What I found refreshing about this conversation is how Evan goes deep into the realities of portfolio management, those large-scale groups of people, time, and company systems, to bring them together to deliver meangingful outcomes.

This is the reason “Adapt to Win” was created – to show how Lean Portfolio Management and an adaptive strategy approach can help companies beat the average of 70% failed initiatives at the Enterprise level.

Subscribe now

Five Takeaways

“Don’t show me your PowerPoint deck. Show me the portfolio investment decisions you’ve made.” (17:53)

1. Funding decisions, not the PowerPoint deck, are your strategy. Companies mistake strategy for an approved artifact. Evan Campbell says all you have to do is read the portfolio to see what the company actually chose to focus on. Large-scale, yearly budgeting with an October “planning season” means that orgs aren’t able to be nimble to capture opportunities that may come up during the year.

“You just scratch the thin little surface veneer of consensus, and you realize they have no alignment at all.” (32:15)

2. When everyone agrees, all you have is a vague strategy with no real choices. Everyone signed off at the strategy offsite because nobody had to make any hard choices. When executives don’t choose, every product and engineering team makes up its own goals, optimizing locally instead of for the whole org. The test: if the opposite of your choice is moronic on its face, i.e, “intuitive interfaces” — you haven’t really chosen anything.

“Strategy decay is obfuscating your signal. […] Inertia is the organization’s inclination to just not change what it’s currently doing.” (05:49)

3. Strategy decay is a failure of both leadership AND management. Evan focuses on the concept of “strategy decay,” something that’s clear when half your executive team can’t tell you what the company’s main objectives are. Five emails, a town hall, and the Mission and Vision printed on the coffee cups won’t fix it. Even a perfectly clear message won’t move an organization, because inertia is a separate force from decay. It’s trying to get your org to do something different from what it’s doing now.

“The silo leaders control goals and rewards within their vertical silo, and they’re all working in opposition to one another. 70% of the strategic initiatives fail, and we don’t realize our strategic objectives as a corporation.”

4. A function can kill its company by hitting every one of its targets. Procurement did nothing “wrong” by following its own rules to the letter — getting three bids, using only approved vendors, volume pricing, all textbook. Applied to a company whose advantage was 48-hour prototype turnaround against single-digit production runs, it destroyed the exact thing the parent bought the company for in the first place. And the blame went to the acquired company. To combat this, Evan suggests using this sequence: strategy first, then business model, then operating model. Procurement was a set of operating model decisions made without any awareness of the other two critical pieces.

“The fastest way to refine the quality of your strategy is to improve execution and speed up your feedback loops.” (22:31)

5. Execution and strategy quality. Evan believes a B-level strategy with A-level execution will beat the opposite every time. Not because execution is more important, but because good execution also runs the continuous improvement learning loop that improves strategy problems. Evan sees companies that are bad at execution that can’t tell whether the strategy or the delivery is the thing holding them back.

Resources Mentioned

Evan Campbell

• Adapt to Win (Wiley, 2026)

• Resources and recommended reading

• Chinook Ranch Consulting · Adaptivity LLC

Books and thinkers named on air

• Roger L. Martin — Playing to Win (Evan’s pick for favorite strategy book)

• Steve Blank — customer development

• Paul Niven and Ben Lamorte — Objectives and Key Results (Evan’s most-gifted book)

• Don Reinertsen — The Principles of Product Development Flow

• Eliyahu Goldratt — The Goal

• Christina Wodtke — on weekly OKR check-ins, from her Zynga work

• Sull, Sull and Yoder, MIT Sloan 2018 — the strategy-decay figures.

Two related pieces on KeyBank’s “Mass Affluent” strategy

• The Hidden $2 Trillion Opportunity KeyBank Discovered In Its Existing Customer Base

• Why The Part of Strategy Nobody Thinks About Is The Part That Decides Who Wins

Full List of Chapters

00:00 Intro

00:53 The CEO who tried five emails and the coffee cups

02:32 Strategy decay: 51% at the top, 13% at the front line

03:50 What leaders skip — silos, rewards, horizontal alignment

05:39 Decay vs. inertia: two different problems

06:29 Procurement won. The company lost.

