essential catch-up reading and my analysis of what’s ahead in Q3.

• The numbers behind The AI Ad Economy: Two-thirds of Gen Z and Millennial AI users have purchased a product they learned about in a chat, according to The New Consumer.

• News of the week: New ad types may be coming to OpenAI. The platform that dominates AI referral traffic. A look inside Amazon’s agentic ad format.

THE MAIN STORY

Q2 2026 Was the Quarter AI Media Started to Scale

Advertising in AI-native destinations and experiences is already starting to look less like a series of experiments and more like an emerging market.

Three months ago, I argued that AI platforms had started acting like media companies. Now, the conversation has already changed. It wasn’t simply that OpenAI introduced ads in ChatGPT and advertisers started to experiment. It was that an entire ecosystem of buying tools, measurement, agency partnerships and commerce infrastructure began forming around AI media.

Here are the three biggest developments in Q2 2026 and what I’m watching in Q3.

#1: OpenAI had an advertising coming-out party

OpenAI got off to a fast start with advertising in ChatGPT in Q2; after launching in February 2026, ads are already available in eight countries around the world. The company spent most of Q2 developing the tools and features advertisers expect from a digital ad business, including a self-serve ad manager, conversions API and CPC bidding.

What types of companies were ChatGPT’s biggest advertisers during the quarter? Retailers, according to Sensor Tower data I wrote about in this newsletter in early June. However, these businesses didn’t advertise themselves; instead, they placed reams of product ads. AI shopping conversations naturally revolve around products rather than stores, suggesting AI media may reward product relevance over retailer brand awareness.

Another important development ad industry executives should pay attention to: OpenAI not only wants to sell ads on ChatGPT; it also wants advertisers and their agencies to use its creative tools to make and manage them. During a Cannes Lions press briefing that I attended, the company showed a demo of how agencies could use Codex and OpenAI’s workspace agents to develop creative assets and variations.

Essential reading

OpenAI’s Dave Dugan explains how ChatGPT is building a new ad category

ChatGPT is making room for ads from regulated verticals—but marketers aren’t ready to jump in at scale

OpenAI moves to automate ad creative

83% of ChatGPT ad triggers don’t exist in traditional search

Takeaway

Q2 2026 was an inflection point for OpenAI. Now it’s time for the company to prove it can earn more than just experimental ad budgets. That will happen if it creates ad formats that are genuinely unique, delivers better measurement and attribution than its competitors and becomes an end-to-end resource for both the creative tools to make ads and the media channels to place them.

#2: The AI media infrastructure race accelerated

Every successful advertising medium has the same supporting infrastructure: Measurement. Buying. Optimization. Attribution. Creative. Commerce connections. AI media made rapid progress toward checking off almost every one of those boxes in Q2.

What’s most interesting is how quickly adtech, creative tech and ad agency partners have moved to partner with OpenAI.

The company created partnerships with Criteo, StackAdapt, Pacvue, Kargo and Adobe. Holding companies Dentsu, WPP, Publicis and Omnicom joined as agency partners, and LiveRamp came on board for measurement (showing that first-party data and identity are becoming foundational to AI-era marketing).

Outside of OpenAI, a few other developments point to the intersection of commerce infrastructure with AI media. The FIDO Alliance, an organization founded to create interoperable password authentication standards, started to work on standards for agentic commerce, alongside Google and Mastercard. And Albertsons and Criteo are putting sponsored products into conversational grocery search.

Essential reading

OpenAI Opens ChatGPT Ads to Self-Service Platform

OpenAI’s ChatGPT ads get their first measurement partner in LiveRamp

Google donates Agent Payments Protocol to FIDO Alliance

Brands Can Now Buy Albertsons Conversational Search Ads Through Criteo

Takeaway

One clear sign of a maturing ad market is the availability of infrastructure that makes it easier to plan, buy and optimize ads and then measure performance and compare results with other types of advertising. Even while advertisers continue to test AI media, the supporting infrastructure is making those experiments easier to execute. Advertisers don’t need to shift major budgets overnight, but they do need to start building institutional knowledge about how advertising in conversational settings performs differently than other forms of advertising.

#3: Google delivered a coherent vision for how advertising will work in AI

Before Q2, Google’s AI initiatives—and the advertising within them—felt disconnected. By the end of the quarter, however, Google connected the dots. It delivered a coherent narrative for how search and AI will coexist and how advertising will work in AI-powered search, shopping and video.

At Google I/O, the company announced the integration of AI tools directly into the search box, calling it “the biggest upgrade to our Search box in over 25 years.” It also plans to introduce information agents that automate basic search tasks and agentic booking capabilities in search.

At Google Marketing Live, the company delivered new ad formats in AI Mode including conversational discovery ads and highlighted answers, and said it plans to introduce AI-powered shopping ads and business agents that exist directly within an ad. In addition, the Direct Offers test is being expanded to travel advertisers, with new features including the ability to bundle promotions and the integration of native checkout for Universal Commerce Protocol merchants.

With these moves, Google clarified how search evolves in tandem with AI, not separate from it, and how the company’s sprawling ecosystem of ad products is starting to evolve toward a conversational future.

Essential reading

Google announces new AI Mode ad formats and agentic commerce tools at Google Marketing Live

A new generation of ads for the AI era of search

A New Era for AI Search

Google pushes AI-generated ads further into search results (paywall)

Takeaway

Google is actively redefining what search means—for consumers as well as advertisers. It’s becoming an AI assistant that does the work for people: answering questions, comparing the products, monitoring news, and soon, completing purchases. That’s teeing up a big change in how search advertising looks and feels.

What to Watch in Q3 2026

AI ad formats start to become more unique. Google’s upcoming new ad products tell me that we’re about to enter a phase when the ad formats in AI media really start to come into their own. Not a replica of search, but truly unique formats that not only exist in AI destinations but use AI as part of the ad experience itself. Recent job listings on OpenAI’s jobs site offer important clues about new image, video and interactive ad products that might be coming.

New ways to attribute conversions. One thorny question in the ad industry is the ability to measure the effect of ads in AI media. What do consumers do next? Do they drive conversions or sales? In Q3 I’m expecting OpenAI, Google and others will push to deliver better attribution tools to help advertisers understand the AI customer journey and the proximity of AI platforms to the decision click.

Will advertisers stick with ChatGPT after initial tests? Growth has already been rapid—more than 1,400 brands have advertised on ChatGPT in the US in the first couple of months, across nearly every industry category, per Sensor Tower. But I’m less interested in the number of new advertisers on the platform (which will naturally go up given its expansion to new markets and the addition of self-serve) and more interested in the percentage of advertisers that stick around (or come back after they evaluate their initial results).

Retailers and AI media get closer together. Amazon and Walmart, the two largest players in retail media, are already taking steps to integrate their AI search tools (Alexa and Sparky, respectively) more deeply into consumers’ everyday shopping behavior. Neither has said much about the role of paid advertising in these AI tools, but I expect that will change. Conversational ads in retail media will enable retailers to capture consumers who are becoming accustomed to searching for and finding products by using natural language queries, not keywords.

Meta AI stumbles (again). Anyone familiar with my past work at eMarketer knows I spent nearly two decades researching social media ad businesses. If you had asked me two years ago whether Meta would be a leader in AI media, I would have said, “Absolutely!” I assumed it would surely figure out how to extend ads into AI-native ad surfaces. But that hasn’t happened. While it’s using AI to make its existing ad surfaces and formats perform better, it hasn’t been able to come up with AI features that consumers want and that don’t get the company in trouble. I don’t expect Meta to materially change that picture in Q3.

The Bottom Line

When I started writing this newsletter last December, I was confident the AI ad economy—the emerging ecosystem of companies, technologies and stakeholders that are shaping how advertising will function and evolve in an AI-driven world—would become the next big ad economy. Everything I saw in Q2 made me even more convinced AI media will soon become a line item in the media budget.

Three things advertisers should care about right now:

1. Ads in AI media are no longer a curiosity; they’re becoming increasingly accessible to advertisers.

2. The supporting infrastructure to optimize, measure and compare performance is arriving faster than expected.

3. Now is the time to build institutional knowledge, because the market is maturing faster than most planning cycles.

Leave a comment

THE NUMBERS BEHIND THE AI AD ECONOMY

AI Chats Drive New Product Awareness

Two-thirds of US Gen Z and Millennial AI users have purchased at least one product they learned about in a chat with an AI app or tool, according to The New Consumer, a publication covering the intersection of technology and consumer brands from journalist Dan Frommer.

HEADLINES

OpenAI looks beyond a single ad format with image, video and conversational ads in the works — Digiday uncovered OpenAI job postings showing the company is building image, video, conversational, native and interactive advertising formats. That suggests OpenAI is thinking beyond sponsored answers toward a broader AI-native media platform.

OpenAI Seeks Ad Format Innovation; Monetizing The Post-Click Experience — AdExchanger highlights Digiday’s reporting alongside OpenAI’s hiring activity and frames it within the company’s broader monetization strategy. The emphasis is less on today’s products than on the architecture of tomorrow’s AI advertising business.

Kickbacks Takes An Outsider’s View While Bringing Ads To AI Agents (AdExchanger) — Kickbacks introduced an advertising model that inserts sponsored experiences into AI agent workflows while sharing revenue with users. Rather than advertising inside chatbot conversations, the company is betting that commercial opportunities will emerge around autonomous task completion.

ChatGPT commands 92% of AI referral traffic. Here’s what 6.77 million sessions reveal. (Search Engine Land) — New analysis of 6.77 million AI-referred sessions found ChatGPT generated more than 90% of referral traffic among major AI assistants while overall AI referrals continued to grow rapidly.

Amazon’s latest ad format offers a glimpse of advertising’s agentic future (Digiday) — Amazon’s new Alexa+ Agentic Ads allow consumers to complete purchases directly inside AI-powered conversations. Although announced during Cannes, the implications continue to unfold as advertisers evaluate the model.

Subscribe now

WHAT’S DEBBIE UP TO?

On Stage

After a hectic May-June I’m taking a pause from speaking to focus on a new slate of presentations for the fall. More soon on those!

In the meantime, here’s a clip from my recent keynote at Skai’s ShopAble conference:

Podcasts

“Future-Proofing Brand Visibility in the Age of AI” with Kevin Lee of the eMarketing Association. We talk about about how AI is reshaping consumer behavior, ad revenue and SEO/GEO and what brands, agencies and publishers must do now to future-proof visibility in an AI-driven world.

Watch my speaking reel

Debra Aho Williamson is the Founder and Chief Analyst of Sonata Insights, where she helps advertisers navigate the next major structural shift in media: the rapid rise and monetization of AI platforms. Over nearly two decades as an analyst at EMARKETER, Debra was among the first to forecast the growth of social media advertising and identify how it would fundamentally reshape marketing investment.

Today, she applies that same pattern-recognition lens to AI, analyzing how consumer adoption is accelerating monetization and redefining brand budgets, media strategy and industry power dynamics. Through Sonata Insights, she partners with leading platforms, agencies and technology companies to deliver research, advisory and strategic narratives that bring clarity and credibility to this transition.

It’s gray and cool in Seattle this week; the joke we always tell is that summer starts here on the 5th of July. But after the heat in Cannes last week, I’m OK with a little bit of cloud cover.

Here’s what’s inside The AI Ad Economy today:

• The hard questions marketers need to ask about AEO: Lately, I have been referring to AEO as a Sisyphean task. In Greek mythology, King Sisyphus angered the gods, and his punishment was to forever push a boulder up a hill, only to see it come crashing back down to the bottom. Imagine if the hill kept moving, the boulder kept changing shape and the gods never revealed the rules. That’s AEO today.

• The numbers behind The AI Ad Economy: North America trails other world regions when it comes to trust in AI recommendations.

• News of the week: Albertsons, Instacart eye ads in AI-infused experiences; brands contemplate AEO; ChatGPT ad dismissals drop 50%.

THE MAIN STORY

The Sisyphus Problem in AEO

“How do you influence an A.I. model?” That was the central question in Marty Swant’s New York Times article about Cannes Lions a few days ago. In it, he explored the implications for marketing when AI chatbots determine which brands and products show up.

It’s an important question. But the more I’ve been writing about the impact of AI on marketing and advertising, the more I think the bigger question isn’t “How do you influence AI?”

It’s this:

“Can anyone prove they know how to influence an AI model?”

I first wrote about the importance of brands showing up in AI search results nearly two years ago, in a column for Procter & Gamble’s Signal 360 magazine (you can read it in the Wayback Machine).

Here’s the relevant excerpt:

Fast forward to last week, and answer engine optimization (AEO) was a hot topic during Cannes Lions.

• At a luncheon I attended, a publishing executive raved about their company’s AEO offering as a way for client brands to get noticed by being associated with verified news content.

• At a dinner hosted by Gumshoe.ai, an AI visibility platform, one conversation around the table was how to manage what already feels like an unending amount of work.

• Profound, one of the largest companies in the AEO space, came to Cannes for the first time this year to pitch its services.

The presence of Profound at Cannes was a big deal for AEO. Here’s what CEO James Cadwallader wrote on LinkedIn this week:

“We had nearly 300 meetings last week with F500 CMOs, agency leaders, and the biggest marketing organizations in the world, all grappling with the same reality: AI is now the most important referrer of brand perception, and most brands still have no idea how they show up. … Last week marked our first time on Cannes’ stage, and we left more convinced than ever that the category we’re establishing is where the whole industry is heading.”

Three hundred meetings sounds impressive, but it makes me wonder. If that many marketing leaders are asking the same question, why don’t we have better answers?

Even more illuminating is Cadwallader’s statement that his category is “where the whole industry is headed.” Of course, the CEO of a company valued at $1 billion would say that to The New York Times. But it also makes me wonder:

Where IS the industry heading when it comes to proving that AI visibility efforts work?

Share

AEO is repeating some of SEO’s history—but with a twist

Two datapoints illustrate the current situation:

1. Only 22% of US marketers believe they have fully integrated SEO and AI search across strategy, execution and reporting, according to an April 2026 survey by Adobe Semrush. That suggests one of two things: Either brands are not convinced of the need to integrate SEO and AEO, or they are finding it difficult to figure out how to do it.

2. Many companies are also actively blocking AI crawlers. Four in 5 websites block traffic from AI bots and crawlers, Nikhil Kolar, Microsoft’s VP of publisher product for AI, said while on stage at AdExchanger’s recent Programmatic AI conference. That suggests some marketers may be rushing toward AI optimization before they’re technically ready for it.

I don’t blame marketers for feeling out of sorts when it comes to AEO. In some ways, today’s AEO industry resembles the early days of SEO: hundreds of vendors, rapidly changing rules, uncertain metrics, conflicting advice and marketers struggling to understand how it all works.

SEO eventually sorted itself out, in part thanks to the arrival of guidance from Google that helped cut the clutter. But AEO is much more complicated because there are multiple AI platforms, each with their own methodologies. Even if guidance eventually emerges, it would be different for every AI platform and would change with each new model that comes out.

It’s time for marketers to start asking for more AEO evidence

I think we’ve reached the point where marketers need to ask deeper questions about what AEO is actually doing for them. It’s not because AI visibility isn’t important. It’s because the market is maturing faster than the evidence.

Before I go on, it’s important to note I have friends who run businesses in this space, including Pete Blackshaw of BrandRank, whose company I mentioned in my P&G Signal article and who writes a terrific Substack, The Answer Economy, about the business of showing up in AI. Gumshoe, whose dinner I attended in Cannes, is a past client. I’ve met with executives from Profound and other AI visibility companies in recent months.

But lately, I have been referring to AEO as a Sisyphean task. In Greek mythology, King Sisyphus angered the gods, and his punishment was to forever push a boulder up a hill, only to see it come crashing back down to the bottom.

Imagine if the hill kept moving, the boulder kept changing shape and the gods never revealed the rules. That’s the parallel to AEO.

Subscribe now

AEO is beginning to look like a business model built on continuous optimization. And the reward for that effort is that when things change, you need to start over. Do marketers want to be Sisyphus forever?

Three hard questions about AEO

This isn’t an argument for doing less. It’s an argument for demanding more evidence. I’ve already posed a few philosophical questions in this article, but now I want to get a little bit harder. These are the three questions I think everyone who has a stake in the AEO business needs to take seriously, now.

What actually moves the needle in AEO? The immaturity of the market has led to a land grab where competing vendor pitches all say different things: Build more credibility on Reddit. Invest in YouTube creators. Get more earned media. Fix your robots.txt. Update your schema. Keep track of 5,000 prompts. If every vendor has a different playbook, how is a CMO supposed to know which one is right?

Can anyone prove, without a doubt, that their AEO work is causing the outcome? Suppose a brand finds that it’s being cited in Claude twice as often as before. How does it separate whether it’s because it published new content, got more reviews on Reddit, or was covered in a few more news articles? Or perhaps Claude’s model changed, or its ability to access structured data improved?

What does winning in AEO look like? Visibility is becoming a common AEO KPI, but it risks becoming the AI equivalent of a social media vanity metric: likes. Marketers need to start asking what business result all those AI mentions actually achieved. More sales? More traffic? Greater purchase intent?

I’m not the only one asking questions. A recently published paper cautions that raw traffic growth figures that seemingly come from AEO efforts could be attributed more simply to “the rapid platform-level growth of the answer engines themselves.” Its conclusion: “headline AEO multiples substantially overstate causal effect.”

I firmly believe AI visibility is important. I believed it two years ago, and I still do today. But the industry’s immaturity is creating a scenario in which brands are pouring time, energy, and marketing dollars into work whose results remain difficult to prove.

I know that in social media (which I covered as an analyst for 17 years) it took a long time for measurement standards to take hold. I get that these are still relatively early days for AEO. But I also believe marketers shouldn’t need to wait years to get accurate metrics.

So if you’re investing in AEO, here’s my advice: Approach the market with curiosity and urgency, but until there are better ways to measure cause and effect and better standards for evaluating competing approaches, be prepared to keep pushing that boulder up the hill over and over again.

Leave a comment

THE NUMBERS BEHIND THE AI AD ECONOMY

In AI trust, North America trails other regions

Nearly two-thirds of consumers in the Middle East already trust AI to match products to their needs, according to new research by ESW. That’s far ahead of consumers in North America.

HEADLINES

Albertsons integrates branded product placement into AI-search tool (Marketing Dive) — Albertsons Media Collective is bringing sponsored product placement into its AI-powered conversational search tool through Criteo, letting brands appear in relevant product carousels as shoppers plan meals and build baskets. The broader signal: retail media is moving from keyword search into AI-mediated shopping moments.