10:14 What good acquirers do differently

11:50 Organizational coherence: strategy → business model → operating model

13:43 Where you must be exceptional vs. where “solid enough” is fine

14:18 Don’t show me the deck, show me the funding decisions

15:47 Monkeys with calculators: the ROI trap

18:27 Reading the real strategy off where the money goes

18:48 The question nobody asks: what have you killed?

20:10 Does AI change the build-vs-fund equation?

21:21 B-strategy + A-execution beats A-strategy + B-execution

23:28 Making KRs leading indicators, not binary checkboxes

27:13 OKRs are a strategy process, not a comp process

28:25 The Viking flotilla

29:36 Why boats should be customer segments, not silos

31:38 The offsite where everyone signed off in blood

32:59 The “moronic opposite” test

35:21 The OKR ghetto: when PMs invent the North Star

38:56 Adaptive portfolio management: capacity, demand, matching

42:48 Trim the tail

43:42 The liar discount

50:32 The Christmas tree effect

52:00 Rapid fire: books and influences

56:35 Where to find Adapt to Win

Share

I was working on a mobile healthcare app for the global market to gamify taking control of your care.

We were behind schedule and over budget as the design and build phases had taken far longer than expected. Worse, patients struggled to use the app, slowing growth and retention. And every fix or feature update took 2-3 times longer to build and push live.

When I dug into the situation, it became clear that our passionate, visionary designers had completely reimagined the user experience, replacing all standard mobile controls with custom ones and relocating nav bars.

Our users, many of whom were neither young nor tech-savvy, couldn’t figure out how to do even the most basic things in the app, much less compete in any challenges or win prizes while improving their health outcomes.

Where had our team really gone wrong?

Meet Mark Pincus

“When Meta comes out with Threads, do you believe that’s going to be a hit product or not? And why? It’s a Proven, Better, New.”

About 20 of us from Lenny’s community were on a call with legendary founder Mark Pincus, having just read his new “Life at the Speed of Play” book. I could immediately see the First Principle behind what Mark shared next:

“...throughout your journey as a product maker, [you’ll need] to hone your instincts. And to keep rinse and repeat, see when you’re right, a little right, very right, wrong.”

Mark continued:

“When do you believe that Snap’s product Spectacles is going to be a hit product? I don’t. I’m a big investor in Snap.”

In minutes, Mark Pincus gave us a quick way to size up Threads and Snap, and to “hone our instincts” through one simple framework.

Subscribe now

Three Words and a Billion Users

Now that AI has made building cheap and frictionless, anyone can build anything, so the job has become building the right thing.

That’s exactly the problem Pincus’ “Proven, Better, New” framework solves. It reverse-engineers products the same way we use the Strategy Choice Cascade to reverse-engineer strategy. But you can run it on any product in under five minutes.

Learning to see products through these three words shows instantly why they win or fail. And how to build your own so it doesn’t.

You may not have heard of Mark Pincus, but his Zynga studio built eight hit games and hosted more than a billion players over four years.

Innovating in the Wrong Places Is Killing Your Success

Pincus’ “Proven, Better, New” framework shows us that too many teams put far too much innovation into the parts customers rely on and want left alone.

And how they further damage their products’ chances by copying the parts they needed to invent to keep customers coming back for more.

How Threads Captured Two Out of Three

Proven in Mark Pincus’ framework includes those things you copy on purpose.

Your users have already developed a set of habits that other products have spent years and fortunes perfecting. And customer habits are one of the most powerful forces you can harness.

Threads literally ripped off X’s posting model so closely that nobody needed a tutorial to figure out how to read or interact with it.

Better is the feature that ten out of ten hardcore users would immediately see as a clear upgrade.

Threads lets creators port their existing Instagram audience over to the platform, instead of making them start over with zero followers. That’s how you generate 100 million signups on a new platform in just five days.

New is the novel hook no one has seen used this way yet that’s exciting enough to get people to switch and come back for. The lack of a truly “new” feature is what caused Threads usage to plummet by half within a month.