Brands told to act fast on AI commerce (Axios) — Brands and publishers are being urged to restructure content, negotiate AI access deals and prepare for agents that reach content before humans do. Agentic commerce is forcing brands to think of content and data as infrastructure, not just marketing assets.

How 6 top CMOs like Coach’s Joon Silverstein are working to optimize their brands for AI platforms (Business Insider) — CMOs at brands including Coach, American Eagle, Chime and Bobbie are testing ways to influence visibility in ChatGPT, Gemini and other answer engines. As I mention in my main story, AEO/GEO is becoming a CMO-level concern, but the market is still early, opaque and full of vendor risk.

How Instacart is cultivating full-funnel marketing with video, AI (Marketing Dive — Instacart introduced new discovery-oriented ad products and outlined plans to experiment with ads inside its AI-powered shopping assistant later this year. Commerce platforms are preparing to monetize AI assistants, but the best players are being careful not to damage trust too early.

OpenAI says ChatGPT ad dismissals have dropped 50% as relevance improves (Search Engine Land) — OpenAI Chief Revenue Officer Denise Dresser said the rate at which ChatGPT users dismiss ads has fallen by 50% since the company launched advertising in February, a metric she says reflects improving relevance rather than simply higher engagement.

Subscribe now

WHAT’S DEBBIE UP TO?

Cannes Lions Edition

I had the huge honor of moderating a panel featuring executives from e.l.f., Ipsos, 360i and Asana at The Female Quotient’s FQ Beach. I just want to say, the FQ’s photographers are incredible!

I recorded a podcast about AI media with CMO Aaron Goldman on Mediaocean’s yacht (watch for the episode coming soon). Thankful for the high-powered fan they had going in between takes so we could cool off!

Beet.tv invited me to share my perspective on AI media and the power of AI platforms to intercept consumer intent. I also moderated a panel about the growth of agentic media buying with executives from OMD, Horizon and Chalice. Much appreciation to Andy Plesser, Lisa Granatstein and the terrific Beet.tv crew.

I added a bit of AI spin to a panel about pharma marketing and pharma retail media, alongside CVS’s Parbinder Dhariwal. Thank you Rob Rasko and 614 Group for the invite!

Finally, it was terrific fun to sit on a panel with my former EMARKETER colleague Andrew Lipsman, current EMARKETER analyst Sarah Marzano and Retail Media Breakfast Club writer Kiri Masters, moderated by Sensor Tower’s Ian Simpson. Thank you Sensor Tower for the invitation!

Watch my speaking reel

Debra Aho Williamson is the Founder and Chief Analyst of Sonata Insights, where she helps advertisers navigate the next major structural shift in media: the rapid rise and monetization of AI platforms. Over nearly two decades as an analyst at EMARKETER, Debra was among the first to forecast the growth of social media advertising and identify how it would fundamentally reshape marketing investment.

Today, she applies that same pattern-recognition lens to AI, analyzing how consumer adoption is accelerating monetization and redefining brand budgets, media strategy and industry power dynamics. Through Sonata Insights, she partners with leading platforms, agencies and technology companies to deliver research, advisory and strategic narratives that bring clarity and credibility to this transition.

Hello from Heathrow airport! I’m on my way back to Seattle after a packed week at Cannes Lions. Apologies this newsletter is a day late—I didn’t get much quiet time to write until today.

Today’s issue digs into OpenAI’s Cannes coming-out party, some interesting new AI ad formats from Amazon and Albertsons and four things about ads in AI, AEO, agentic commerce and consumer AI adoption that I heard over and over in Cannes. Since I’m racing to get this out, I’m skipping the chart of the week and news headlines.

If you were in Cannes too, I hope you had a safe trip home!

THE MAIN STORY

Here’s How AI Media’s Cannes Coming-Out Party Went

The Cannes Advertising Festival was buzzing about ads in AI this week. Here’s the news — and also, what I heard whispered in private meetings, dinners and parties.

OpenAI Debuts in Cannes

“More coffee than rose.” That’s how OpenAI’s head of ad sales, David Dugan, described the company’s first official presence as a media seller at Cannes Lions. No yacht. No beachfront activation.

In presentations and meetings with advertisers and agencies, Dugan and chief revenue officer Denise Dresser hammered home their key messages: Ads don’t affect the organic results. They’re clearly labeled as sponsored. And things are changing very fast. In just 4 months, the company has gone from “early pilot” to “early platform,” as Dugan put it. Ads just went live in Japan and South Korea, and Brazil and Mexico are coming soon.

A few other tidbits:

OpenAI not only wants to sell ads on ChatGPT, it also wants advertisers and their agencies to buy its creative tools to make and manage them. At the press briefing, OpenAI showed a demo of how agencies could use Codex and OpenAI’s workspace agents to develop creative assets and variations. One key finding Dugan shared about ChatGPT’s early ad performance is that the more creative variations brands use, the better results they see. “Creative volume will need to increase,” he said. “If you have too few, it’s probably not going to feel tailored enough; if you have enough volume, it feels tailored.”

Adtech partner Criteo now has 2,000 advertisers on ChatGPT, up from 1,000 a month ago. Adtech must evolve to be able to process semantic language and conversations, Criteo CEO Michael Komasinski said during a panel discussion with Dugan aboard Criteo’s yacht. “Keywords will go away,” he said bluntly.

This was a historic moment for OpenAI and the ad business. The big reveal is over. Now it’s time for OpenAI to prove it’s a partner worth investing ad budgets in.

Share

Amazon, Albertsons Introduce New AI-Native Ad Formats

I spend a lot of time thinking and writing about ads in ChatGPT, but AI media is much bigger than chatbots.

Two announcements from Amazon and Albertsons this week show how ads in AI-native experiences are expanding within retail media.

Amazon launched Alexa+ Agentic Ads, a format that lets users go from seeing an ad to making a purchase in the same voice conversation. Papa Johns is among the first to test the format, with an ad that walks users through a pizza order at a local store.

Grocery retailer Albertsons is starting to place product ads in its AI search feature, via its partnership with Criteo. I chatted with Jill Pavlovich, Albertsons’ senior vice president of digital experiences, in Cannes, and she told me the retailer is purposely going very slow in introducing product ads into conversational environments, preserving customer trust by ensuring ads respect dietary preferences and personalization data.

Subscribe now

AI Whispers In Cannes

Over the course of the week I heard a lot about ads in AI, AEO, and consumer adoption in my conversations with agency, brand and adtech executives. Here are four big things that kept coming up in conversations:

Many advertisers plan to test ads in AI media, but they’re taking a slow, deliberate approach. The stat from Criteo that I shared above says a lot—2,000 advertisers are already running ads through Criteo’s ChatGPT integration. But when I started asking around about allocating actual budget, no one was willing to say they were committing anything more than experimental dollars. The common refrain I heard: We’re watching performance closely and then we’ll decide. And when I checked in with larger advertisers (or their agencies), the word was “we’re waiting until Q3 or Q4 to decide.”

Here’s the interesting thing, though. That slowness leaves the door open for smaller, more nimble advertisers to step in. That’s what happened in the early days of social media; when Facebook opened self-serve, the kinds of advertisers that started popping up were definitely not household names. And a few upstart companies were able to build huge market share on the back of smart investments in social media ads.

The same could be the case in ChatGPT. Big brands, take note.

Agentic commerce is far off most advertisers’ radar. Several people told me they thought “agentic” was the word of Cannes this year. In my conversations, it didn’t come up much (except in the context of media buying, where agentic techniques are making huge strides since last year).

For Albertsons, customers so far aren’t using any agents to make purchases. “None of them are [taking their] hands off the wheel,” Pavlovich told me. Still, the retailer is building towards making those capabilities available, with an eye toward putting full control in users’ hands.

My point of view: Agentic purchasing will eventually be here, and at least some consumers will use it. It’s early, but there’s still valuable learnings to be had by experimenting where you can.

Brands are starting to get frustrated with their AEO efforts. Rules and guidelines don’t exist, tools keep changing and CMOs keep pestering when they try a prompt and don’t see their brand (or worse, they only see competitor brands). While most brands now realize the importance of visibility in AI platforms, they are having serious trouble seeing a path to consistent, positive results.

In my conversations, the topic that kept coming up was less about the “why” of AEO and more about the “how.” Specifically: “How do I turn this into a sustainable business activity that doesn’t suck all of our time and energy?” Good question…putting a pin in that for future exploration.

Gen Z’s AI skepticism is getting stronger—and that’s impacting how they relate to brands in AI. On the Female Quotient rooftop at the Hotel Martinez, I chatted with a 26-year-old brand strategist who told me she rarely trusts what she sees in AI conversations and “always goes to Google or Wikipedia” to check the answers. She said her friends were similar. When I asked if she’d trust a product recommendation from Claude or Gemini without doing her own research, she bluntly said, “No.”

That sort of sentiment comes up a lot in surveys. I’m planning to look into it more deeply in a future newsletter article.

Well, that’s it. My flight back to Seattle boards soon!

Leave a comment

THE NUMBERS BEHIND THE AI AD ECONOMY

Back next week!

HEADLINES

Back next week!

Subscribe now

WHAT’S DEBBIE UP TO?

In the News

Cannes Briefing: Cannes is selling AI transformation. Off stage the talk is about the bill (Digiday)

Podcasts

“Future-Proofing Brand Visibility in the Age of AI” with Kevin Lee of the eMarketing Association. We talk about about how AI is reshaping consumer behavior, ad revenue and SEO/GEO and what brands, agencies and publishers must do now to future-proof visibility in an AI-driven world.

Speaking

• RetailClub AI Festival, Sept. 24, 2026, Huntington Beach CA. I’m facilitating a conversation focusing on “When Chat Becomes the Channel: Advertising in the Age of AI Assistants.” Check out the full agenda here.

Watch my speaking reel

Debra Aho Williamson is the Founder and Chief Analyst of Sonata Insights, where she helps advertisers navigate the next major structural shift in media: the rapid rise and monetization of AI platforms. Over nearly two decades as an analyst at EMARKETER, Debra was among the first to forecast the growth of social media advertising and identify how it would fundamentally reshape marketing investment.

Today, she applies that same pattern-recognition lens to AI, analyzing how consumer adoption is accelerating monetization and redefining brand budgets, media strategy and industry power dynamics. Through Sonata Insights, she partners with leading platforms, agencies and technology companies to deliver research, advisory and strategic narratives that bring clarity and credibility to this transition.

When the ad industry’s leaders convene in Cannes, advertising on AI platforms will be a front-and-center conversation. I’ll be there, gathering intel and speaking on several panels. Want to meet up? Reply or send me a DM.

Here’s what’s inside The AI Ad Economy today:

• AI media enters Cannes: This year, the rise of ads in AI will be impossible to avoid.

• Who’s advertising on ChatGPT? Last week, I wrote that shopping-related brands accounted for nearly 40% of all ChatGPT mobile ad impressions in the past few months. The article generated a lot of questions on LinkedIn, so I dove back into the Sensor Tower ChatGPT ads database to find some answers.

• The numbers behind The AI Ad Economy: AI traffic is more engaged and worth more in revenue, new data from Adobe shows.

• Headlines: Facebook has a new AI-powered search feature, Claude now shows product cards, plus more AI ad economy news you need to know.

THE MAIN STORY

AI Media Gets Its Cannes Spotlight

Last year at Cannes, advertising in AI platforms barely registered as a topic of conversation. Instead, people were more focused on the plumbing: AI for creative development, for campaign planning or as a tool within the marketing organization.

Those topics will still be on the Cannes table next week, but the rise of ads in AI will be impossible to avoid.

Here are four new signals that AI media isn’t a future opportunity. It’s becoming a media channel today.

AI media becomes a forecasted ad category

WPP Media just issued its first global forecast for AI search ad spending, estimating that it will amount to nearly 40% of search ad spending by 2031. WPP’s forecast predicts that AI search will top $100 billion more quickly than any other major category, including search, social and retail media (a topic I wrote about last month).

EMARKETER also recently published its own US forecast and along with it, a new chart describing ads in AI as the “fourth wave” of digital media—on top of search, social and retail media. My former EMARKETER colleague Andrew Lipsman famously dubbed retail media the “third wave” of digital media.

Clearly, advertiser interest in AI media is growing. In Mediaocean’s just-issued 2026 H2 Market Report, 60% of respondents said they were planning to increase spending on AI media, or ads in AI agents. It was the strongest advertising growth area in the study, ahead of stalwarts like CTV, display, social, retail and search.

Of course, marketer interest doesn’t automatically translate to action, but other data in the Mediaocean study shows that marketers are paying close attention to consumer AI behavior, which is an important factor in the decision to advertise in AI media. AI was the top consumer trend for marketers in the study, cited by 75% (up from 70% in November 2025), while interest in other consumer trends fell, such as CTV (57% in May 2026, down from 63%) and social video (37%, down from 43%).

OpenAI arrives in Cannes as a media company

Media companies come to Cannes to attract advertisers and partners. OpenAI won’t have a yacht or splashy beach activation, but its executives will be pitching its ad business all over the Croisette.

On Monday morning, Criteo CEO Michael Komasinski and OpenAI’s head of global ads, David Dugan, will share early learnings from their partnership and discuss what’s next for AI-assisted commerce. Immediately afterward, OpenAI is hosting a press briefing at the Belle Plage hotel. Dugan and Denise Dresser, OpenAI’s chief revenue officer, will appear on other stages as well, including the official Cannes Lions program in the Palais, Shelly Palmer’s Innovation Breakfast and LinkedIn’s rooftop.

Conversational AI ads expand into new geographies and formats

ChatGPT ads are now available in the US, Canada, Australia and the UK (the latter as of a couple weeks ago), and Japan, Germany, Brazil and Mexico are coming soon. Amazon is extending conversational ads across its Alexa product suite, after merging its AI shopping assistant Rufus into Alexa and rebranding it as “Alexa for Shopping.” Google just started testing conversational discovery ads in AI Mode and expanded its Direct Offers pilot to include promotion bundling and native checkout.

As publishers and platforms look for new monetization opportunities, I expect conversational ads to spread rapidly this year.

Learn more about the power of conversational ads here:

GEO/AEO and paid AI media are converging

In prior digital media waves, brands had several years to get smart about organic before they ramped up paid. That also meant they treated organic and paid as separate, with separate teams.

That’s not what’s happening in AI media. Instead (as I predicted in this newsletter back in February), GEO/AEO and paid AI advertising are coming together. Here are just two examples:

• AI visibility companies like Profound are touting solutions for incorporating paid ads data into AEO.

• OpenAI is moving to bring paid and organic closer by linking company brand names directly to homepages—dramatically boosting referral traffic and building a valuable dataset of brand clicks (read more on LinkedIn here).

I expect a lot of chatter in Cannes about the true value of GEO/AEO, the need for AI-friendly landing pages for ads, and whether it’s a better strategy to buy ads that relate directly to a chat, or to try to conquest another brand. Linking organic AI and paid AI will also be on that discussion topic list.

The common thread behind all of these signals is clear: Last year, AI was primarily discussed as a utility for making marketing work better. This year it’s becoming a place where marketing happens.

Your questions answered: More from Sensor Tower’s ChatGPT ads data

After I published my first look at Sensor Tower’s new ChatGPT Ads Intelligence product last week, a few great questions came in on LinkedIn. So, I dove back into the database to find some answers.

“How has the shopping category’s share of ad impressions trended over time?”

The finding that caught the most attention in my article was that 40% of mobile ad impressions between February 10-May 24, 2026, were from shopping-related brands. Several people wanted to know more about how that share changed over time.

What I discovered was this: While shopping brands are still the top ad category on ChatGPT, their share has declined since February. Shopping brands represented 46.8% of ad impressions between February 10-28, 2026. But by May, their share had fallen to 31.9%.

What changed? Two categories increased their presence and two more joined the mix.

• Software advertisers grew from a 7.7% share in February to 13.1% for the month ended May 31.

• Media advertisers increased from 7.5% to a 10% share in the same period.

• Two new categories—business & industrials and financial services—began advertising on ChatGPT, accounting for 5.9% and 6.1% of ad impressions in May, respectively.

What this means: ChatGPT advertising is diversifying quickly. Retailers were among the earliest adopters (perhaps for the learning opportunity or out of concern about disruption), but other categories are ramping up fast.

“How much of this [advertising] is retailer-funded vs supplier-funded offsite retail media?”

— Courtland Dearing

The Sensor Tower data doesn’t capture where the ad budgets come from, but it does show the headline and text of each ad. Based on my scan of the top retail advertisers during the period, virtually all creative executions used by top advertisers such as Best Buy or Home Depot were product ads. I saw no ads for the retail brand itself.

This meshes with ChatGPT’s focus on delivering product-related advertising. People researching products will likely be more responsive to an ad for a product related to their query, rather than an ad for a retailer that sells the product.

For retail media networks, this suggests an interesting possibility. If conversational environments become strong venues for product ads, retailers may have a new incentive to develop conversational features of their own and monetize them with ads.

“What you’re seeing in this early going is that Home Depot is seeing some conversions (hence the steady increase across the month), and Best Buy is not (hence the drastic reduction in spend). I’ll be interested to see the June trends,”

— Bill Bledsoe

What this question is referring to is the fact that Best Buy and Home Depot used very different strategies to test ChatGPT ads. Best Buy went big out of the gate, while Home Depot took a slow and steady approach.

The Sensor Tower dataset doesn’t assess ad performance, so it can’t address the first part of Bill’s question.

What it does do is show advertiser participation and the intensity of spending, based on ad impressions. Best Buy accounted for a whopping 28.3% of all ad impressions on ChatGPT in the week ended February 15, 2026. But by mid-May, Sensor Tower was no longer detecting any Best Buy ads at all.

A peek at the June trends shows that Home Depot has now risen to become the No. 1 advertiser on ChatGPT (based on data from June 1-14), with a 3.1% share of ad impressions.

One of the most interesting aspects of Sensor Tower’s ChatGPT ad product is the visibility it provides into how advertiser participation ebbs and flows over time. That kind of visibility didn’t exist when Facebook first launched.

It’ll be interesting to see whether advertisers come in, test ads and leave, or if they stay and increase their commitments to ChatGPT. That will tell me a lot about the future monetization potential of this platform.

Leave a comment

THE NUMBERS BEHIND THE AI AD ECONOMY

New Data on AI’s Impact on Conversion, Engagement and Revenue

Adobe is out with yet another of its quarterly AI Traffic Trends reports and the impact of AI keeps getting stronger. Here are four notable stats from the report:

• “In May 2026, AI-referred retail visitors converted at a rate 54% higher than non-AI traffic.”

• “AI-referred retail visitors showed 15% higher engagement than non-AI traffic in May 2026, the strongest advantage recorded since tracking began.”

• “AI-referred retail visitors spent 53% more time on site than visitors from other channels.”

• “AI-referred retail visitors generated 53% more revenue per visit than non-AI traffic in May 2026, the strongest RPV advantage on record and a sharp climb from 37% in March.”

HEADLINES

AI Advertising Signals

OpenAI investigated by coalition of state attorneys general (WSJ) — A coalition of state attorneys general is investigating OpenAI, with document requests covering advertising strategies, user engagement, retention, consumer and health data, and treatment of minors and seniors. The probe comes as OpenAI’s ad and commerce ambitions are becoming more central to its business model.

AI Discovery & Platform Signals

Facebook’s new AI Mode search gets its info from public posts (The Verge) — Meta introduced an AI-powered search mode in Facebook that answers questions using public posts, Groups, Reels and content from across Meta apps. This turns social content into an AI discovery layer, not just a feed or search archive.

A Court Has Ruled That Google Is Liable for False Statements Generated by AI Overviews (Wired) — A Munich court ruled that Google can be held responsible for false statements generated in AI Overviews, after publishers said the summaries linked them to scams and disreputable activity. Google plans to appeal, but the case highlights a new accountability problem for AI search.

Data and Measurement Signals

Can ChatGPT Ads Get People to Buy? OpenAI Partners With LiveRamp to Prove They Do (Adweek) — OpenAI has partnered with LiveRamp to give ChatGPT advertisers access to conversion measurement that connects ad exposure inside ChatGPT to real-world purchases. Through LiveRamp’s Conversions API Hub, brands can match campaign data with transaction outcomes, helping determine whether ChatGPT ads drive measurable sales rather than simply generating clicks or engagement.

80% of ChatGPT product recommendations change when search is enabled: Study (Search Engine Land) — A study of 20,000 ChatGPT responses found that product recommendations changed 80.2% when search was enabled, with recommendation overlap falling to 19.8%. That suggests AI shopping results can vary sharply depending on whether the model relies on live retrieval or its underlying model knowledge.

AI Commerce Signals

BREAKING: Claude Has Product Cards (Retailgentic newsletter) — Trackers from agentic commerce expert Scot Wingo’s ReFiBuy discovered that Anthropic’s Claude is now showing product cards in response to certain queries. The cards are “pervasive across categories,” Wingo writes. This move puts Claude more explicitly in the mix of product research and could be a step toward a more explicit commerce play.

Subscribe now

WHAT’S DEBBIE UP TO?

CANNES LIONS EDITION

I’ll be on stages up and down the Croisette this year. Here’s where you can catch me (or you can look for me in my snazzy new The AI Ad Economy hat).

Monday June 22, 3-6pm: “Meet the People Defining Commerce Media’s Next Era” (Sensor Tower, CPG Guys and EMARKETER)

I’m excited to join independent analyst Andrew Lipsman (my former EMARKETER colleague), current EMARKETER analyst Sarah Marzano and Retail Media Breakfast Club’s Kiri Masters in a panel discussion that’s part of a 3-hour event focusing on retail media, AI and agentic commerce.

Tuesday June 23, 3-4pm: Beet.tv’s AI Summit

I’ll be presenting “Beyond Performance Marketing: AI and the New Consumer Decision Journey” and then moderating a panel discussion called “AI, Automation & the New Performance Playbook,” featuring Michael Cohen, EVP of performance media at Horizon Media; Bradley Rogers, CEO of OMD US; and Ali Manning, COO of Chalice AI.

Wednesday June 24, 2-2:30pm: “The New Prescription — How Pharma Marketing Is Being Redefined,” from 614 Group.

I’m moderating a panel discussion featuring Parbinder Dhariwal of CVS Media Exchange, Kelly Millett of PHM and Khara Hutchinson of Bayer.

View the full 614 Group event schedule and request an invite here.

Thursday June 25, 9:30am: “The Human Advantage: Creativity in the Age of AI” at FQ Beach at Cannes Lions

I’ll be discussing how companies can embrace AI in the creative process, how to maintain cultural nuance and authenticity when we use AI in marketing, and how AI can help unlock new ways of being creative rather than replacing current methods with Ekta Chopra, Chief Technology & AI Officer, E.L.F. BEAUTY; Christine Cotter, Managing Director, 360i & Head of Social Innovation, dentsu; Lindsay Franke, President & Deputy Chief Executive Officer, Ipsos North America; and Prachi Gore, Chief Marketing Officer, Asana.

Watch my speaking reel

Debra Aho Williamson is the Founder and Chief Analyst of Sonata Insights, where she helps advertisers navigate the next major structural shift in media: the rapid rise and monetization of AI platforms. Over nearly two decades as an analyst at EMARKETER, Debra was among the first to forecast the growth of social media advertising and identify how it would fundamentally reshape marketing investment.

Today, she applies that same pattern-recognition lens to AI, analyzing how consumer adoption is accelerating monetization and redefining brand budgets, media strategy and industry power dynamics. Through Sonata Insights, she partners with leading platforms, agencies and technology companies to deliver research, advisory and strategic narratives that bring clarity and credibility to this transition.

Hello from Seattle! I’m happy to be back home for several days before the Cannes Lions craziness starts. I’ll be participating in panels and events with Sensor Tower/EMARKETER (Monday), Beet.tv (Tuesday), Mediaocean (Wednesday), The Brand and AI Safety Summit (Wednesday) and The Female Quotient (Thursday). Get all the details at the bottom of the newsletter.

Today you’re getting a first look at data from Sensor Tower’s new ChatGPT Advertising Intelligence product (currently in beta). The number of advertisers that have started testing ads on ChatGPT is much higher than I expected, showing that ChatGPT has already crossed an important threshold: From early-adopter experiment to significant testing across a broad base of advertisers.

THE MAIN STORY

More than 1,400 Brands Have Tested ChatGPT Ads Since Launch

Retailers are among the businesses most threatened by ChatGPT’s growing role in product discovery, but they are also the platform’s biggest advertisers.

Sensor Tower data provided to The AI Ad Economy shows that shopping-related brands accounted for nearly 40% of all ad impressions between February 10 and May 24, 2026, making them by far the largest ad category on the platform.

The data also reveals that more than 1,400 brands have already advertised on ChatGPT in the US, across nearly every major industry category.

While that represents only about 1% of the roughly 135,000 brands that have advertised digitally in the same timeframe, per Sensor Tower’s tracking, the findings suggest that experimentation with ChatGPT is quickly spreading across multiple sectors that want to understand how advertising in AI-native platforms can influence purchase behavior.

The number of ChatGPT advertisers is growing quickly

Sensor Tower identified a total of 1,419 unique brands that have advertised to US audiences on ChatGPT’s Android mobile app in the February 10-May 24, 2026 timeframe.

Brand participation in ChatGPT has risen dramatically in just a few months.

In the first week after the ads launch, Sensor Tower tracked 30 advertisers on the platform. By the week ending March 1, there were 117 advertisers—significant but not notable.

By early April, the total tracked advertiser count ballooned to over 600 and then topped 950 by the start of May.

Advertiser activity has ramped up even more in the past few weeks, after OpenAI introduced new ad tools and opened ads to more advertisers. In the week ending May 24, 2026, Sensor Tower tracked 266 new brands advertising on the platform, almost four times the number of new entrants tracked in the week ending May 10.

That increase followed the introduction of a wave of new advertising capabilities from OpenAI, including a self-serve ads manager, CPC bidding and conversions API, as well as product feed ads and new ecommerce creative formats.

Retailers are embracing ChatGPT

Shopping brands tower over other categories when it comes to advertising on ChatGPT. These brands accounted for 39.5% of tracked ad impressions in the February-May time period. Software companies ranked second (12%), followed by media & entertainment firms (10.3%).

However, brands from nearly every major industry category are represented in the database, including travel and tourism, which accounted for 4.6% of ad impressions, auto (3.2%) and CPG (1.8%).

Top advertisers are testing different ad strategies

More than half of the top 10 advertisers on ChatGPT are retail brands. Best Buy was the biggest advertiser during the February 10-May 24 time period, with a 2.43% share of ad impressions. Target, SCHEELS and Home Depot ranked next.

Digging deeper into the Sensor Tower data, I’m seeing diverging strategies for testing ChatGPT ads. Best Buy went big in the first weeks after the ad launch, accounting for a whopping 28.3% of ad impressions in the week ended February 15, 2026 (the first week ads were shown). By early April, Best Buy’s share had fallen below 2%, and it trailed off to zero by mid-May.

Home Depot, on the other hand, maintained a lower but relatively steady presence. Its share of weekly ad impressions averaged under 1% through early May, but then began increasing, surpassing a 4% share by the end of the month.

The remaining advertisers in the top 10 were Speechify, a text-to-speech AI assistant; Preply, an online language education business; Top10.com, an online comparison company; and Jotform, an online survey provider.

What Does this Data Mean for the Growth of AI Media?

ChatGPT ads are attracting far more advertiser interest than skeptics expected. The fact that more than 1,400 brands have already started to place ads on ChatGPT shows that AI media—AI-native platforms and experiences where consumers research and make purchase decisions—is something advertisers are very open to testing. Advertisers are clearly eager to understand how AI-native media environments work.

Retailers are benefiting from early learnings. I’m not surprised that shopping-related brands have been the strongest to embrace ads on ChatGPT. In January, when I published an analysis predicting which industry categories should be the first to advertise in AI media, retailers were one of the four categories I identified.

One reason retailers have emerged as early adopters of ChatGPT ads is that they fear losing visibility into consumers’ research and purchase behavior. They know that AI platforms can intercept shopping behavior. Advertising on ChatGPT gives them visibility into how consumers research, what triggers them to click on ads (versus just reading the organic responses) and the types of messaging that can influence them.

Tech advertisers are also gaining first-mover advantage. The fact that software brands accounted for 12% of ad impressions shows that they are embracing the potential for conversational discovery as an advertising vehicle. In my January analysis, I advised tech advertisers to put AI media on the front burner. These advertisers rely heavily on research-driven discovery, so there is significant overlap with AI use cases such as comparison, troubleshooting and buying guidance.

Media brands are a surprising entrant. What I didn’t anticipate was how quickly media & entertainment brands have moved to test ChatGPT advertising, accounting for 1 in 10 ad impressions in the Sensor Tower dataset. Just like retailers, these companies are also wary of ChatGPT’s impact on referral traffic. It’s possible they are testing ChatGPT ads to gain insight into how they work so they can mount a defense. If that’s the case, it’s a smart strategy.

Another explanation could be related to understanding how discovery works in conversational environments. If ChatGPT becomes a significant discovery layer, media companies need to understand it.

No single advertiser dominates ChatGPT. Best Buy started out dominant, but it ended up accounting for just 2.43% of impressions during the period as other advertisers entered and its initial burst of advertising slowed. The data suggests that there is broad experimentation across brands and categories rather than concentrated spending by a few.

The data measures advertiser participation, not advertiser success. What Sensor Tower’s data shows is that interest in ChatGPT advertising is substantial, given the newness of the ad offering, but it’s also in the early stages, since only about 1% of tracked brands have advertised there so far.

In addition, the industry categories that are investing first—and strongest—are among the ones that are most vulnerable to the shift in discovery patterns that AI platforms are creating. They are gaining early visibility into a discovery environment that has the potential to sit between them and their consumers.

What the data doesn’t show is if advertisers are getting results from these ads. Whether they continue spending will depend on performance and consumer acceptance.

To sustain that momentum, OpenAI will need to continue expanding its measurement and optimization capabilities. But for now, the initial interest in ChatGPT ads is significant.

If you’ll be in Cannes, I’ll be discussing this data as well as how AI platforms are affecting commerce media alongside analysts Andrew Lipsman, Sarah Marzano and Kiri Masters at the EMARKETER/Sensor Tower event “Meet the People Defining Commerce Media’s Next Era.” It’s on Monday June 22, from 3-6:30pm at The CPG Guys’ residence. Register your interest here.

Leave a comment

THE NUMBERS BEHIND THE AI AD ECONOMY

How Much Do People Trust AI Across Product Categories?

When it comes to recommending products or services, confidence in AI chatbots is fairly uniform across age groups. US adults are generally more likely to trust recommendations for things like electronics, entertainment/media or appliances and generally less likely to trust recommendations for medical- and financial-related products or services, according to Ipsos. People 55 and up are less trusting, overall than than younger age groups; in fact, trust levels in the 18-34 and 35-54 groups are remarkably similar.

HEADLINES

AI Advertising Signals

ChatGPT ads land in UK as OpenAI outlines EU privacy rules — OpenAI is expanding advertising while adapting to Europe’s stricter privacy environment. AI advertising is moving beyond testing and beginning to confront the same regulatory constraints that shaped digital advertising.

OpenAI thinks ChatGPT will collect about $50bn in US ad revenues in 2030. We think their entire addressable market will be 1/10th that — A new EMARKETER forecast for US AI ad spending predicts that “AI ads will more than double from over $32bn in 2026 to more than $68bn in 2030. But more than 80% of AI advertising in 2026 will appear next to AI content, such as traditional search ads next to Google AI Overviews, rather than inside AI chatbots or conversations,” EMARKETER analyst Nate Elliott writes on LinkedIn.

AI Commerce & Agentic Signals

Visa to secure payments for shoppers on ChatGPT in OpenAI partnership — OpenAI is extending from product discovery toward transaction execution. Control of payments may become as strategically important as control of recommendations in the emerging agentic commerce stack.

Emerging Players & Ecosystem

Perplexity planning IPO by 2028 regardless of what happens to Anthropic or OpenAI, CEO says — Perplexity is positioning itself as an independent platform rather than a feature acquisition target. The signal is growing confidence that AI-native discovery companies can support durable standalone businesses.

Google gets a dose of social media with Search Profiles to capture new audiences — Search Profiles create direct relationships between Google and creators. As AI-generated answers reduce clicks, Google appears to be strengthening incentives for content creation inside its own ecosystem.

How publishers are modeling and mitigating a future with significantly less Google Search traffic — Publishers are shifting from temporary traffic declines to long-term planning assumptions. The signal is that AI-driven referral loss is increasingly being treated as a structural market change rather than a cyclical disruption.

Data & Performance Signals

Google’s AI opt-out leaves publishers with a choice they can’t safely use — People Inc. lost 800 million Google visits between Q1 2024 and Q1 2026. Publishers face growing dependence on AI visibility even as AI-generated answers reduce referral traffic.

Subscribe now

WHAT’S DEBBIE UP TO?

SPECIAL CANNES LIONS EDITION

I’ll be on stages (and yachts) up and down the Croisette this year. Here’s where you can catch me:

Monday June 22, 3-6pm: “Meet the People Defining Commerce Media’s Next Era” (Sensor Tower, CPG Guys and EMARKETER)

I’m excited to join my former EMARKETER colleague Andrew Lipsman, current analyst Sarah Marzano and Retail Media Breakfast Club’s Kiri Masters in a panel discussion that’s part of a 3-hour event focusing on retail media, AI and agentic commerce.

Register your interest here.

Tuesday June 23, 3-4pm: Beet.tv’s AI Summit

I’ll be presenting “Beyond Performance Marketing: AI and the New Consumer Decision Journey” and then moderating the panel discussion “AI, Automation & the New Performance Playbook,” featuring Bob Lord, president of Horizon Media; Bradley Rogers, CEO of OMD US; and Ali Manning, COO of Chalice AI.

Check out the full Beet.tv Cannes agenda here.

Tuesday June 23, 8:30pm: Mediaocean dinner

I’ll be delivering an opening provocation focused on my AI media research at a private dinner hosted on Mediaocean’s yacht.

Wednesday June 24, 2:15-3pm: “Closing the Loop — AI, Programmatic, and Measurement in Pharma,” from 614 Group and the Brand and AI Safety Summit

I’m moderating a panel discussion featuring Parbinder Dhariwal, VP & GM, CVS Media Exchange, and Anvita Karara, VP, Commercialization, BMS.

View the full Brand and AI Safety Summit and request an invite here.

Thursday June 25, 9:30am: “The Human Advantage: Creativity in the Age of AI” at FQ Beach at Cannes Lions

I’ll be discussing how companies can embrace AI in the creative process, how to maintain cultural nuance and authenticity when we use AI in marketing, and how AI can help unlock new ways of being creative rather than replacing current methods with Ekta Chopra, Chief Technology & AI Officer, E.L.F. BEAUTY; Lindsay Franke, President & Deputy Chief Executive Officer, Ipsos North America; and Prachi Gore, Chief Marketing Officer, Asana.

Register your interest here.

Will you be in Cannes too? If you want to meet up, just reply to this email and we’ll make it happen!

Watch my speaking reel

Debra Aho Williamson is the Founder and Chief Analyst of Sonata Insights, where she helps advertisers navigate the next major structural shift in media: the rapid rise and monetization of AI platforms. Over nearly two decades as an analyst at EMARKETER, Debra was among the first to forecast the growth of social media advertising and identify how it would fundamentally reshape marketing investment.

Today, she applies that same pattern-recognition lens to AI, analyzing how consumer adoption is accelerating monetization and redefining brand budgets, media strategy and industry power dynamics. Through Sonata Insights, she partners with leading platforms, agencies and technology companies to deliver research, advisory and strategic narratives that bring clarity and credibility to this transition.

Can I just say: Chicago in early summer is absolutely blissful. I lived here for 12 years (for college and then my first job), and all the memories have been flooding back while I’m in town to speak at Path to Purchase Institute’s Retail Media Summit. Last time I was in the Grand Ballroom at the Palmer House was my Northwestern University sorority formal. Kind of a different vibe…!

In today’s issue, I’m shedding light on a looming problem for retail media platforms and their advertisers: AI platforms are changing the path to purchase by translating consumer intent before traditional digital advertising ever gets a chance to see it.

THE MAIN STORY

The Signals I Sent ChatGPT (That Ads Did Not See)

There I was one Saturday morning, right before my birthday, standing in front of my bathroom mirror and thinking, “Your eyes look terrible. You have dark circles, bags, wrinkles, all the things. Maybe you need to start using eye patches.”

I went downstairs, got my breakfast and coffee, opened my phone and flipped to my Instagram Stories. And there, sandwiched in between my friends’ missives, was Gwyneth Paltrow staring up at me, wearing eye patches, in an ad for her beauty brand Goop.

This isn’t a story about why I decided to buy Gwyneth’s patches after seeing that Instagram ad. Nor is it a story about how social media advertising has become incredibly, suspiciously well-targeted (apparently your phone now listens to your inner thoughts, not just what you’re saying).