And this is the area Threads still hasn’t gotten right. They haven’t been able to create anything compelling enough to keep people on the platform that X doesn’t do better.

Spectacles – Only One Out of Three and Still Trying

You can see Spectacles as having the opposite problem, stuck as an all “new” product leveraging no “proven” customer habit, and offering nothing “better” than the phone already in the user’s pocket.

Snap has been at this for 10 years already (since 2016!), and took a $40 million write-down on roughly 300,000 unsold pairs back in 2017.

Innovation is essential, but watch out for too much “new” in the wrong places. I’ve repeatedly seen talented product leaders, UI designers, engineers, and founders feel they have to “innovate” and make something “super cool” to wow peers or win design awards.

Words With Friends – What Having All Three Looks Like

Former console game developers Paul and David Bettner rebuilt Scrabble for the iPhone, skillfully mixing in key iOS platform features, including push notifications, chat, and turn-based play.

“Words With Friends” launched in 2009 with no marketing and a tiny team yet amassed over 12 million downloads and dominated the budding mobile gaming market within a year. At the time, Zynga was raking in millions with its desktop-based Facebook games like FarmVille and Texas Hold ‘Em.

Knowing he had to make a bold move to stay relevant in the emerging phone-based ecosystem, Mark Pincus acqui-hired the Bettner brothers’ entire 23-person Newtoy team for $53 million in December 2010, establishing Zynga overnight as a force in mobile gaming.

The proven piece “Words With Friends” nailed was taking advantage of Scrabble’s eight decades of tile racks and letter bonuses that everyone knew.

The “10 out of 10” better lay in completely reimagining Scrabble for mobile, making it free to play, and compressing games into short, bite-sized 5-to-15 minute asynchronous sessions, so you could fit games into your schedule instead of requiring several hours to play.

The new feature was integrating with the Facebook social graph, so you could connect with your real-world, like-minded friends waiting in your feed to play with you.

At its peak, “Words With Friends” boasted 13 million daily active users.

Astoundingly, 17 years after launch, “Words With Friends” still ranks among the top mobile games (within the top 25 word games in the iOS App Store, and the #4 top-grossing word game in the Android Play Store).

Get Your Reps In. And Build Better Products.

Once you start using this framework, you’ll see every product around you in a new light.

And more clearly understand when you’re “...right, a little right, very right, wrong.”

Your most painful realization will be seeing how much effort you’ve wasted trying to reinvent what’s already been “proven” somewhere else. Even worse is stepping back and seeing where you’ve copied other products’ “new” features, explaining why the unique “hook” you hoped would keep people coming back isn’t working at all.

Refocusing on our healthcare app’s target persona, we brought back standard iOS and Android platform controls. We also simplified, made buttons bigger, and moved nav items back where our users expected them to be. Using these “proven” pieces allowed patients to begin embracing the app and gaining lasting benefits from taking ownership of their health.

Pick up Mark Pincus’ “Life at the Speed of Play” book here, and watch his interview with Lenny Rachitsky here.

Share

In 2000, Netflix was teetering on the brink of bankruptcy in its “DVDs-by-mail” era.

Looking for options to stay afloat, flip-flop-clad Reed Hastings and Marc Randolph strode into Blockbuster CEO John Antioco’s office and offered to sell Netflix for a mere $50 million. At the time, Blockbuster was worth $8.4 billion with 9,000 stores and over 50 million customers. Netflix was on track to lose $50 million that year.

Antioco “flatly declined” Hastings’ and Randolph’s offer.

Crushed, Hastings managed to lead Netflix from that setback and gradually gained momentum with a series of small strategic bets.

By getting video onto any screen in any form factor, and adding the Netflix button to TV remotes, it began to build an audience. It next licensed original content from existing studios tailored for that audience. And deeply understanding its audience allowed it to create original content with its own studio, eventually winning over 200 Emmys for its TV shows, and 33 Academy Awards for its films.

The Netflix of 2000 could never have made “House of Cards.” But it learned to shoot for that next goal just out of reach.