This is a story of how I discovered an entirely new brand, researched it and bought it—all without ever seeing any of the traditional digital advertising that brands have used for decades.

It’s also a lens into a looming problem.

AI platforms do more than influence purchases. They intercept the purchase journey—and along with it, the intent signals advertisers and retail media platforms normally see. That’s a challenge brands like Goop and retail media platforms are only beginning to grapple with.

A Social Media Ad Leads Me On An Unexpected Journey

It’s true. The Instagram ad for Goop does pique my interest. But instead of clicking it, I open ChatGPT and type: “Tell me about the Goop under eye pads. … I have deep set eyes with dark circles and some puffiness.”

What happens next sends me down a completely different path. ChatGPT immediately, without hesitation, tells me that no, these pads are NOT what I need. “Good for temporary changes, not structural improvement,” it says.

Hmm.

Then ChatGPT transforms into my personal beauty advisor. What you really need, it says, is this other product…The Ordinary Caffeine Solution 5% EGCG.

Huh. I’ve never heard of The Ordinary. I’ve never tried caffeine under my eyes. I have no idea what EGCG means. But I’m intrigued. ChatGPT goes on to say I also need retinol (I bypass that – I’m sensitive to it). And then, confidently, it advises that I also need yet another product by The Ordinary.

At this point, I’m convinced. I veer off to Amazon to research prices (I could have done this part on ChatGPT, but I’m a Seattle girl and I love Amazon). The products are incredibly inexpensive, so I go ahead and buy both.

After using The Ordinary for a few weeks, I have to say ChatGPT might have been right. My eyes look a lot better.

The Retailer Still Captures the Transaction, But the AI Platform Shapes the Decision

Here are the signals I sent ChatGPT that ads never had a chance to see:

1. I have a concern: I don’t like the way my eyes look.

2. I’m open to new products.

3. I’m ready to buy.

If you think this isn’t happening very often, think again: A new study by Pacvue found that 28% of US respondents (44% of whom had incomes of $100,000 or above) use AI tools daily for shopping research. And when shoppers use AI tools, 64% see a new product in-session, IAB and TalkShoppe found.

Many of those consumers will go on to buy that new product, just as I did.

Retailers and brands still have valuable customer relationships and transaction data, and retailers may still end up getting the sale. But the decision layer is shifting toward AI platforms.

Share

What This Means for Retail Media Companies and Their Advertisers

As more consumers use AI platforms for these sorts of decisions, intent formation is migrating out of the reach of traditional media environments—including the retail media destinations that many advertisers rely on. Search platforms see intent. Amazon sees intent. Retailers see intent. But increasingly, AI platforms are seeing intent before any of them.

You might argue that search pretty much always saw intent before retailers did. But the signals it saw were expressed intent. What consumers are giving AI platforms is different: Feelings. Motivations. Preferences. Questions.

In my situation, I didn’t ask ChatGPT what to buy. I told it I had a problem and I asked if the product I thought I wanted to buy would fit my needs.

Strategically, that means the following:

1. Understand how and where AI systems are forming opinions about your brand. Optimizing for AI visibility isn’t an option anymore, it’s a requirement.

2. Ensure your systems capture not only expressed intent, but implied intent. If you’re only mining keywords, you’re only getting partial insight. Put conversational signals on the analytics roadmap.

3. Don’t skimp on customer loyalty. I did all the research through ChatGPT but chose to go to Amazon for the purchase because I’m a loyal customer. Be the brand consumers think of when they are ready to purchase.

The other important takeaway is this:

Even if AI platforms can and do see intent before other media can, who owns the customer journey in this new scenario is still up for grabs.

Yes, AI platforms can take people from brand awareness to purchase in a single destination. But consumer behavior is typically more nuanced. People still move between AI and traditional media, and purchases are still more likely to take place with a retailer or marketplace than on an AI platform.

My experience is a perfect example: Instagram started it, ChatGPT took me in a new direction, The Ordinary won my consideration and Amazon took the sale.

No one company owned the entire journey. But one company saw the most important signals.

And that’s the challenge for brands and retail media companies. AI platforms don’t need to own the customer relationship to gain influence. They only need to intercept the motivations, questions and preferences that shape the decision.

Read next:

Leave a comment

THE NUMBERS BEHIND THE AI AD ECONOMY

Parents With Kids Under 18 Lean Into AI

Looking for AI early adopters? Look no further than US parents. They are much more likely than people who aren’t parents or guardians of under-18s to use AI for a wide range of use cases and tasks, according to a study by YouGov. One factor may be age (parents of under-18s could be younger than those without children of that age). Another may be time-savings; AI can help busy people, including parents, get things done.

HEADLINES

AI Advertising Signals

Meta Has Struggled at Selling Anything Other Than Ads. Will AI Be Different? (CNBC) — Meta is testing whether AI can become a second revenue engine alongside advertising. The broader signal is that major AI platforms are converging on hybrid monetization models rather than relying solely on ads.

OpenAI is working with Skai to bring retail and commerce advertisers into ChatGPT (Digiday) — Skai gives commerce advertisers an existing buying path into ChatGPT. AI ads are moving from pilot inventory to platform-integrated performance channels.

Google begins testing healthcare ads in AI Mode (Search Engine Land) — Google is extending AI Mode monetization into one of digital advertising’s most regulated categories. The signal is growing confidence that conversational search can support high-value verticals beyond commerce and retail.

AI Commerce & Agentic Signals

Amazon offers AI agent tech to other retailers (Retail Dive) — Amazon is turning its shopping assistant infrastructure into a retailer-facing product. The agentic commerce fight is shifting from consumer interfaces to who supplies the underlying retail stack.

Emerging Players & Ecosystem

This startup is taking on TikTok Shop and eBay Live with AI-powered auctions. Read its $26 million pitch deck. (Business Insider) — Tilt’s $26 million round points to AI-native commerce beyond chatbots. AI is being embedded into listing creation, seller guidance, search and social distribution.

Phoebe Gates’ App Raises $35 Million as Khloé Kardashian Says She Was Inspired to Invest by Mom Kris Jenner (People) — Phia’s 1.5 million users and 9,600 brand partners show AI shopping startups gaining consumer scale. Discovery is fragmenting into assistants that compare price, resale value and availability.

Subscribe now

WHAT’S DEBBIE UP TO?

On Stage

• Path to Purchase Institute’s Retail Media Summit, June 3, 2026, Chicago. I just delivered “Who Owns the Customer in the Age of AI-Fueled Commerce?”.

• Sensor Tower, CPG Guys and EMARKETER event, June 22, 2026, Cannes. I’m excited to join my former EMARKETER colleague Andrew Lipsman, current analyst Sarah Marzano and Retail Media Breakfast Club’s Kiri Masters for a 3-hour event focusing on retail media, AI and AI commerce.

• Beet.tv’s The New Age of AI, June 23, 2026, Hotel Majestic, Cannes. I’m hosting and moderating a panel discussion “AI, Automation & the New Performance Playbook.” Check out the full Beet.tv Cannes agenda here.

• FQ Beach at Cannes Lions, June 25, 2026. I’m moderating “The Human Advantage: Creativity in the Age of AI” for The Female Quotient. Register your interest here.

• RetailClub AI Festival, Sept. 24, 2026, Huntington Beach CA. I’m facilitating a conversation focusing on “When Chat Becomes the Channel: Advertising in the Age of AI Assistants.” Check out the full agenda here.

Watch my speaking reel

Podcasts

• “AI is the New Ad Space” with AdExchanger: In a conversation with Allison Schiff, we explore why AI media can be powerful in the middle of the funnel, what organizational changes brands should make to be AI-ready and what needs to happen to make OpenAI’s $100 billion ad forecast feasible.

Research

• IAB + Sonata Insights: The AI Ad Gap Widens (read our 2024 study here)

• Mediaocean: 2026 Advertising Outlook Report (I wrote the foreword)

Debra Aho Williamson is the Founder and Chief Analyst of Sonata Insights, where she helps advertisers navigate the next major structural shift in media: the rapid rise and monetization of AI platforms. Over nearly two decades as an analyst at EMARKETER, Debra was among the first to forecast the growth of social media advertising and identify how it would fundamentally reshape marketing investment.

Today, she applies that same pattern-recognition lens to AI, analyzing how consumer adoption is accelerating monetization and redefining brand budgets, media strategy and industry power dynamics. Through Sonata Insights, she partners with leading platforms, agencies and technology companies to deliver research, advisory and strategic narratives that bring clarity and credibility to this transition.

Hello from Seattle, where Mark Zuckerberg’s superyacht has just arrived amid heckling and tomato-throwing threats. I’m enjoying a few days at home in between speaking gigs. If you’ll be at Retail Media Summit next week, be sure to catch my presentation “Who Owns the Customer in the Age of AI-Fueled Commerce?” on June 3rd. I used to live in Chicago and I can’t wait to go back for the Path to Purchase Institute’s awesome event. Check out the full agenda here.

Next month it’s on to Cannes, and I’m excited to announce I’m partnering with Beet.tv for “The New Age of AI,” a 2-hour event on June 23 at the Hotel Majestic. Learn more here.

This week’s issue is adapted from a recent presentation I gave on how AI is reshaping media monetization—and what that means for advertising. In previous digital transformations, the publisher monetization infrastructure lagged consumer behavior by years. But that’s not what’s happening in AI media.

THE MAIN STORY

The AI Monetization Shift: Why Media Won’t Get 8 Years This Time

I spend most of my time studying one question: What happens to advertising and media when consumer behavior shifts faster than the industry infrastructure around it? I think we’re entering one of those moments right now. Not just because of AI itself, but because AI is compressing the amount of time media companies have to react.

In prior platform shifts, the media industry had breathing room. In the early internet era, for instance, there was chaos, certainly. But there was also time.

Time for experimentation.
Time for business models to emerge.
Time for buyers and sellers to figure out standards.

The monetization infrastructure lagged consumer behavior by years.

Search took years to monetize. Social took even longer. Amazon took nearly two decades to fully build its ad machine. ChatGPT? Roughly 3 years from launch to ads.

This is why I don’t think AI is just another platform shift. It’s a compression event.

Al is creating massive shifts in consumer adoption, discovery behavior, monetization, commerce and advertising infrastructure in a much shorter time frame than this industry is accustomed to.

One of the biggest mistakes in media right now is treating AI platforms as utilities. Yes, they help you get lots of things done. But they are also becoming destinations.

And once a platform becomes the place where discovery happens, it starts influencing what gets seen, what gets recommended, what gets trusted and eventually what gets purchased.

That’s media power.

Share

Consumers are normalizing AI-assisted decision-making much faster than many media organizations internally appreciate. And if AI platforms become the front door, media companies risk becoming supply.

So the strategic question becomes: How do you stay inside the decision layer?

When decisions move into AI systems, several familiar assumptions begin breaking down simultaneously: Forecasting. Yield dynamics. Measurement.

Historically, digital media forecasting relied heavily on visible demand signals: things like traffic patterns, search behavior and referral flows. But if intent is increasingly intermediated by AI, some of that visibility weakens. You may still see outcomes. But you lose visibility into how the decision formed.

One of the biggest risks for the media industry is invisible influence. You still see conversions and outcomes, but fewer observable signals in the middle. And if you can’t fully see the decision journey, forecasting, pricing and attribution become harder.

The risk to media companies isn’t simply traffic disruption. It’s role disruption. AI changes who participates in the transaction itself and who owns the relationship.

In previous cycles, the media industry had time to observe. To debate. To wait. This time, everything is happening simultaneously. That’s why waiting is no longer a neutral strategy. Because while media companies are still evaluating the shift, the AI platforms are already monetizing it.

I don’t think AI eliminates the need for media companies. But I do think it changes where value accrues. And the companies that adapt fastest to the decision layer will have disproportionate influence over the next era of advertising.

Leave a comment

THE NUMBERS BEHIND THE AI AD ECONOMY

Gen Z’s AI skepticism is growing

I’m seeing lots of studies about Gen Z’s AI skepticism lately. Part of me thinks this is a natural pendulum, since periods of negativity tend to follow periods of high adoption and usage. What stands out for me, however, is Gen Z’s concern about how AI impacts critical thinking.

40% of Gen Z respondents to a survey by GoTo and Workplace Intelligence said they can’t get by without AI at work, but 62% believe they’ve become too reliant on it and 46% believe AI is eroding their skills and making them less intelligent. The survey was conducted November 2025-January 2026 among 2,500 knowledge workers and IT decisionmakers in the US, Canada, the UK and seven other countries.

HEADLINES

AI Advertising Signals

OpenAI is building a new kind of ad machine. Early data suggests Google should worry (Business Insider) — ChatGPT ads are monetizing conversational intent, not keywords. The structural shift is that demand can be created mid-dialogue, before users express commercial intent.

OpenAI is preparing conversion-focused ads for ChatGPT (Search Engine Land)— OpenAI is moving from brand pilots toward leads, bookings and sales. ChatGPT ads are being positioned as performance media, not experimental sponsorship inventory.

Google tests new conversational ad formats in AI Mode and Search (Search Engine Land) — Google is turning AI search ads into answer-like units. The competitive frame shifts from ranking ads near answers to embedding paid persuasion inside the answer experience. (There’s a great discussion about this on LinkedIn here.)

AI Commerce & Agentic Signals

Google’s latest commerce moves deepen the battle over agentic shopping (Digiday) — Google’s Universal Cart extends agentic commerce across Search, Gemini, YouTube and Gmail. The shopper journey is becoming a platform-controlled orchestration layer.

Google upgrades AI search ads: What marketers need to know (Marketing Dive) — Direct Offers, native checkout and UCP move AI Mode closer to transaction capture. Google is linking conversational discovery directly to merchant conversion infrastructure.

Data & Performance Signals

ChatGPT ads API and context hints (Adthena) — Adthena CMO Ashley Fletcher went under the hood of ChatGPT’s Ads API and found “standard PPC plumbing” metrics, but no conversion data and no competitive tracking. Plus “context hints” instead of keywords.

Users behave differently in AI Overviews vs. AI Mode (Search Engine Land) — A study of 846,000 Google sessions found AI Overviews create comparison behavior, while AI Mode drives acceptance. Visibility strategy now depends on the AI surface, not just the query.

Subscribe now

WHAT’S DEBBIE UP TO?

On Stage

• AdExchanger’s Programmatic AI, May 19, 2026, Las Vegas. I delivered the opening keynote, “Consumers Aren’t Where Ads Expect Them Anymore.”

• Path to Purchase Institute’s Retail Media Summit, June 3, 2026, Chicago. I’m delivering “Who Owns the Customer in the Age of AI-Fueled Commerce?”. Check out the agenda and register here.

• Beet.tv’s The New Age of AI, June 23, 2026, Hotel Majestic, Cannes. I’m hosting and moderating a series of panels focusing on the impact of LLMs and agentic commerce on marketing and paid advertising. Check out the full Beet.tv Cannes agenda here.

• FQ Beach at Cannes Lions, June 25, 2026. I’m moderating “The Human Advantage: Creativity in the Age of AI” for The Female Quotient. Register your interest here.

• RetailClub AI Festival, Sept. 24, 2026, Huntington Beach CA. I’m facilitating a conversation focusing on “When Chat Becomes the Channel: Advertising in the Age of AI Assistants.” Check out the full agenda here.

Watch my speaking reel

Podcasts

• “AI is the New Ad Space” with AdExchanger: In a conversation with Allison Schiff, we explore why AI media can be powerful in the middle of the funnel, what organizational changes brands should make to be AI-ready and what needs to happen to make OpenAI’s $100 billion ad forecast feasible.

Research

• IAB + Sonata Insights: The AI Ad Gap Widens (read our 2024 study here)

• Mediaocean: 2026 Advertising Outlook Report (I wrote the foreword)

Debra Aho Williamson is the Founder and Chief Analyst of Sonata Insights, where she helps advertisers navigate the next major structural shift in media: the rapid rise and monetization of AI platforms. Over nearly two decades as an analyst at EMARKETER, Debra was among the first to forecast the growth of social media advertising and identify how it would fundamentally reshape marketing investment.

Today, she applies that same pattern-recognition lens to AI, analyzing how consumer adoption is accelerating monetization and redefining brand budgets, media strategy and industry power dynamics. Through Sonata Insights, she partners with leading platforms, agencies and technology companies to deliver research, advisory and strategic narratives that bring clarity and credibility to this transition.

Today, I’m welcoming an amazing guest author to The AI Ad Economy. When Retail Media Breakfast Club writer Kiri Masters read Andrew Lipsman’s myth-busting guest post about the “collective hallucination” of agentic commerce a few weeks ago, she immediately pinged me:

It was an equally immediate yes from me. Ever since I discovered Kiri’s writing last year, I’ve been obsessed with her assertive and provocative viewpoints about the future of commerce in the age of AI. I love her stuff as much as I love Andrew’s, and I’m thrilled to share it with you.

Grab some popcorn and enjoy. (And then, subscribe to Kiri’s newsletter.)

P.S. Because I wanted you to be able to read Kiri’s full article, I’m skipping the Numbers and Headlines sections this week. They’ll be back next week.

THE MAIN STORY

The Hallucination Is That Everything Is Just Fine As It Is (guest post by Kiri Masters)

Andrew Lipsman and I agree on most things in retail and commerce. We’ve worked the same beat for years. We’re friends in real life — even when we’re publicly arguing.

But on agentic commerce, we keep landing in different places. We’ve debated it in print. We’ve debated it in person too, most recently at the Digital Shelf Summit in Atlanta, where we shared a stage with Rob Gonzalez. So when Debbie asked if I wanted to write a response to Andrew’s guest post (read it at the link below), I jumped at it. Healthy debate is how an industry stress-tests its assumptions, and this one needs the stress test.

Here’s where Andrew is right, where I think he’s wrong, and why this disagreement matters more than the headlines he picked apart.

What Andrew got right

Three quick concessions, because they’re earned.

The Adobe number was thin. A 693% increase in Gen AI referral traffic sounds enormous until you learn it’s growth off a 0.1% base. Sensor Tower’s read — 0.1% to 0.6% — is the more honest framing. Andrew is right to flag it.

The ChatGPT-as-#2-ecommerce-site chart was bad math. Comparing OpenAI’s self-reported prompt volume to retailer site traffic, while conveniently leaving Google out of the comparison, was sloppy. He’s right to call it cherry-picking.

Sensational headlines deserve scrutiny. This is Andrew’s core methodological point and I agree with it. The retail media industry has a habit of grabbing the most extreme datapoint and running with it. We should all read more carefully.