The all-powerful Blockbuster of 2000? It gradually slid into irrelevance because it was stuck trying to exploit an old business model when it should have been reaching for the next one.

Each company’s destiny was the result of the systems and processes it put in place to achieve its goals, its “machinery.”

More than goals

Some people think having a vague mission is enough – “We’ll win an Academy Award one day!”

Strategy is about designing a small number of choices, a bet that gets you from here to there – “Start producing original content.”

Once you’ve tested those choices by asking the right questions, we know it takes a story to rally your team – “Annabelle is searching for bold, original entertainment. The Netflix studio will uniquely serve Annabelle’s viewing needs by bringing her content from daring, creative visionaries.”

What’s still missing? A clear set of a few key things you have to be uniquely great at to deliver against that bet.

And to consciously make sure you’re building the underlying machinery that allows you to achieve everything.

Faster DVDs?

If all Netflix had tried to do was make mail-order DVD rentals run faster, the streaming future we take for granted today wouldn’t exist.

Designing a bet just past what your team can do now, and putting the systems in place to measure and improve your ability to deliver against it allows any team to do the impossible. What may surprise you is you don’t have to do everything perfectly.

You only need to focus on the main 3-5 things that get you closer to your bet.

In 2000, Blockbuster was wringing as much as it could from its customers, earning no less than one-sixth of its total revenue, or $800 million, from late fees.

But despite its dominant position, Blockbuster didn’t realize it was relying on an outdated set of systems and machinery to keep doing what it was already great at (collecting late fees, running stores) instead of innovating to remove friction from the user video-rental experience.

Netflix instead evolved a very different and unique set of machinery through its “No Rules Rules.” And with the breakout success of its original “K-Pop Demon Hunters,” Netflix has proven it’s on the right track.

The idea of “machinery” might sound abstract, but it comes down to just three calls: Where the money goes, who gets to decide, and what you keep score on.

Where the money goes

Netflix invests in a few key areas to keep it ahead of other streaming services.

Its $18 billion content budget for 2025, which it expects to increase to $20 billion in 2026, is the largest content budget dedicated to driving and sustaining its core subscriber growth.

It also pays top-of-market salaries to hire and keep the best people across creative, design, and technical leadership. Before his departure in early 2024 to start his own media company, Scott Stuber reportedly earned between $16 million and $18 million annually to lead Netflix’s original film division, and engineering leaders earn an average of $1.1 million. And that’s cash, not deferred stock.

It then empowers those people with budgets and an entire support network, gives them free rein, and holds them accountable for real results.

Who gets to decide

Netflix empowers its top leadership talent by allowing them to greenlight ambitious projects without committee approval.

When Netflix made its first foray into original content with “House of Cards,” metrics played a factor, but it wasn’t the only thing it paid attention to. Ted Sarandos noted that Netflix’s original-content bet was fueled by a mix of data matched with creative intuition.

The Atlantic Wire estimated that Netflix’s “House of Cards” bet paid off within three months, despite bankrolling two seasons for $100 million without a pilot.

What you keep score on

A gym counting sign-ups might know its revenue.

But only a gym that tracks who keeps showing up will know whether it’s going to survive.

Netflix has downplayed traditional industry metrics like subscriber counts or broadcast ratings, instead focusing on viewer engagement and churn metrics. Reed Hastings understood Blockbuster’s “gym membership run backward” business model was fundamentally flawed and wouldn’t survive because it preyed on its users’ forgetfulness.

Netflix regularly reviews and updates its analytics approach, using its “Weekly Top 10” to showcase metrics for every region in which it operates.

Your turn

All it takes is answering these four questions in one sentence each for you to start building your own machinery:

What are the most important things you’re doing with your team?

What is your team uniquely good at that allows you to deliver against those things?

How do you manage decision-making? Who gets to decide what happens?

What and how do you measure?

Where to go from here

You probably can’t answer those yet. Almost nobody can on the first pass.

Here is how the next section gets you there, as you’ll learn how to:

• Figure out what you need to be great at vs. what you need to just be good enough at.

• Understand how ready your team is to do the things that matter.

• Use three ways to close that gap.

• Use three lenses to design your unique machinery to achieve your goals.

Let’s design a bet for your team

I’m connecting with a few select product teams to run through these concepts.

If you’re interested, hit “reply” and introduce yourself.

Subscribe now

The few vs. the many

I was coaching a leader who’d developed a reputation for relentlessly driving their team to do as much as possible. Team morale was consistently among the lowest at the company.

When I suggested the leader could get better results by focusing on fewer, more important tasks and emphasizing effectiveness over efficiency, the leader shot back:

“I demand both efficiency AND effectiveness.”

The manager never let up on their team. Morale stayed low.

Not long afterward, they were let go.

A classic mistake is confusing the few things you need to invest in and be great at with the many things you can afford to be efficient with.

Reinforcements

Think in terms of “up” vs. “through” to figure out what your team needs to be great at to play your part in supporting the rest of the org.

“Up” means the few specialized things your team must be great at that the rest of the organization depends on. “Through” are the shared things every team relies on, which need to be solid and reliable enough not to hold anyone back.

The breakthrough here is that one team’s “up” becomes another team’s “through.” But any “through” layer that isn’t up to a decent level will set off a chain reaction, killing every other team’s ability to deliver.

Think about the relationship between airports and airlines. Airlines compete in a few areas of shared excellence, which really boil down to customer experience on the ground and in the air.

But no airline needs to build its own runway. Even the best airline will suffer at subpar airports, limited by too few runways or broken baggage systems.

If the teams across your org aren’t consistently served fast, professionally run, plentiful, and readily available “through” services like analytics, networking, or software management, your mission-critical product teams simply won’t be able to deliver.

This is how “efficiency” rapidly degrades into a series of rolling problems – Team A doesn’t get analytics, doesn’t understand what to build, so your entire retail network is stuck operating with outdated software.

“Up” vs. “through” is the first cut you make on your own list. If you’re working on a key strategic initiative, you’re delivering “up.” If you’re an operations team, everything you do is a “through” service the whole company leans on.

That “through” has to be dependable without being world-class.

“Up” is where Tool One starts. Here’s how one bank arrived at its own list.

Tool One: Name the few things you have to be great at to deliver your bet

When I first joined KeyBank, we were following a “financial wellness” strategy that helped clients be more money-savvy. But the bank tried to sell broadly to anyone possible.

Key shifted to pursue a strategy of “growth through focus” by targeting mass affluent clients with anywhere from $250k to $2MM to invest. As other banks were traditionally focused on either lower-income or ultra-high-net-worth individuals, they were completely overlooking the mass affluent sector. Even Key was initially serving only 10% of the available mass-affluent clients in its user base.

Key identified three things they needed to be great at to win with mass-affluent clients:

• Have the right investment talent

• Adapt the branch network

• Raise the digital experiences to match

Each of those three is an “up.” Adapting the branch network for mass affluent clients isn’t something the rest of Key leans on, and the choices that worked here would be wasted chasing ultra-high-net-worth families.

But doing those three things would take an enormous amount of investment and effort.

Here’s how Key would know this was a strategy worth shooting for.

Key’s Mass Affluent Bet Boosting Volume Money Story

We can use Rich Mironov’s “Boosting Volume” Money Story from his excellent “Money Stories” book:

Size of Audience X Value Per Transaction X Guesstimated Increase Range = Guesstimated Impact Range

Narrative: If we shift our focus to mass affluent clients in the local markets served by our branches, we can better serve them and increase our total fee income.

Size of Audience: 10% of 1 million, or 100k customers

Value Per Transaction: 1% annually of $250k – $2MM, roughly $2.5k – $20k

Guesstimated Increase range: 10-20% increase

Guesstimated yearly impact range: = $25MM–$400MM, a solid 8-9-digit impact.

The Money Story lets us know Key was on the right track with its mass affluent strategy.

It’s pretty easy to see what Key’s three things it needed to be great at where. Grading each of them is harder, because any grade only has meaning based on your unique bet. And you need shared words before you can even grade at all.

“We’re good at this” describes a team of eight who all think they can handle everything they need to do well.

But the real test might not happen until someone takes a break.