So far, so good. Where I part ways with Andrew is what he does next.

The category error

Andrew uses these debunked headlines to argue that retailers can stop worrying about agentic commerce as a structural threat. He reaches for an eMarketer forecast showing commerce media ad spend growing by $48 billion through 2029, versus $28 billion in “AI platform on-site retail ecommerce sales,” and concludes that retailers should keep building RMNs and stop fretting.

That’s a forecast about transaction share in 2029. The argument I’ve been making — and the one I made on that stage in Atlanta — is about where consumers do their narrowing in 2026.

A forecast can tell you how much money will move through agentic ecommerce in three years. It can’t tell you where the buying decision is being made today. And that’s the part which retail media’s business model actually rides on.

What Criteo’s data actually shows

Criteo recently published their 2026 Commerce & AI Trend Report. Their top-line view — that AI is additive, that buying stays centralized at trusted retailers, that the hype overstates what’s happening — reads almost like an endorsement of Andrew’s piece.

But their own network data tells a sharper story. Across roughly 500 US Criteo merchants in January and February 2026, more than 70% of LLM-referred users now land directly on product pages. In mid-2025, that figure was 50%. So in roughly six months, the share of AI-referred shoppers who skip everything else on the retailer’s site and go straight to the PDP went from half to more than seven in ten.

Criteo frames this as good news. Higher conversion, less friction, shoppers arriving with intent. From a pure ecommerce P&L view, that’s fair. But for retail media specifically, that pattern is actually a problem.

Adobe’s Q1 2026 numbers, which came out after Andrew’s original article published, make the same point from a different angle. AI-referred visitors spend 48% more time on retail sites, view 13% more pages, and convert 42% higher than non-AI traffic. At first glance, that sounds like onsite retail media is fine.

But Adobe doesn’t tell you which pages.

An AI-referred shopper arrives with a shortlist. They open three PDPs to compare reviews, specs, and price. They pick one. They check out. That’s 12 pages viewed, 15 minutes on site.

What they don’t do is browse the category page, scroll the homepage, or click into the deals section. Those are the surfaces where onsite retail media’s premium inventory lives.

That’s a different problem than “agentic commerce will replace ecommerce.” And it’s the one Andrew’s transaction-share math doesn’t capture.

“But upstream influence isn’t new”

The objection I hear most when I make this argument, and one that Andrew brought up on-stage, says: consumers have always done research before they hit a retailer site. TV, radio, social, influencers, reviews on third-party blogs. Upstream influence is as old as retail itself. But this time is different for a few reasons:

The Criteo data is what this looks like in the wild. The direct-to-PDP behavior pattern doesn’t show up for TV-influenced shoppers, or social-influenced shoppers. It’s specific to AI referrals, and it’s growing.

The OpenClaw signal

One more thing Andrew dismisses too quickly: the early behavior at the edges.

Andrea Leigh, an ex-Amazon advisor with three teenagers, automated her Fred Meyer shopping run with ChatGPT and a browser extension. These are hacks that early users are cobbling together, because off-the-shelf options don’t really exist yet.

We’ve seen this movie. Napster wasn’t a viable business model either. What it was, and what early agentic shopping is now, is proof that consumers want a thing the industry hasn’t figured out how to deliver. Emerging behavior precedes the business model.

People adopt new behaviors when there’s pain in the current way of doing things, and new technology emerges that makes a better experience which is easy to adopt. We haven’t yet seen the technology that meets these requirements. OpenAI’s Instant Checkout wasn’t it, but that doesn’t mean that the code is forever destined to be unlocked.

Amazon, for what it’s worth, seems to think this is real, too. In late April, Amazon joined the Universal Commerce Protocol Tech Council alongside Meta, Microsoft, Salesforce, and Stripe — joining founding members Google, Shopify, Etsy, Target, and Wayfair. UCP is the body writing the open rules for agentic commerce. The biggest walled garden in ecommerce and retail media just took a seat at the table writing the open rules.

So what?

Andrew closed his piece with a warning: “Don’t be surprised when you get burned.”

I’d hand that back to him.

The hallucination he’s worried about — the one where the industry talks itself into agentic commerce as imminent disruption — is real.

But the bigger hallucination is the assumption that nothing about the current system needs to change. That the surfaces retail media is built on will keep generating the same impressions, at the same volumes and prices, for a journey that looks like it did in 2023.

Andrew is right that the headlines are oversold. But he’s wrong that retailers should blindly continue with the same playbook they’ve been running for the past 5 years.

The hallucination isn’t that agentic commerce is coming. It’s the belief that nothing about the current system needs to change.

-----

Kiri Masters is a leading voice in the retail media industry, with a particular interest in how the space is evolving with emerging AI-enabled shopping behavior. As host of the Retail Media Breakfast Club podcast and newsletter, she helps retail & brand leaders navigate industry disruption with pragmatic optimism. Her insights are informed by a decade in the industry, including founding an award-winning agency (acquired in 2022). She writes a weekly column for The Drum, and has been named a Top Retail Expert by RETHINK Retail for four consecutive years.

Leave a comment

THE NUMBERS BEHIND THE AI AD ECONOMY

Back next week!

HEADLINES

Back next week!

Subscribe now

WHAT’S DEBBIE UP TO?

Quoted

OpenAI makes it easier to run shopping ads in ChatGPT (Krystal Scanlon, Digiday)

OpenAI adds product feed ads to ChatGPT (Anu Adegbola, Search Engine Land)

On Stage

• AdExchanger’s Programmatic AI, May 19, 2026, Las Vegas. I just delivered the opening keynote, “Consumers Aren’t Where Ads Expect Them Anymore.”

• Path to Purchase Institute’s Retail Media Summit, June 3, 2026, Chicago. I’m delivering “Who Owns the Customer in the Age of AI-Fueled Commerce?”. Check out the agenda and register here.

• FQ Beach at Cannes Lions, June 25, 2026. I’m moderating “The Human Advantage: Creativity in the Age of AI” for the Female Quotient. Register your interest here.

Watch my speaking reel

Podcasts

• “AI is the New Ad Space” with AdExchanger: In a conversation with Allison Schiff, we explore why AI media can be powerful in the middle of the funnel, what organizational changes brands should make to be AI-ready and what needs to happen to make OpenAI’s $100 billion ad forecast feasible. Plus, learn my foodie recommendation for the restaurant you need to go to after visiting Pike Place Market in my city, Seattle.

Research

• IAB + Sonata Insights: The AI Ad Gap Widens (read our 2024 study here)

• Mediaocean: 2026 Advertising Outlook Report (I wrote the foreword)

Debra Aho Williamson is the Founder and Chief Analyst of Sonata Insights, where she helps advertisers navigate the next major structural shift in media: the rapid rise and monetization of AI platforms. Over nearly two decades as an analyst at EMARKETER, Debra was among the first to forecast the growth of social media advertising and identify how it would fundamentally reshape marketing investment.

Today, she applies that same pattern-recognition lens to AI, analyzing how consumer adoption is accelerating monetization and redefining brand budgets, media strategy and industry power dynamics. Through Sonata Insights, she partners with leading platforms, agencies and technology companies to deliver research, advisory and strategic narratives that bring clarity and credibility to this transition.

Hello from New York and a big thank you and welcome to my 50 new subscribers who joined the AI Ad Economy community after I spoke at Skai’s terrific ShopAble conference. Today I’m keynoting at Operative’s Inview 2026: The Media Intelligence Summit. Next week, it’s on to Vegas (baby!), where I’m delivering the opening keynote at AdExchanger’s Programmatic AI.

Today, I’m exploring something that’s come up a lot recently in my social feeds and in surveys and the media: How can advertisers and AI platforms navigate the tricky balance between trust in AI answers and the desire to monetize those answers? It’s a thicky topic (did I just coin that word? thick+sticky). I’ll continue to keep close tabs on how this debate evolves — and report what I find.

THE MAIN STORY

The Commercialization Challenge Facing AI Media

Over the past couple of weeks, the conversation about ads in AI media has shifted. We’ve gone from “How will ads appear?” to “What happens when consumers push back?”

The industry is confronting the reality that what makes AI so useful — the expected objectivity of the answers — is also what will be the most impacted by the push to monetize.

Even though I call it “AI media,” the consumer experience of using AI platforms is actually more like getting advice than engaging with media. And once you enter the business of offering advice — especially alongside advertising — you enter delicate territory.

The situation becomes even more complicated when you look at recent data showing that consumer trust in AI platforms declines as ads enter the experience.

A survey by Quad Graphics and Harris Poll, conducted in February 2026, found that 45% of US adult internet users would trust an AI agent to make better product recommendations than an in-store associate would. But advertising diminished brand trust; 75% said they would trust brands less if they knew those brands had paid to influence the AI’s recommendations.

The tension between usefulness and concern also appears in other recent research:

• “Trust in AI drops significantly for high-stakes decisions and sensitive information, where customers prefer human involvement and limits on AI autonomy.” (Adobe, 2026 AI and Digital Trends report)

• “Consumers are willing to trust AI – and increasingly, they do. But that trust is provisional. They verify. They cross-check. They bounce. And when they find inconsistencies? They move on.” (Yext, 2026 Consumer Search Behaviors Report)

The takeaway:

Trust in AI is contextual.

It’s also fragile.

Monetization is starting to change the psychological contract between consumers and AI platforms

63% of US adults believe ads in AI search results would make them less trusting of the results, a January 2026 Ipsos poll found.

While that figure may be artificially high because the survey base consisted of both AI users and non-users, concern about the presence of ads could end up keeping those holdouts on the sidelines forever. And AI platforms still need new users in order to expand monetization.

So to me, the developing contract between consumers and AI platforms boils down to this:

Consumers have realized that AI platforms can deliver information faster and more easily than traditional methods. The more they use AI, the more attached they become to that convenience. And that attachment grows stronger if they feel confident that what they see isn’t tainted by advertiser influence.

AI platforms are realizing they are sitting on one of the richest sources of consumer intent data ever created. The allure of monetizing that intent is almost impossible to ignore.

In this situation, the agreement with consumers – that the information they see won’t be marred by paid interlopers – becomes ever more difficult to maintain.

The ad industry already senses this tension

Much of this debate currently focuses on ChatGPT, which recently launched a self-serve ad manager that’s bound to increase the types and variety of ads users will see.

To OpenAI’s credit, it recognizes the delicate situation it’s in. Every communication about its ad business states that ads are clearly labeled and separate from answers.

Smartly, an advertising creative partner with OpenAI, similarly emphasizes both relevance and conversational fit in a recent blog post: “ChatGPT presents a huge opportunity for advertisers in that it’s a one-to-one interaction where ads are contextual and matched to the topic of the conversation.”

Industry observers are also calling for trust and transparency. In his excellent Substack article, “Before AI Advertising Scales, It Needs Rules Consumers Can Trust,” Pete Blackshaw of BrandRank writes, “The issue is whether consumers can clearly distinguish between independent answers and commercially influenced ones. … Once users begin questioning whether answers are commercially distorted, trust erosion accelerates quickly.”

It’s not the same as social media — or is it?

In the social media era, the rhetoric centered on maximizing attention and engagement. In AI, the early message is credibility and balance.

The shift in the language from the social media era to the AI media era is important.

Because AI increasingly feels less like a media channel and more like a conversation with an advisor, there’s even more pressure on the information to be trusted, reliable and not influenced by outside advertiser pressure.

On the consumer side, the conversational interface changes expectations for ad load, perceived conflicts of interest and the credibility of recommendations.

This makes the concern over whether ads in social feeds were adjacent to negative or unpleasant content seem quaint, in retrospect. Preserving integrity in AI recommendations is much more consequential than whether a travel ad appeared next to a post about a bad resort experience.

That leads me to this conclusion:

The platforms that scale advertising successfully may be the ones that commercialize most carefully.

I agree with Pete Blackshaw’s assertion that “The industry needs measurable standards around answer integrity. Independent researchers and auditors should be able to test whether advertiser participation influences recommendation frequency, sentiment or answer quality.”

Establishing these guardrails early could help define the relationship between consumers, advertisers and AI platforms before trust erodes further.

However, there is one important wild card that I feel compelled to share, even if I believe careful monetization of AI media is the safest route.

My experience analyzing consumer attitudes toward social media advertising (which I did for 17 years before shifting to focus on AI) convinced me that trust doesn’t correlate to usage of social platforms — or to the willingness to engage with ads there.

For several years while I was at eMarketer, we conducted primary research into “digital trust” — essentially, whether consumers trust social platforms to protect their privacy, shield them from misinformation and disinformation and be good stewards in the world. And every year, some of the industry’s largest platforms ranked among the least trusted across nearly every metric we measured.

But when we looked at usage data and ad engagement, we didn’t see declines. Trust didn’t impact the metrics that advertisers care about the most: Are enough people seeing my ad and are they engaging with it?

I believe there’s going to be a similar situation with AI platforms, where trust will be tested (and the media and pundits will crow about that when it happens), but in the right circumstances, an ad could potentially be additive to a conversation, and become useful rather than something to distrust.

When I was shopping for a frying pan for a holiday gift for my parents, I knew what I wanted to purchase: a Made In 12-inch professional skillet. I went to ChatGPT with one question: What’s the cheapest price right now on this skillet?

It showed me images of the pan I was looking for, as well as a few other Made In varieties, and I ended up buying the pan I wanted on Amazon. But if I had seen an ad at that very moment offering a discount or free shipping from another retailer, it could have easily influenced where I ended up buying.

Why? ChatGPT compressed my decision path, cutting out the comparison paralysis from visiting various websites and checking prices. I trusted the information. And the offer of a discount would have been qualitatively different from seeing a search ad.

That experience tells me that the right ad shown at the right time may not lower trust.

Balancing organic advice with sponsored responses will be difficult for AI platforms. AI media may reward restraint more than previous digital media eras did, and integrity may outweigh the allure of boosting ad load.

But it’s possible that there is a path forward where ads and organic content can actually work together and build trust, not destroy it.

Leave a comment

THE NUMBERS BEHIND THE AI AD ECONOMY

A Different Lens on Trust

Continuing on the topic of trust, new research from Prophet indicates that even as more consumers are using AI (up from 45% of consumers in China, Germany, Singapore, the UK and the US in 2024 to 73% in 2026), their enthusiasm is diminished by concerns about its impact on human life. As I wrote in the main story, trust is contextual.

HEADLINES

Signals from The AI Ad Economy

AI Platform Monetization

OpenAI solidifies ad platform ambitions with ChatGPT Ads Manager — ChatGPT ads added self-serve buying, CPC bidding and expanded measurement. OpenAI is moving from curated pilot to performance-platform infrastructure.

OpenAI rolls out ads on ChatGPT, but marketers remain unconvinced — Advertiser hesitation is shifting from access to proof. ChatGPT’s ad opportunity now depends on whether conversational intent can produce measurable performance.

Testing ads in ChatGPT — New: OpenAI is expanding ChatGPT ads to the UK, Mexico, Brazil, Japan and South Korea. The pilot is becoming a cross-market test of AI-native ad acceptance.

AI Commerce & Agentic Commerce

Amazon ditches Rufus AI chatbot in favor of Alexa shopping agent — Amazon is embedding conversational commerce directly into its core shopping interface instead of maintaining a standalone AI assistant. The shift signals that AI shopping experiences are moving from experimental chatbot features toward persistent, agent-like infrastructure

Raleigh startup ReFiBuy raises $13.6 million in seed round — ReFiBuy, led by ecommerce entrepreneur Scot Wingo, raised $13.6 million to help merchants surface inside AI shopping flows. AI visibility is becoming an important ecommerce infrastructure layer.

Data & Performance

8 GEO metrics to track in 2026 — AI search visibility is being measured through citation frequency, answer inclusion and share of model voice. The 8% click-rate datapoint reframes organic performance around presence inside answers, not traffic alone.

AdClarity tracks ChatGPT ads — AdClarity says it now tracks 1,000+ ChatGPT advertisers and 10 billion+ ad impressions. Third-party measurement is forming around ChatGPT as paid media inventory.

Adthena launches ChatGPT ads intelligence platform — Advertisers are building competitive-intelligence infrastructure specifically for ChatGPT ads. AI search is beginning to develop the monitoring, benchmarking and share-of-voice tooling that defined mature search advertising.

OpenAI Ads update, a wild stat and something we just shipped at AirOps — Sponsored-card density fell from roughly 3% of ChatGPT queries to 47 ads per 100,000 searches after the self-serve launch, according to AirOps data cited by CEO Alex Halliday on LinkedIn. OpenAI appears to be shifting from aggressive inventory testing toward controlled monetization and scalable advertiser onboarding.

Subscribe now

WHAT’S DEBBIE UP TO?

Quoted

OpenAI makes it easier to run shopping ads in ChatGPT (Krystal Scanlon, Digiday)

OpenAI adds product feed ads to ChatGPT (Anu Adegbola, Search Engine Land)

On Stage

• AdExchanger’s Programmatic AI, May 19, 2026, Las Vegas. I’m delivering the opening keynote, “Consumers Aren’t Where Ads Expect Them Anymore.” Use my code JOINME20 for 20% off your pass.

• Path to Purchase Institute’s Retail Media Summit, June 3, 2026, Chicago. I’m delivering “Who Owns the Customer in the Age of AI-Fueled Commerce?”. Check out the agenda and register here.

• FQ Beach at Cannes Lions, June 25, 2026. I’m moderating “The Human Advantage: Creativity in the Age of AI.”

Podcasts

• “AI is the New Ad Space” with AdExchanger: In a conversation with Allison Schiff, we explore why AI media can be powerful in the middle of the funnel, what organizational changes brands should make to be AI-ready and what needs to happen to make OpenAI’s $100 billion ad forecast feasible. Plus, my foodie recommendation for the restaurant you need to go to after visiting the Pike Place Market in my city, Seattle.

• The AI Edge with U of Digital. It’s been a while since I did a purely news-driven podcast and this one was rapid-fire fun. U of Digital’s Shiv Gupta and Myles Younger and I debated whether Sam Altman is a liar, why Anthropic’s Mythos announcement might have been a marketing tactic, whether AEO/GEO has jumped the shark, and the two big developments in consumer behavior and advertising that will tell me if we’ve gotten to a social media-like turning point in AI. Catch it on YouTube, Apple podcasts or Spotify.