Tool Two: Grade them against the Traffic Light Audit

You can use a simple “traffic-light” system to audit your crucial three to five areas:

Red means there’s no one on your team who can consistently do it, and don’t yet know how to find the right person to do it.

Amber means you can do it, but it’s inconsistent or limited to a single person.

Green would mean you have multiple people across your group who can reliably and consistently deliver the needed work at a high level.

Here’s a test to really see what “green” means. If one person going on vacation would drop your grade and people are scrambling, you were amber the whole time.

Living in amber

One operations team I worked with years ago had several key strengths that showed up as solid “green.”

On paper, we were an unbeatable team. But in one major area crucial to our strategy, we were bright “amber.” Most of the process was stuck in one person’s head and kept changing.

Unfortunately, we never accepted we weren’t able to handle this, nor did we ever manage to simplify, streamline, and turn that area into a repeatable process that anyone could take on.

Being able to honestly see which of your areas are red or amber doesn’t mean you’ve failed. It just means you have a to-do list.

Here’s how you start to close them.

Tool Three: Close the gap by Building, Borrowing, or Buying

Think of these in terms of three options you’d have for preparing family dinner.

“Build” means shopping, finding recipes, and preparing food yourself, as well as sharing the task with your partner or child so they can also learn how to prepare food.

“Borrow” could involve having your mother-in-law come over to cook for you and your family.

And nowadays, “Buy” means “DoorDashing,” getting a hot, ready-made dinner delivered straight to your door.

They could all even be the exact same dinner, but each choice carries a different speed and price tag for getting dinner on the table.

Facing hard truths and responding

Had we done an honest review of our team, we would have quickly understood we simply couldn’t reliably pull off what was needed.

We didn’t see the three options at the time. “Build” would have meant sitting down with the one person who managed that area, simplify and standardize what was in their head, then document it as a repeatable process anyone could run. “Borrow” would have meant bringing in a contractor to review the process, tear it apar, and hand us back a better one. “Buy” would have meant hiring someone who’d already created and standardized complex processes like elsewhere.

For our team struggling with this process, we would have based our decision whether to build, borrow, or buy on how crucial this work was to every other team in the org, how quickly it was needed, how much it was costing us not to do this right, and how central this work was to our strategy.

Had we graded and accepted we simply couldn’t pull this off, we should have immediately stopped and designed a different bet.

Now that you’re starting to close your gaps around your few crucial areas, here’s how to build the machinery that holds them in place.

Tool Four: Three calls that build the machinery

Linear provides a great example for building a unique set of machinery that created and sustains its edge in world-class work management tools.

Where the money and effort go. Linear invests in Polishing Seasons, where the entire team stops shipping new features several times a year to improve overall product quality. It has dedicated performance-engineering specialists and a team fully focused on its Linear Sync Engine. Linear also invests heavily and continuously in its design system and offers anywhere from 1- to 5-day paid work trials during its hiring process to maintain its “talent density.”

Who gets to decide. Co-founder Karri Saarinen calls it “taste-driven development” guided by “finely-tuned intuition,” never optimizing through A/B testing. While Linear leadership listens carefully to customers, it holds final say on public roadmap decisions. Paid work trials require a unanimous “strong yes” from every team member the candidate worked with.

What they keep score on. Linear rejects classic SaaS metrics “for metrics’ sake,” like MAU and DAU, instead mainly tracking a “North Star” metric of active, engaged adoption among target “rising star” tech companies, CSAT on the daily experience, speed and performance benchmarks, and bug fix turnaround speed. Linear does track Net Revenue Retention (NRR), considering it a strong customer value indicator. Linear’s 145% NRR means existing customers consistently stay longer and spend more.

Through these choices, Linear achieved a $1.25 billion valuation, profitable since 2021 with around 100 employees, equalling $12.5 million in value per employee as of late 2025. When fintech giant Brex adopted Linear, they used it 47% more, achieved 55% greater clarity in connecting work to goals, and rated it 63% more satisfying than their previous tool.