Research

• IAB + Sonata Insights: The AI Ad Gap Widens (read our 2024 study here)

• Mediaocean: 2026 Advertising Outlook Report (I wrote the foreword)

Watch my speaking reel

Debra Aho Williamson is the Founder and Chief Analyst of Sonata Insights, where she helps advertisers navigate the next major structural shift in media: the rapid rise and monetization of AI platforms. Over nearly two decades as an analyst at EMARKETER, Debra was among the first to forecast the growth of social media advertising and identify how it would fundamentally reshape marketing investment.

Today, she applies that same pattern-recognition lens to AI, analyzing how consumer adoption is accelerating monetization and redefining brand budgets, media strategy and industry power dynamics. Through Sonata Insights, she partners with leading platforms, agencies and technology companies to deliver research, advisory and strategic narratives that bring clarity and credibility to this transition.

The AI monetization push keeps getting stronger at OpenAI. This week, I’m exploring what its latest announcements mean for advertisers.

First, two announcements:

1. Possible Miami may have just ended, but Cannes Lions is right around the corner. Want to join my growing list of partners for presentations, panels and yacht events? I have a few more openings in my schedule. Reach out if you’d like to work with me!

2. Next week I’ll be in New York for two exciting events. First, Skai’s ShopAble conference on May 13th, where I’m presenting “AI, Consumers and Commerce: A Reality Check.” Then on May 14th, I’m keynoting Operative’s Inview 2026: The Media Intelligence Summit. Will you be at these events? Let me know so we can say hi.

Ok, let’s dive into the main story.

THE MAIN STORY

OpenAI is Opening the Window on Its Ad Ambitions

OpenAI made several major moves in its ad business this week. In an interview with The AI Ad Economy, the company’s ad leaders revealed important details about new ad formats, why ChatGPT ads behave differently than search ads, and why OpenAI is keeping tight control over decisioning.

Here’s what OpenAI announced:

1. Ads manager: Now beta testing in the US, the ads manager will allow advertisers to create accounts, set budgets/bids, upload ads, launch and manage campaigns and view ad performance details.

2. CPC bidding, conversions API and pixel-based measurement: These features bring OpenAI closer to the standard buying and measurement capabilities that other digital ad companies already offer.

3. Adtech and agency partners: Previously announced adtech partner Criteo has been joined by Kargo, Pacvue and StackAdapt on the ad buying side and Adobe on the ad creative side. Holding companies Dentsu, WPP, Publicis and Omnicom join as agency partners.

Subscribe now

What it means for advertisers, OpenAI and digital media

For advertisers: It’s now much easier to start advertising on ChatGPT, and to compare results across platforms. Pacvue clients, for example, can “manage ChatGPT campaigns alongside Amazon, Walmart and 100+ channels in one place,” president Melissa Burdick wrote on LinkedIn. And the introduction of CPC pricing means advertisers will be able to use a buying method they’ve long been familiar with, lowering the psychological barrier for performance marketers to start testing ChatGPT ads.

For OpenAI: These steps show how urgent it is for OpenAI to demonstrate that advertising can become a significant revenue stream. Remember, this is a company that has set a $100 billion revenue goal by 2030. By adding things like an ads manager, CPC bidding, pixel measurement and CAPI, OpenAI is demonstrating that it understands the basic building blocks that are necessary for advertisers to feel comfortable testing on ChatGPT. The list of partners is also a strong signal. It doesn’t guarantee OpenAI will automatically get a big chunk of advertiser budgets, but it gives the company a wide-open window of opportunity.

For digital media: OpenAI is no longer behaving like a company cautiously experimenting with ads. It’s building the infrastructure for an entire advertising system extremely quickly. That’s something that every digital media player needs to watch closely.

My Conversation with OpenAI’s Ad Leaders

I spoke to Asad Awan, head of monetization, and Benji Shomair, VP of monetization, this week. Both used to work at Meta, and the conversation offers a revealing look at how OpenAI views the ad market and where it sees meaningful differences between conversational ads and search ads. (Interview is lightly edited for clarity.)

Debra Aho Williamson: In the first part of your test, the ads were largely search ads that advertisers were importing into ChatGPT. Is there anything new coming up about the types of ads that people can run?

Asad Awan: Right now we still have the same format, which is the business name and a few lines of title and subtitle with a small image. But of course, we are iterating on different formats. We don’t have a date around when we will launch some of those, but it’s certainly something which is top of mind.

Benji Shomair: One comment on the idea that this was all just search: I think what we found as we’ve been progressing through pilot is that the folks who are just taking their search ads and pouring them directly onto our platform are not the ones who are performing the best. What we’re finding is that this isn’t the same as search for a host of different reasons. How people are using the platform is different and therefore the ads are different. And then also the way the ranking engine works, because of the contextual targeting. On our platform, we’re finding that one of the things that really helps an advertiser perform well is using a lot of creative variation.

Share

Debra: Tell me more about how you’re working with the adtech partners in the announcement.

Benji: Criteo is the only one that has been out there officially, and they were the first of our technology partners. We work with Kargo, StackAdapt and Pacvue in the same way as Criteo: Access to the same API, providing budgets, bids, creative and bringing advertising onto our platform. Adobe is slightly different; Adobe is a creative partner. A lot of advertisers use Adobe to manage their creative assets across different channels, and now ChatGPT is an additional channel where they can do that in the same place.

The role the partners play is bringing budgets, bids and creative. There’s no decisioning taking place. All the ad delivery decisions take place within the ChatGPT system. A partner like StackAdapt will work differently with other platforms, and they’ll do real-time bidding, but we don’t have real-time bidding on our platform. They’re bringing the budgets and we control the decisioning.

Debra: Why is it important for OpenAI to retain control of decisioning? Is it because you’re still building this out, or is control of decisioning something you want to maintain going forward?

Asad: The main question in decisioning is that when an ad comes in, you have a conversation from the user and you need to make sure you match these ads in a relevant way, because we don’t share that conversation [with the advertiser]. We believe that this matching will evolve as the platform evolves and as the formats become more native. All the way from matching the ad, the ranking, the relevance­—all of that is an insight. And a big part of that is driven by [the fact that] how people use this platform is different, and we want to make these [ads] very native, both from an experience perspective and from a ranking perspective, and how they fit in the user journey through these conversations.

Debra: Digiday reported that you are getting ready to expand ads into Europe. Can you comment on that?

Benji: We plan on expanding more broadly to other markets. We’re not commenting on specific countries just yet. What Digiday was pointing towards is actually a pretty standard thing that was happening in the app, and it may or may not point toward going to Europe in the short term. [UPDATE ON 5/7: OpenAI announced that ads will soon expand to United Kingdom, Mexico, Brazil, Japan, and South Korea.]

Debra: It’s been reported that retail advertisers are heavy users of ChatGPT ads, and travel as well. Is there any change in the categories of advertisers that you’re seeing hopping onto ChatGPT?

Asad: No, I think those verticals—travel, retail, ecomm, SaaS—are doing quite well and the demand has been pretty strong.

Debra: One of my main theses about AI media is that this is moving much faster than people are expecting. What are you seeing in terms of advertisers and their ability to keep up? Do they need a lot of handholding or are they ready to jump in?

Benji: There’s the excitement that you only see a platform of this scale arrive every decade or so, and curiosity to figure out what works on that new platform. Folks are leaning in pretty aggressively in terms of iterating and working with us as we identify things that are working and then shifting their strategy.

Takeaways

OpenAI is moving beyond experimentation. It’s rapidly assembling the components of a full advertising system and preparing for scale. Advertisers don’t need to shift major budgets overnight, but they do need to start building institutional knowledge about how advertising in conversational settings performs differently than other forms of advertising.

Leave a comment

Here’s what others are saying about OpenAI’s announcements:

• OpenAI Applications CTO Vijaye Raji calls out the engineering team (and lots of AI tokens) for the speed of the ad features launch. Plus: monetization jobs at OpenAI.

• Digiday’s Krystal Scanlon digs deeper into OpenAI’s rationale for controlling decisioning.

• Retail media guru Roger Dunn explains why the list of OpenAI’s new partners is actually “the bigger story.”

THE NUMBERS BEHIND THE AI AD ECONOMY

Regular AI Usage is On the Rise

The percentage of US adults using generative AI chatbots is still increasing, though more slowly than last year, according to Ipsos. Its most recent Consumer Tracker study, conducted in April 2026, found that 59% of US adults who were familiar with the concept of artificial intelligence use a chatbot “often” or “sometimes.” While that’s up 15 points from March 2025, it’s only a 3-point difference from November 2025.

However, the percentage using chatbots “often” is widening, standing at 24% in April. double where it was last March.

HEADLINES

Signals from The AI Ad Economy

AI Advertising Signals

OpenAI is Now Sharing Its Users’ Data With Advertisers (Adweek) — OpenAI’s privacy update formalizes purchase-data sharing and marketing-partner relationships. The structural shift is clear: measurement and targeting are becoming part of ChatGPT’s ad platform foundation.

Google updates links within AI Overviews, AI Mode (Search Engine Land) — Google is making publisher, Reddit and expert-source links more prominent inside AI answers. AI search is evolving from a closed-answer interface into a blended discovery layer that may better preserve outbound exploration and content visibility.

AI Commerce & Agentic Signals

Stripe updates Link, a digital wallet that autonomous AI agents can use, too (TechCrunch) — Stripe is giving agents controlled payment access through Link instead of raw credentials. Agentic commerce is shifting from checkout experimentation to delegated-payment infrastructure.

Meta readies AI shopping agents for Instagram to rival TikTok Shop (EMARKETER) — Meta is developing Instagram shopping agents that could identify products in feeds and complete purchases. Social commerce is moving toward AI-mediated product recognition and transaction routing.

Data Signals

Reddit expects revenue above estimates as AI tools fuel ad growth (Reuters) — Reddit’s active advertiser count rose 75%, Q1 revenue grew 69% and performance ads now exceed 60% of ad revenue. AI-assisted ad tools are strengthening Reddit’s performance-marketplace position.

Shopify expects agentic commerce to lift ecommerce adoption (EMARKETER) — Shopify says AI-driven traffic rose 8x and AI-search orders grew 13x in Q1. AI discovery is beginning to produce measurable merchant demand, not just exploratory traffic.

Subscribe now

WHAT’S DEBBIE UP TO?

Quoted

Yahoo says its AI assistant has a secret weapon in the new search wars (Lara O’Reilly, Business Insider)

OpenAI opens up ChatGPT ads manager to the U.S. while promising third-party measurement, CPA bidding (Krystal Scanlon, Digiday)

Podcasts

• “AI is the New Ad Space” with AdExchanger: In a conversation with Allison Schiff, we explore why AI media can be powerful in the middle of the funnel, what organizational changes brands should make to be AI-ready and what needs to happen to make OpenAI’s $100 billion ad forecast feasible. Plus, my foodie recommendation for the restaurant you need to go to after visiting the Pike Place Market in my city, Seattle.

• The AI Edge with U of Digital. It’s been a while since I did a purely news-driven podcast and this one was rapid-fire fun. U of Digital’s Shiv Gupta and Myles Younger and I debated whether Sam Altman is a liar, why Anthropic’s Mythos announcement might have been a marketing tactic, whether AEO/GEO has jumped the shark, and the two big developments in consumer behavior and advertising that will tell me if we’ve gotten to a social media-like turning point in AI. Catch it on YouTube, Apple podcasts or Spotify.

• Retailgentic: I joined ReFiBuy’s Scot Wingo for a wide-ranging discussion about how AI is reshaping advertising and commerce and what brands should be doing now. Watch the full podcast here!

Research

• IAB + Sonata Insights: The AI Ad Gap Widens (read our 2024 study here)

• Mediaocean: 2026 Advertising Outlook Report (I wrote the foreword)

On Stage

• Skai’s ShopAble, May 13, 2026, New York City. I’ll be presenting “AI, Consumers & Commerce: A Reality Check.” RSVP at my personal link here.

• Operative’s Inview 2026: The Media Intelligence Summit, May 14, 2026, New York City. I’m giving the opening keynote, “The AI Monetization Shift: Why Media Won’t Get 8 Years This Time.” Request an invitation to attend here.

• AdExchanger’s Programmatic AI, May 19, 2026, Las Vegas. I’m delivering the opening keynote “Consumers Aren’t Where Ads Expect Them Anymore.” Use my code JOINME20 for 20% off your pass.

• Path to Purchase Institute’s Retail Media Summit, June 3, 2026, Chicago. I’m delivering “Who Owns the Customer in the Age of AI-Fueled Commerce?”. Check out the agenda and register here.

Watch my speaking reel

Debra Aho Williamson is the Founder and Chief Analyst of Sonata Insights, where she helps advertisers navigate the next major structural shift in media: the rapid rise and monetization of AI platforms. Over nearly two decades as an analyst at EMARKETER, Debra was among the first to forecast the growth of social media advertising and identify how it would fundamentally reshape marketing investment.

Today, she applies that same pattern-recognition lens to AI, analyzing how consumer adoption is accelerating monetization and redefining brand budgets, media strategy and industry power dynamics. Through Sonata Insights, she partners with leading platforms, agencies and technology companies to deliver research, advisory and strategic narratives that bring clarity and credibility to this transition.

Wow! You’re one of nearly 1,500 people who have subscribed to The AI Ad Economy since I launched on Substack in December and LinkedIn in January. I’m thrilled you’re here! My goal is to reach 2,000 subscribers by the end of June. You can help me get there by sharing this newsletter with your colleagues, clients, partners and friends.

This week is a big one for me because I’m celebrating the 2nd anniversary of my advisory and research business, Sonata Insights. In today’s issue, I’m looking back at what has changed since April 2024 — and what might be coming next in AI media. Scroll down for my top predictions for the next two years.

Subscribe now

THE MAIN STORY

What 2 Years of Analyzing AI Behavior Taught Me About the Future of Advertising

If you worked in AI this time two years ago, here’s what might have been top of mind for you:

• Perplexity’s plan to launch ads

• Google’s “search generative experience”

• AI hallucinations

• SearchGPT

And here’s what no one was talking about (yet):

• AEO

• GEO

• Agentic commerce

• ChatGPT ads

In 24 months, advertising has become a reality on ChatGPT, Google AI Overviews and other AI-infused experiences, transforming from a vague idea to a real (albeit still small) source of revenue.

In search, AI has gone from being adjacent to search results to fully embedded. It is rewriting how discovery happens, surfacing intent in new ways and creating a cottage industry of AI visibility specialists.

And on the consumer side, generative AI has crossed from curiosity to mass usage. In Q1 2024, generative AI usage in the US was still emerging, with 24% of US internet users using a generative AI tool in the past 12 months, according to a S&P Global Markets Intelligence Kagan survey. By Q3 2025, that share had nearly doubled to 46%.

For the first time since the rise of search, consumer intent is shifting away from traditional ad surfaces.

These shifts didn’t come as a surprise to me when they happened. The signals were already there.

Here are a few things I predicted:

The rise of AEO: In July 2024, I wrote, “Businesses of all sizes that market to consumers need to make sure they show up in the places consumers are going when they need an answer fast: generative AI platforms.”

The continued dominance of Google: In my first research collaboration with Datos in July 2024, I concluded, “Google … benefits from a powerful trifecta: a traditional search engine, a companion AI platform and the top site for social search, YouTube.”

The challenges Perplexity would face selling ads: In October 2024, a few months after Perplexity launched its ads test, I attended an Advertising Week session where Andrew Beck, then the GTM leader for ads, shared a few early results from the brand-advertising test. My impression? “It’s taking a risky approach.”

That ChatGPT would be much more than an “AI search platform.” In a January 2025 conversation with ChatGPT, I was introduced to a new facial moisturizer, researched it, compared it to other options, and bought it via a direct link to Amazon. My takeaway: “ChatGPT isn’t just disrupting search, it’s a full funnel marketing platform.”

Share The AI Ad Economy

A Shift That’s Happening Faster Than Anyone Expected

When I opened Sonata Insights, it was a big shift for me personally and professionally. I pivoted from 17 years of being a social media analyst at eMarketer to analyzing (as a solopreneur) the growth of AI platforms as new destinations for discovery, intent and purchasing.

I recognized the AI media shift early because I was paying attention to consumer signals. It’s the same methodology I used to identify the massive change in consumer behavior that social networks were causing in 2006, when I founded the social media research practice at eMarketer.

Today, I’m applying that same pattern recognition lens to analyze how changes in consumer AI behavior will impact marketing and paid advertising.

What has become clear is that this shift is happening faster than anyone (including me) expected. In mid-2024, I gave a presentation to the AI Marketers Guild, the community created by David Berkowitz, my former eMarketer colleague (watch the clip below).

In just two years, that opportunity has already become a reality. There’s a lot that’s still unknown about how to market in this new environment, but what’s no longer debatable is the existence of the opportunity.

What I’m Predicting for the Next Two Years

With two years of analyzing the signals under my belt, I’m looking ahead to the future. Here are a few predictions for 2028: what AI media will look like, how marketers will use it and how consumers will experience it. Do you agree? Disagree? What have I missed?

Leave a comment

1. AI media won’t be a destination. It will be a utility within the devices and environments consumers use to create, decide and act.

2. No single company will own the majority of the AI media ad market. Distribution, models and interfaces will fragment too quickly for one dominant player to emerge as in search and social. This will be a market with several semi-large players and a long tail of many more.

3. A meaningful share of high-intent consumer decisions will happen without the user ever seeing a traditional ad unit. That’s the foundation for agentic advertising (ads aimed at agents, not humans), which will become a reality over the next two years.

4. Agencies built around human-facing creative will struggle. New “agentic ad agencies” will emerge to design and optimize ads for machines, and many will operate with minimal human involvement.

It’s been a momentous two years in AI, advertising and consumer behavior, and the next phase of AI media will be about institutionalizing the changes already under way and preparing for what comes next.

What do YOU think is ahead for the ad industry?

Leave a comment

THE NUMBERS BEHIND THE AI AD ECONOMY

Nearly Half of Consumers Google a Brand That AI Recommends

One of my favorite stories to tell is about the time ChatGPT introduced me to the Paula’s Choice skincare brand — simply because I asked it to recommend a moisturizer that was similar to a brand I already used. I went from unaware the brand even existed to purchasing a product ChatGPT recommended in just a few minutes, and one year later, I’m still a loyal Paula’s Choice customer.

That’s why Idea Grove’s study caught my eye. Only 2% of US adults would buy without doing any other research, but nearly half said their next step after seeing an AI recommendation would be to Google the brand.

This takeaway from the study is important: “Getting mentioned by ChatGPT or Gemini is just the starting gun for a trust evaluation that plays out across Google, review sites and the brand’s own website.”