Making your own three calls

Linear’s Polishing Seasons are how it gets great at craft, which is the one thing it needs to create a product the best software teams are happy to live in all day.

Write each of your own three the same way, as a single sentence in the form “we do X, which is how we get great at Y.” where Y is one of the few things you named in Tool One above. If a call won’t attach to one of your few, you inherited it and never intentionally chose it.

Then run one test on each. If a competitor with a different bet could adopt your call tomorrow and come out ahead, the call is generic and isn’t uniquely supporting your bet. Linear’s unanimous “strong yes” hiring gate would slow down a larger company competing on breadth of features.

But for Linear’s smaller team, built on taste and talent density, that same decision gate is what makes its talent density compound.

The importance of uniqueness

Linear’s choices are unlike those of any other SaaS app. Making unique calls across these three areas and matching them to its bet to win with the best software teams is exactly what powers its success.

Linear never stops improving, and neither should you. Because customers and markets never stand still, your machinery will decay, and you’ll be stuck going through the motions serving a business model that no longer exists, just like Blockbuster did.

But there is a way to make sure your machinery continuously improves and compounds like Netflix’s or Linear’s.

Reflecting leads to compounding

I was asked to come in as product and delivery lead for a new group being spun up.

We were vastly underfunded, stocked with new and untested talent, and working in obscurity on a much lower-priority product that had been stuck in the weeds for years.

All we had going for us was a few people who deeply cared and were invested in getting better.

Guided by a simple set of choices, and using a regular cadence of retrospectives every two weeks, we looked for ways to improve both our product and how we built it, our “machinery” that structured us for success.

Our product eventually became the place where everyone else across the org wanted to be featured.

And you see this with startups – these teams start with groups of inexperienced people, but they’re passionate and believe deeply in what they’re doing. They’re also deeply committed to learning and improving.

When you focus your cycles of reflection on the metrics that matter, and work to get just 1% better on a regular basis, the law of compounding kicks in.

That’s how Netflix went from near-failure to the streaming and original content market leader.

You know the theory and see the possibilities. Now dedicate a short length of time to identify your own key areas and build the machinery to keep it in place and improving.

15 Minutes to Build Your Machinery

Take just fifteen minutes to step through these four passes on a single page.

1. Name the Few Things You Have to Be Great At

• Input: your matched bet you’ve developed over earlier issues.

• Output: a clean list of three to five things you need to be great at.

2. Measure your team where it matters with the Traffic Light Audit

• Input: The 3-5 things you need to do to consistently deliver against your winning bet:

• Output: Where you’re doing well, and where you need to quickly get better

3. Close the gap through Build, Borrow, or Buy

• Input: Your scored top 3-5 things

• Output: Know which ones you need to build internally through long-term training, borrow from consultants, or buy by hiring or acquiring

4. Build The Unique Machinery

1. Input: Your top 3-5 things

2. Output: How to support them with a unique set of the right machinery

If pass two comes back all green, you’ve only named what you’re already good at. Go back to pass one.

But after you run this exercise and have your action items to move on, be aware of three ways teams start wrong or slide back once they’ve started.

Common Machinery Pitfalls and Fixes

Designing your strategy for a team that doesn’t exist

Business is littered with the remains of companies that pinned their hopes on bets designed for the teams they wish they had, instead of the teams they had.

The difference is thinking big but acting just a step or two ahead of where you are today.

As Netflix demonstrated, if you have the right machinery, you can achieve the impossible one step at a time.

But for the failing Netflix of 2000 to try to “swing for the fences” and create original content before it had a built-in, loyal viewing audience would only have spelled disaster.

Trying to be efficient with your few, critical things

We touched on this earlier with my experience, but one of the most damaging mistakes a team can make is trying to be too efficient with the 3-5 things you have to be great at to deliver on your bet.

You’ll see this play out when a team is young and innovative, achieves some success, and then brings in “professional” management. Traditionally-trained MBAs are used to looking at spreadsheets to estimate cost and profit. They might see large sums “wasted” on expensive design talent, and advise management to be more “efficient,” farming out design to agencies.

But as soon as those design people are let go, the user experience suffers, and customers leave in droves.