HEADLINES

Signals from The AI Ad Economy

AI Advertising Signals

ChatGPT ads expand to logged-out users (Search Engine Land) — OpenAI is widening ad inventory beyond signed-in users, addressing advertisers’ early struggle to spend against limited frequency and supply. The signal: ChatGPT ads are being scaled like a performance channel.

AdMarketplace Is Piloting Performance Ads In AI Chat (AdExchanger) — AdMarketplace’s AI Discover beta ran 2 million sponsored placements across roughly 700,000 AI queries, priced on CPC. This pushes AI chat ads toward intent-based performance, not impression-based sponsorship.

Snapchat Users Will Soon Be Able to Chat Directly with Agentic Ads Served to Them Via DM (Adweek) — Snapchat is turning inbox ads into conversational agents, letting users query brands directly inside DMs. This shifts social ads from static impressions to interactive discovery, embedding product research and intent capture inside chat.

AI Commerce & Agentic Signals

The Race Is on to Keep AI Agents From Running Wild With Your Credit Cards (Wired) — FIDO Alliance, Google and Mastercard are pushing standards for authenticating agent-led payments. Agentic commerce is moving from “can agents buy?” to “who proves authorization, intent and liability?”

Platforms, brands accelerate agentic commerce push as fintechs plug payment gaps (The Economic Times) — In India, Flipkart, Bigbasket, Ajio and Firstcry are building AI storefronts for LLM-driven shopping, while Razorpay pilots agentic commerce with 15–20 merchants. India is becoming a live testbed for AI storefront infrastructure.

Data & Performance Signals

AI traffic to US retailers rose 393% in Q1, and it’s boosting their revenue too (TechCrunch) — Adobe data shows AI-referred retail traffic converted 42% better than non-AI traffic in March 2026. AI referrals are becoming a higher-value commerce channel.

The biggest wall between enterprise ecommerce and agentic AI is trust (EMARKETER) — There’s a sharp tension: 65.5% of US consumers still have misgivings about agent-led payments, while 52% would let an AI agent purchase without final approval. Readiness and anxiety are rising together.

Subscribe now

WHAT’S DEBBIE UP TO?

Podcasts

• “AI is the New Ad Space” with AdExchanger: In a conversation with Allison Schiff, we explore why AI media can be powerful in the middle of the funnel, what organizational changes brands should make to be AI-ready and what needs to happen to make OpenAI’s $100 billion ad forecast feasible. Plus, my foodie recommendation for the restaurant you need to go to after visiting the Pike Place Market in my city, Seattle.

• The AI Edge with U of Digital. It’s been a while since I did a purely news-driven podcast and this one was rapid-fire fun. U of Digital’s Shiv Gupta and Myles Younger and I debated whether Sam Altman is a liar, why Anthropic’s Mythos announcement might have been a marketing tactic, whether AEO/GEO has jumped the shark, and the two big developments in consumer behavior and advertising that will tell me if we’ve gotten to a social media-like turning point in AI. Catch it on YouTube, Apple podcasts or Spotify.

• Retailgentic: I joined ReFiBuy’s Scot Wingo for a wide-ranging discussion about how AI is reshaping advertising and commerce and what brands should be doing now. Watch the full podcast here!

Research

• IAB + Sonata Insights: The AI Ad Gap Widens (read our 2024 study here)

• Mediaocean: 2026 Advertising Outlook Report (I wrote the foreword)

On Stage

• M Agency webinar, May 6, 2026, 11am PT/2pm ET (virtual). I’m joining senior leaders from M Agency for “AI is Shaping the Customer Journey: Is Your Brand Visible Where It Counts?” Register for the webinar here.

• Skai’s ShopAble, May 13, 2026, New York City. I’ll be presenting “AI, Consumers & Commerce: A Reality Check.” RSVP at my personal link here.

• Operative’s Inview 2026: The Media Intelligence Summit, May 14, 2026, New York City. I’m giving the opening keynote, “The AI Monetization Shift: Why Media Won’t Get 8 Years This Time.” Request an invitation to attend here.

• AdExchanger’s Programmatic AI, May 19, 2026, Las Vegas. I’m delivering the opening keynote “Consumers Aren’t Where Ads Expect Them Anymore.” Use my code JOINME20 for 20% off your pass.

• Path to Purchase Institute’s Retail Media Summit, June 3, 2026, Chicago. I’m delivering “Who Owns the Customer in the Age of AI-Fueled Commerce?”. Check out the agenda and register here.

Watch my speaking reel

Debra Aho Williamson is the Founder and Chief Analyst of Sonata Insights, where she helps advertisers navigate the next major structural shift in media: the rapid rise and monetization of AI platforms. Over nearly two decades as an analyst at EMARKETER, Debra was among the first to forecast the growth of social media advertising and identify how it would fundamentally reshape marketing investment.

Today, she applies that same pattern-recognition lens to AI, analyzing how consumer adoption is accelerating monetization and redefining brand budgets, media strategy and industry power dynamics. Through Sonata Insights, she partners with leading platforms, agencies and technology companies to deliver research, advisory and strategic narratives that bring clarity and credibility to this transition.

If you want more from Andrew in your inbox (which you should!), I encourage you to subscribe to his Substack Media, Ads & Commerce here.

PS: Because I wanted you to be able to read Andrew’s full article, I’m skipping the Numbers and Headlines sections of this week’s newsletter. They’ll be back next week.

Subscribe now

THE MAIN STORY

An Agentic Commerce Reality Check (guest post by Andrew Lipsman)

I’ve previously described agentic commerce as a “collective hallucination”—a not-so-popular characterization among the pro-agentic crowd.

The most famous examples from history are the Dancing Plague of 1518, the Salem Witch Trials, and the Mad Gasser of Mattoon. Collective hallucinations don’t arise out of thin air—rather, they manifest within communities subject to intense external stress, high suggestibility, and social contagion.

In late fall 2025, all three converged in retail and ecommerce. Retailers were under growing pressure to squeeze more revenue from retail media networks, even as performance gains became harder to sustain. At the same time, the surging adoption of AI search made the idea of “agentic commerce” sound not just plausible but inevitable—a narrative too compelling for an anxious industry to resist. The Fall conference season echo chamber—amplified on LinkedIn—converted speculation into conviction almost overnight.

The sentiment soon crystallized as every major consulting firm weighed in with its prognosis of retailers’ existential risk:

Kearney (Aug 2025): Agentic Commerce: From Brand Loyalty to Bot Logic

“Agentic commerce will be the most disruptive retail shift since e-commerce. 60% of consumers expect to use AI agents for shopping within the next 12 months. The real threat comes from cross-platform ‘super agents’ that sit between retailers and consumers. For unprepared retailers, the risk is up to 500 bps EBIT erosion from margin compression, traffic dispersion, and brand commoditization.”

BCG (Oct 2025): Agentic Commerce is Redefining Retail – How to Respond

“The advancements in agentic shopping present substantial challenges for traditional retailers, primarily through risks associated with disintermediation… The growth of zero-click search and agent-driven interactions is eroding direct traffic—along with the retailer’s ability to observe, influence, and understand consumer behavior at scale… AI agents’ emphasis on utility weakens traditional brand loyalty, limits cross-selling opportunities, and puts retail media revenues under pressure, all while intensifying the race to the bottom on price and delivery speed.”

McKinsey (Oct 2025): The Agentic Commerce Opportunity: How AI agents are Ushering in a New Era for Consumers and Merchants

“Agentic commerce also threatens to bring about a decline in traditional revenue streams, particularly from advertising. Retail media networks, which rely on ad-based models, could face challenges as consumers shift toward agent-driven experiences that bypass traditional ad channels.”

Bain (Nov 2025): Agentic AI Poised to Disrupt Retail, Even With 50% of Consumers Cautious on Fully Autonomous Purchases

“Of the three types of AI agents – third-party ‘objective’ agents (e.g. ChatGPT, Perplexity), on-site retailer agents, and off-site retailer agents – third-party agents threaten to disintermediate retailers most.”

The result was a near-industry-wide belief that agentic commerce would rewrite the rules of retail within months. An ongoing array of headline-grabbing charts and statistics have only added fuel to the fire.

But few of these stats stand up to scrutiny.

So, as with any hallucination, what feels visceral and real in the moment warrants a sober second look.

In this article, we’ll put three recent agentic commerce headline findings to the test.

Share

HEADLINE 1: Adobe: Holiday Gen AI Ecommerce Referral Surged 693% vs. Prior Year

Throughout the 2025 holiday season, Adobe reported on various ecommerce sales and shopping trends. Garnering the most headlines this year wasn’t the increase in ecommerce spending but the dramatic rise in Gen AI referral traffic: “Generative AI tools drove a 693.4% increase in traffic to retail sites”.

The implication was clear: that AI shopping is a megatrend increasingly influential to the record $257 billion in US holiday ecommerce spending.

I received pitches from Adobe’s PR firm throughout the season highlighting the growth rates. On at least one occasion, I asked what the basis of the growth rate was but never received the relevant supporting data. While ~700% growth might sound impressive, year-over-year growth rates like that tend to only occur from a very small base.

At the close of the season, my curiosity led me to check another reliable source for web traffic: Sensor Tower. This data provided the clarity needed to put this trend in context, revealing that Gen AI referral numbers increased from 0.1% of traffic to 0.6% between Holiday 2024 and Holiday 2025.

The overall growth rate in this traffic jumped ~200%, an impressive number to be sure. (Not quite as high as Adobe’s estimate, but directionally similar—the difference likely explained by Adobe’s footprint of ecommerce sites not including Amazon and other large retailers.) And yet, the total Gen AI referral still barely registers in the overall composition of ecommerce traffic.

Adobe’s headlines weren’t inaccurate, but they did lack important context.

Leave a comment

HEADLINE 2: Agentic Commerce is an Existential Risk to RMNs

One school of thought says that agentic capabilities will shift shopping to LLMs, drive ecommerce traffic to zero, and vaporize on-site retail media revenue in the process. And this will all happen by the end of the decade!

The directive is clear: retailers should forsake current initiatives—such as RMN buildouts—to adapt to this zero-traffic future.

Is this good advice? How should retailers assess the tradeoffs?

I decided to consult Emarketer, a reputable third-party source of forecast data, which has projections for retail/commerce media and agentic commerce through 2029. According to those estimates, “agentic commerce” (“AI platform on-site retail ecommerce sales”, per Emarketer) will reach $28 billion by that date. Meanwhile, commerce media ad spend will increase by $48 billion from its 2025 total. Thus, if retailers allocate resources based on the revenue potential, then retail media represents a bigger opportunity to pursue.

But retailers aren’t just accountable to revenue growth—their valuations depend on profits. And currently, ecommerce revenue on its own isn’t very profitable. In fact, many retailers are still in the red, and Walmart is only now going into the black on this channel. Meanwhile, retail media has huge gross margins—especially for O&O media like their ecommerce sites.

If we apply typical margins of 10% for ecommerce and 60% for retail media, the gap between the two opportunities gets more dramatic. Under this scenario, commerce media would produce $29 billion in operating margin vs. $2.8 billion for ecommerce occurring on AI platforms. (Feel free to use different margins assumptions, but the disparity will still be significant.)

There’s no way to predict this future with certainty, but retailers can make decisions rooted in rationality over hyperbole. And when you do the math, the answer should be clear.

Check out Andrew's Substack

HEADLINE 3: ChatGPT is Already the #2 Ecommerce Site

A chart began circulating on Linkedin late last year suggesting that ChatGPT’s meteoric rise as a commerce channel made it the #2 ecommerce site in web traffic after Amazon.

A back-of-the-napkin calculation used OpenAI’s self-reported stats of 2.5 billion prompts per day and 2.1% pertaining to purchasable products to get 52.2 million queries. This was then compared to Euromonitor’s US daily visits to Amazon, Walmart, Ebay and other leading retailers. Only Amazon topped 52.2 million.

It made for a great headline—if only it were close to accurate.

A comparable analysis using Sensor Tower data—based on US visits/sessions for desktop, mobile web, and mobile app—painted quite a different picture. (ChatGPT visit data was similarly multiplied by 2.1% to estimate product-oriented traffic.)

Here’s what it showed:

• Amazon traffic is 27x the size of ChatGPT’s product-oriented traffic.

• Walmart, Ebay, Target, and Home Depot also top ChatGPT by healthy margins.

• Google—which was conveniently omitted from the prior analysis—is bigger than all of them, including Amazon. (This was calculated based on total traffic multiplied by a reported 14.5% of searches being commercial.)

It’s probably not that useful to compare traffic for ecommerce sites to search and answer engines in the first place. But I get the point in drawing the comparison. However, what’s included shouldn’t be cherry-picked. If ChatGPT is included in this list, then Google should be too—and probably social media platforms, as well.

As coverage of AI and agentic commerce continues to intensify, sensational headlines will proliferate. Now more than ever, these headlines require a careful reading, a critical lens, and a grounding in facts.

Otherwise you’re just buying into belief. And when you believe too easily, you’re prone to hallucinate.

So don’t be surprised when you get burned.

Share

THE NUMBERS BEHIND THE AI AD ECONOMY

Back next week!

HEADLINES

Back next week!

Subscribe now

WHAT’S DEBBIE UP TO?

Podcasts

• The AI Edge with U of Digital. It’s been a while since I did a purely news-driven podcast and this one was rapid-fire fun. U of Digital’s Shiv Gupta and Myles Younger and I debated whether Sam Altman is a liar, why Anthropic’s Mythos announcement might have been a marketing tactic, whether AEO/GEO has jumped the shark, and the two big developments in consumer behavior and advertising that will tell me if we’ve gotten to a social media-like turning point in AI. Catch it on YouTube, Apple podcasts or Spotify.

• Retailgentic: I joined ReFiBuy’s Scot Wingo for a wide-ranging discussion about how AI is reshaping advertising and commerce and what brands should be doing now. Watch the full podcast here and check out the Retail Media Breakfast Club article above for even more!

• Genuin: I recently sat down with CMO Matt Wurst for an episode of Genuin Connections. I first got to know Matt during his time at 360i, and it was a ton of fun to catch up and unpack the rapid rise of AI media and the AI advertising economy. Watch the full video here, but if you only have time for the highlights, here’s a clip I posted to LinkedIn.

Research

• IAB + Sonata Insights: The AI Ad Gap Widens (read our 2024 study here)

• Mediaocean: 2026 Advertising Outlook Report (I wrote the foreword)

On Stage

• M Agency webinar, May 6, 2026, virtual. I’m joining senior leaders from M Agency for “AI and the Customer Journey: Make Your Brand Visible Where It Counts.” Register for the webinar here.

• Skai’s ShopAble, May 13, 2026, New York City. I’ll be presenting “AI, Consumers & Commerce: A Reality Check.” RSVP at my personal link here.

• Operative’s Inview 2026: The Media Intelligence Summit, May 14, 2026, New York City. I’m giving the opening keynote, “The AI Monetization Shift: Why Media Won’t Get 8 Years This Time.” Request an invitation to attend here.

• AdExchanger’s Programmatic AI, May 19, 2026, Las Vegas. I’m delivering the opening keynote “Consumers Aren’t Where Ads Expect Them Anymore.” Use my code JOINME20 for 20% off your pass.

• Path to Purchase Institute’s Retail Media Summit, June 3, 2026, Chicago. I’m delivering “Who Owns the Customer in the Age of AI-Fueled Commerce?”. Check out the agenda and register here.

Watch Debra's speaking reel

Debra Aho Williamson is the Founder and Chief Analyst of Sonata Insights, where she helps advertisers navigate the next major structural shift in media: the rapid rise and monetization of AI platforms. Over nearly two decades as an analyst at EMARKETER, Debra was among the first to forecast the growth of social media advertising and identify how it would fundamentally reshape marketing investment.

Today, she applies that same pattern-recognition lens to AI, analyzing how consumer adoption is accelerating monetization and redefining brand budgets, media strategy and industry power dynamics. Through Sonata Insights, she partners with leading platforms, agencies and technology companies to deliver research, advisory and strategic narratives that bring clarity and credibility to this transition.

OpenAI is continuing to make waves in the ad market. It’s projecting $100 billion in ad revenue by 2030, its CEO of Applications, Fidji Simo, is speaking at Cannes Lions, and it recently launched an ads manager for ChatGPT. This week, I’m analyzing what OpenAI’s aggressive goals mean for the growth of global AI media ad spending.

Speaking of Cannes, it’s just 2 months away. I’m working on a few fun speaking appearances and I have room for a couple more. Reply to this email or DM me on LinkedIn and let’s strategize!

Speaking of speaking, I’m thrilled to announce that I’m keynoting at Operative’s Inview 2026: The Media Intelligence Summit. On May 14 in NYC, I’ll be breaking down why AI is accelerating media transformation, why AI platforms are becoming media destinations and what it means for other media companies. Learn more about Inview 2026 and request an invite here. Hope to see you there!

Subscribe now

THE MAIN STORY

Can AI Media Become a $200 Billion Ad Market?

If OpenAI achieves its goal of becoming a $100 billion ad business by 2030, it would become one of the largest digital media companies in the world in just five years.

Whether it can do that is a topic many others have debated since the news broke late last week.

What I’m more interested in is this: What does a $100 billion ad business for one company imply about the total potential size of the next big market for monetizing consumer decision-making—AI media?

Before I go any further, I’m not creating a forecast for AI media ad spending. It’s far too early for that. However, this much is clear: No major media category ends up with only one player capturing the entire market. Instead, multiple platforms compete for a segment of the pie.

OpenAI achieving its $100 billion goal implies that the total global market for AI media ad spending could reach $200 billion by 2030, including China, where AI usage and AI commerce are already quite strong.

That would be an extraordinary growth trajectory, and it would put AI media into direct competition with legacy digital media categories at a much more rapid pace than anyone had expected.

I reached out to WARC for a few forecasts so I could create a comparison between AI media and global ad spending on other media channels. At $200 billion, AI media spending in 2030 would be:

• Equivalent to what advertisers will spend on retail media in 2026

• 70% of 2026 search ad spending

• 52% of 2026 social media spending

Search and social have been ad businesses for more than two decades, and retail media has been around for more than a decade. AI media is an entirely new category, and it could scale before its ad formats are fully defined.

Let that sink in.

A category that didn’t even exist until 2025 could scale faster than other prior media categories.

Who will participate in AI media’s revenue pool?

Just like search, social media and retail media, AI media won’t be a winner-take-all category. Yes, each legacy category has a dominant company (Google, Meta, Amazon), and there will be one in AI media too. But many other types of companies will exist in the category. They might include:

Distribution giants: OpenAI, Google, Microsoft, Anthropic, Baidu, Tencent, Bytedance

Retail leaders: Amazon, Walmart, Target, Temu

Social media stalwarts: Meta, TikTok, Reddit

Vertical decision platforms: Expedia, Yelp, Wirecutter

High-authority content sources: The New York Times, The Wall Street Journal

The long tail: Perplexity, Character.ai and other smaller AI apps and services

Unknowns: AI-infused properties, experiences and services that haven’t been invented yet

All of these players will compete for the moment of AI-driven decision and monetize that intent in different ways.

What will it take to get to a $200 billion ad market?

Even if OpenAI doesn’t achieve its $100 billion goal, the combination of all the players could still add up to $200 billion within five years.

What will it take to get there? Three factors will determine both the growth trajectory of AI media and which players become dominant:

1. User growth. A lot of people are focused on ChatGPT and how many users it needs to have to get to $100 billion. But OpenAI’s ambitions don’t lie solely with ChatGPT. Over the next few years, I believe it will build out an ad-supported media stack by launching and/or acquiring additional media properties. [See last week’s article about the TBPN acquisition for more on this.] Other companies will also need to demonstrate increasing use of their AI features and services and an expanding portfolio of ad-supported offerings.

2. High-intent interactions and metrics. The power of AI media won’t come from its ability to drive engagement metrics like impressions, visits per user or even time spent. Instead, it will come from harder metrics like recommendations acted on or transactions completed. AI media still needs to prove it works for these types of interactions, but if it succeeds, ad spending will scale quickly.

3. Advertiser trust. This is where legacy players have the potential to shine. Advertisers will need to trust that the companies will do right by them and their users, and that the ads will deliver positive results. Incumbents such as Google, Amazon and Meta that are investing in developing AI-infused advertising surfaces have the benefit of longstanding advertiser relationships. But if AI-native interfaces can develop ad formats that perform even better than the legacy players’ offerings, budgets will shift.

Leave a comment

What are the roadblocks to a $200 billion AI media ad market?

It becomes difficult to separate AI media from other types of media. As these experiences grow and expand, they may become more intertwined with existing ad categories, becoming an AI layer on top of things like search, social or retail media. Something like this already happened in social media, when video started becoming a bigger feature. Now ads on TikTok or Instagram Reels straddle the line between social and video.

One or more major AI companies face significant regulatory or legal challenges. Both Anthropic and OpenAI have recently warned that AI models are becoming so smart that there is the potential for them to do more harm than good (either on their own or in the hands of bad actors). Any impact on the companies’ ability to develop new products or build consumer-facing surfaces will limit their potential as advertising media.

Privacy scandals and other negative news that impact consumer usage. Consumers are less skeptical of AI than they used to be, but that could change if a major privacy breach were disclosed or if an AI model were to cause significant disruption or a security breach.

Bottom line

AI platforms and services are already disrupting the media landscape, and the pace of change is faster than expected. While the ultimate size of the market has yet to be determined, consumer usage patterns, the number of companies pursuing this opportunity and the rapid development of AI-driven ad products all point toward AI media becoming a transformative global business.

More about OpenAI’s $100B bid

Here are three excellent analyses of OpenAI’s ad revenue goal:

• Sizing OpenAI’s Ad Business (Tom Triscari/Quo Vadis)

• How can ChatGPT reach $100BN in advertising revenue? (Eric Seufert/Mobile Dev Memo)

• OpenAI’s $100 Billion Ad Ambition: Highly Unlikely, But Not Impossible (Brian Wieser/Madison and Wall subscription newsletter)

And here are my own thoughts on LinkedIn.

Share

THE NUMBERS BEHIND THE AI AD ECONOMY

Gen Z’s Worsening AI Sentiment

Whenever I tell my two Gen Z daughters about some cool thing I did with AI, the inevitable response is “Mom, don’t let AI be your brain!” New research from the Walton Family Foundation and GSV Ventures, conducted by Gallup, sheds statistical light on why they might be saying that. The percentage of Gen Z internet users who feel excited about AI has plummeted since last year. They’re also more concerned about the impact of AI on jobs than they were, and more skeptical of its impact on productivity.

My takeaway: Young people are often on the leading edge of tech usage, so it’s wise to pay attention to whether their worsening sentiment ends up impacting their willingness to use AI.

HEADLINES

Signals from The AI Ad Economy

AI Advertising Signals

OpenAI projects $100 billion in ad revenue by 2030 (Axios) — The new element is the scale of the forecast: OpenAI is now telling investors ads could reach $2.5 billion in 2026 and $100 billion by 2030, making its monetization ambition much clearer than the pilot test revenue-run rate story.

OpenAI has quietly launched its ads manager (Digiday) — This is the operational follow-through: a subset of advertisers is now testing an early ads manager, showing OpenAI is moving beyond sponsored-message sales toward the tooling needed for a scalable self-serve ad business.

Google to retire Dynamic Search Ads in favor of AI Max (Search Engine Land) — Google said AI Max is exiting beta after adoption by “hundreds of thousands” of advertisers, with eligible DSA, ACA and campaign-level broad match settings set to migrate automatically in September.

AI Commerce & Agentic Signals

Starbucks launches beta app in ChatGPT to fuel new drink discovery (CNBC) — Starbucks is shifting discovery upstream by embedding itself inside ChatGPT, where users start with moods or “vibes” instead of menus. This is one of the clearest examples yet of a brand trying to intercept intent before search, navigation or even menu browsing happens.

Pacvue Enters AI Agent Race With Amazon-Focused Tool — Pacvue’s new agent can analyze performance, generate Amazon Marketing Cloud SQL from plain-language prompts and automatically adjust campaigns, a sign that commerce media management is shifting from dashboard work toward governed automation.

Amex tries to “grease the skids” for agentic commerce (American Banker) — Amex launched a developer kit to enable AI agents to complete transactions using its network, with built-in agent verification, intent validation and payment credentialing. The key shift is its “closed-loop” model and purchase protection, positioning trust and liability as core infrastructure for scaling agentic payments.

Data & Performance Signals

The most “ethical” AI company might also be the web’s biggest freeloader (Business Insider) — Cloudflare’s early-April crawl-to-refer ratios were 8,800:1 for Anthropic and 993:1 for OpenAI, quantifying how much AI platforms can extract from publishers without sending equivalent traffic back.

Under the hood of Google AI Mode Ads (Adthena on LinkedIn) — New data from AI ad tracker Adthena, shared by CMO Ashley Fletcher, shows what triggers ads to appear in AI Mode and how text vs. product listing ads differ. Key finding: exact query match is rare across both formats in AI Mode.

This startup wants to replace marketing agencies with AI. Read the pitch deck it used to raise $4.5 million (Business Insider) — Uplane says clients improved return on ad spend by 30% in about six weeks versus agency performance, with one client, Aonic, seeing a 60% improvement, giving the AI-agency-replacement pitch fresher performance proof.

Emerging Players & Ecosystem

Dentsu Revamps Its AI Platform for an Agentic Future (Adweek) — Dentsu.Connect is being rebuilt to let planning, strategy, audience and workflow systems communicate inside a shared agentic environment, signaling that holding companies are preparing for more automated media and marketing operations.

Subscribe now

WHAT’S DEBBIE UP TO?

In the Media

OpenAI’s Big Ambitions; Tricks Of The Trade (AdExchanger)

OpenAI has quietly launched its ads manager (Digiday)

The AI Ad Economy Is Here (Retail Media Breakfast Club)

Podcasts

• The AI Edge with U of Digital. It’s been a while since I did a purely news-driven podcast and this one was rapid-fire fun. U of Digital’s Shiv Gupta and Myles Younger and I debated whether Sam Altman is a liar, why Anthropic’s Mythos announcement might have been a marketing tactic, whether AEO/GEO has jumped the shark, and the two big developments in consumer behavior and advertising that will tell me if we’ve gotten to a social media-like turning point in AI. Catch it on YouTube, Apple podcasts or Spotify.

• Retailgentic: I joined ReFiBuy’s Scot Wingo for a wide-ranging discussion about how AI is reshaping advertising and commerce and what brands should be doing now. Watch the full podcast here and check out the Retail Media Breakfast Club article above for even more!

• Genuin: I recently sat down with CMO Matt Wurst for an episode of Genuin Connections. I first got to know Matt during his time at 360i, and it was a ton of fun to catch up and unpack the rapid rise of AI media and the AI advertising economy. Watch the full video here, but if you only have time for the highlights, here’s a clip I posted to LinkedIn.

• Coming soon: I recently recorded a podcast with AdExchanger. Will share links as soon as I have them!

Research

• IAB + Sonata Insights: The AI Ad Gap Widens (read our 2024 study here)

• Mediaocean: 2026 Advertising Outlook Report (I wrote the foreword)

On Stage

• M Agency webinar, May 6, 2026, virtual. I’m joining senior leaders from M Agency for “AI and the Customer Journey: Make Your Brand Visible Where It Counts.” Register for the webinar here.

• Skai’s ShopAble, May 13, 2026, New York City. I’ll be presenting “AI, Consumers & Commerce: A Reality Check.” RSVP at my personal link here.

• Operative’s Inview 2026: The Media Intelligence Summit, May 14, 2026, New York City. I’m giving the opening keynote, “The AI Monetization Shift: Why Media Won’t Get 8 Years This Time.” Request an invitation to attend here.

• AdExchanger’s Programmatic AI, May 19, 2026, Las Vegas. I’m delivering the opening keynote “Consumers Aren’t Where Ads Expect Them Anymore.” Use my code JOINME20 for 20% off your pass.

• Path to Purchase Institute’s Retail Media Summit, June 3, 2026, Chicago. I’m delivering “Who Owns the Customer in the Age of AI-Fueled Commerce?”. Check out the agenda and register here.

Watch Debra's speaking reel

Debra Aho Williamson is the Founder and Chief Analyst of Sonata Insights, where she helps advertisers navigate the next major structural shift in media: the rapid rise and monetization of AI platforms. Over nearly two decades as an analyst at EMARKETER, Debra was among the first to forecast the growth of social media advertising and identify how it would fundamentally reshape marketing investment.

Today, she applies that same pattern-recognition lens to AI, analyzing how consumer adoption is accelerating monetization and redefining brand budgets, media strategy and industry power dynamics. Through Sonata Insights, she partners with leading platforms, agencies and technology companies to deliver research, advisory and strategic narratives that bring clarity and credibility to this transition.

OpenAI didn’t buy TBPN to get into podcasting. In this week’s issue of The AI Ad Economy, I break down three ways the acquisition advances its push into advertising. Plus, a new and improved news recap that delivers what you need to know across advertising, commerce, data and emerging players in the AI Ad Economy.

Also, big news! I’m thrilled to announce that I’m speaking at Skai’s ShopAble conference in New York on May 13th. This year’s theme is “Where Commerce Media Meets AI,” and I’ll be breaking down what’s actually changing, what’s still emerging and what retail leaders should realistically be paying attention to. Learn more about ShopAble and register here. Hope to see you there!

Subscribe now

THE MAIN STORY

OpenAI, TBPN and the Future of Media

OpenAI has acquired a tech podcast called TBPN, and many people are now puzzled. Some have called it a bid to create an owned media platform. Others say it’s a way to cultivate favor with the tech elite.

But it’s more than that.

The move to buy TBPN is also a foundational element in OpenAI’s quest to build an ad business. That’s something a lot fewer people are talking about—and something every marketing executive needs to be aware of.

Now: Build OpenAI’s credibility as a media business

At its most basic level, TBPN gives OpenAI a platform for discussion and commentary about tech and media issues that it cares about.

But this isn’t about OpenAI getting into the podcast business. The acquisition also brings several assets that will boost OpenAI’s credibility among agencies and brands as it accelerates the development of its ad business:

• Audience: High-value tech investors and executives

• Attention: A daily 3-hour podcast with consistent tune-in

• Credibility: Hosts that have deep connections in the tech space

Yes, OpenAI needs to build a lot of infrastructure, measurement and reporting capabilities to show that ads drive results on ChatGPT. TBPN gives it something that’s less quantifiable but goes a long way in ad sales: buyer confidence.

Next: A venue for upper-funnel ads built on conversation and trust

OpenAI didn’t buy TBPN because it would be a source of immediate new revenue. The Wall Street Journal reported that TBPN was expecting to generate just $5 million in ad revenue in 2025.

What it does provide is a way to gain experience with upper-funnel advertising. So far, ads on ChatGPT have been more tied to mid-funnel and lower-funnel goals such as research and purchasing.

Podcast advertising sits at the top of the funnel. And, even more compelling, the kind of advertising that exists on podcasts—conversational, host-read—is a prototype for what OpenAI wants to do with AI advertising. It has pledged to deliver ad formats on ChatGPT that are conversational, not interruptive, and trusted, not resisted.

“We see an opportunity for advertising in ChatGPT to be uniquely valuable for people. In a conversational interface, ads can be more relevant and useful, connecting people to new products and services that fit naturally with what they’re trying to do.” — OpenAI, Feb. 9, 2026

Personal voices and creator validation are also incredibly important to the success of podcast advertising. There’s a lot OpenAI will learn about how to integrate authentic points of view and creator voices into its ad offerings.

And a final thought bubble: TBPN could become a sandbox for testing ads that eventually link back to ChatGPT for conversion and purchase opportunities.

Future: An OpenAI media stack

This week, journalist Alex Kantrowitz published an interview with OpenAI president Greg Brockman, where he likened OpenAI to Disney, which has media businesses, studios, theme parks, etc. The context was more about OpenAI’s broader business than specifically about its media bets, but I believe that with time (and future acquisitions), OpenAI could create a full-fledged media stack with various revenue models, including paid advertising.

As part of this push, OpenAI’s “build-it-or-buy-it?” strategy will be worth paying attention to. It recently killed off Sora, its video-generation app, which some believed could eventually become a new type of social network. I always thought the appeal of an app where you could watch AI-generated videos was fleeting, but I still think the idea of making AI use more social has merit.

Other ad surfaces that could evolve from within OpenAI include its planned superapp that combines work and personal features; the OpenAI browser, Atlas; and its agentic commerce offerings.

Layer those with a few more acquisitions, and OpenAI will have a significant (and broad) set of media properties to deliver advertising.

Final thoughts

Share

THE NUMBERS BEHIND THE AI AD ECONOMY

More AI use = more trust

Two-thirds of US adults say they’re more proficient with AI than they were a year ago, according to a February 2026 survey by TD Bank. With greater proficiency comes more trust: The percentage who trust AI to provide honest and reliable information is up 12 points compared to 2025.

HEADLINES

THE AI AD ECONOMY WEEKLY SIGNAL TRACKER

AI Advertising Signals

Google: AI ads driving up to 80% sales lift for some brands (Search Engine Land) — Google says Aritzia saw an 80% revenue increase after adopting AI Max, a sign that AI-matched search ads are moving from abstract promise to case-study-backed performance claims.

ChatGPT ads favor clarity over creativity, new data shows (Search Engine Land) — Adthena’s analysis of 40,000+ daily placements suggests ChatGPT ads are standardizing around short, utility-first copy, with average headlines of 30 characters and body copy of 116 characters. Read more from Adthena CMO Ashley Fletcher on LinkedIn.

AI Commerce & Agentic Signals

Visa unveils AI agent payment platform (Axios) — Visa’s new Intelligent Commerce Connect lets AI agents make payments across multiple card networks, signaling that payments players are preparing for agents to sit inside the checkout flow.

Mondelez overhauls its $3.5 billion digital commerce strategy in era of AI search (Digiday) — Mondelez said retail partners expect 30% of site traffic to be agentic by 2028, prompting a rebuild of its AI visibility, product data and measurement systems.

What do you think of the new format for this section?

Data & Performance Signals

One in five ChatGPT clicks go to Google: Study (Search Engine Land) — Semrush found ChatGPT referral traffic grew 206% from January 2025 to January 2026, but 21.6% of outbound clicks went to Google and over 30% went to just 10 domains.

AI talk at retail events shifts to proving real results, defining a true strategy (Digiday) — At Shoptalk, Macy’s said users of its “Ask Macy’s” assistant spend 400% more, reinforcing that AI commerce discussion is shifting toward measurable outcomes and strategy proof.

Emerging Players & Ecosystem

CNN readies AI agents to buy and sell ads (eMarketer) — CNN is actively building agent-to-agent ad trading infrastructure with testing planned this year and scaled transactions targeted for 2027, signaling publishers are moving from exploration to execution timelines.

Subscribe now

WHAT’S DEBBIE UP TO?

Video Interviews and Podcasts

• Retailgentic: I joined ReFiBuy’s Scot Wingo for a wide-ranging discussion about how AI is reshaping advertising and commerce and what brands should be doing now. Watch the full podcast here.

• Genuin: I recently sat down with CMO Matt Wurst for an episode of Genuin Connections. I first got to know Matt during his time at 360i, and it was a ton of fun to catch up and unpack the rapid rise of AI media and the AI advertising economy. Watch the full video here, but if you only have time for the highlights, here’s a clip I posted to LinkedIn.

• Coming soon: I’m recording podcasts with AdExchanger and U of Digital in the next couple of weeks. Will share links as soon as I have them!

Research

• IAB + Sonata Insights: The AI Ad Gap Widens (read our 2024 study here)

• Mediaocean: 2026 Advertising Outlook Report (I wrote the foreword)

On Stage

• Skai’s ShopAble, May 13, 2026: I’ll be presenting “AI and Commerce: A Reality Check” at ShopAble, at Center 415 in New York, NY.

• AdExchanger’s Programmatic AI, May 19, 2026: I’m delivering the opening keynote presentation “Consumers Aren’t Where Ads Expect Them Anymore” at AdExchanger’s Programmatic AI, Las Vegas NV. Use my code JOINME20 for 20% off your pass.

• Path to Purchase Institute’s Retail Media Summit, June 3, 2026: I’m delivering “Who Owns the Customer in the Age of AI-Fueled Commerce?” at P2PI’s Retail Media Summit, Chicago IL.

  Watch Debra's speaking reel

Debra Aho Williamson is the Founder and Chief Analyst of Sonata Insights, where she helps advertisers navigate the next major structural shift in media: the rapid rise and monetization of AI platforms. Over nearly two decades as an analyst at EMARKETER, Debra was among the first to forecast the growth of social media advertising and identify how it would fundamentally reshape marketing investment.

Today, she applies that same pattern-recognition lens to AI, analyzing how consumer adoption is accelerating monetization and redefining brand budgets, media strategy and industry power dynamics. Through Sonata Insights, she partners with leading platforms, agencies and technology companies to deliver research, advisory and strategic narratives that bring clarity and credibility to this transition.