Probably one of the most “expensive” ways to be “efficient” imaginable.

Copying your machinery from someone else

Early on, when John Doerr was brought in to advise Sergey Brin and Larry Page, he introduced the OKR goal-setting and measurement system that had worked so well at Intel. Google, already leading with the greatest business model in history, was raking in cash by selling words. OKRs just poured gasoline on the fire of Google’s success. As with any system or process, what works great for one team may not work for yours.

Try out different things to find the right machinery and measures that fit tightly with and support your bet.

And learn from Blockbuster’s example. As customers and markets shift, shift what you keep score on and where your strengths lie to stay ahead.

As it found out too late, being great at collecting late fees would be a liability in the streaming era.

AI Prompts

These four prompts run in sequence to step you through the logic of the 15-minute exercise.

Each follows a Role, Context, Goal structure. “Role” tells the AI who to think as and which ideas to hold it to. “Context” is where you fill in your situation. “Goal” tells it what to produce. The <role>, <context>, and <goal> tags are intentional, so include them when you paste.

Every prompt opens by asking you questions before it produces anything. Answering those questions honestly is the hard work. If you find yourself unable to answer, you’ve exposed the gap the prompt was designed to surface.

Each produces a set of answers that feed the next, so you can’t start in the middle.

You can use them with Claude, ChatGPT, or Gemini.

Prompt 1: Name the Few Things You Have to Be Great At

<role>

You are product strategy advisor Michael Goitein, who combines Roger L. Martin's

"Playing to Win" strategy work with Rich Mironov's Money Stories.

You hold to three ideas.

A bet on where and how to win is one key cut for one lock, so the few things a

team has to be great at are the ones that would be wasted on a different bet.

No team wins by being great at everything. An album has ten tracks and two or

three singles that carry it. The list runs three to five and never more.

Most of what a team does is table stakes it keeps good enough, so calling table

stakes critical is the most common way teams dilute their focus.

You push back on vague answers. If I give you something that would be true of any

team in my industry, say so and make me sharpen it.

</role>

<context>

My bet:

Where to Play: [who you serve, what problem, in what context]

How to Win: [how you're different, and what trade-off that requires]

The Guesstimated Impact Range from the Money Story I built for this bet: [paste

the narrative and the range, or write "none yet"]

Where my team's time and money go today:

1. [activity]

2. [activity]

3. [activity]

4. [activity]

5. [activity]

What we've deliberately chosen not to do that competitors do: [describe, or write

"nothing yet"]

</context>

<goal>

Step 1. Ask me 4-6 questions about the gap between my bet and what my team can do

today. Focus on where we fall short. Useful questions include what breaks when we

try this now, what we've promised customers we can't yet deliver, and what a

competitor could do next quarter that we couldn't answer. Wait for my answers.

Step 2. From my answers, draft a longlist of what would have to be true for my bet

to work. Draft more than five.

Step 3. Cut the longlist to three to five using two tests, showing the reasoning

for every cut.

The bet test. Would this still matter if I switched to a competitor's bet? If yes,

it's table stakes I keep good enough. Cut it and say where it belongs instead.

The stretch test. Can my team already do this reliably? If yes, it's a strength,

not a stretch. Cut it, because the few are things we have to become great at.

If I gave you an impact range, use it once here as a calibration check. A list

that looks timid against a large prize usually means I named safe work. Say so and

push me to name the harder thing. Don't otherwise do arithmetic with the number.

Step 4. Produce the artifact as a numbered list of three to five, each written as

one sentence in this form:

"We have to be great at [specific thing] because our bet depends on [specific

consequence]."

Then name the one item that would hurt us most if we never got great at it, and

say what breaks first when it fails.

If I've given you too little to work with, tell me what's missing rather than

guessing.

</goal>

Expect one or two items to get cut as table stakes. Learn from that cut. Keep the final list open, because Prompt 2 grades it.

Prompt 2: The Traffic Light Audit

<role>

You are product strategy advisor Michael Goitein, who combines Roger L. Martin's

"Playing to Win" strategy work with Rich Mironov's Money Stories.

You grade what the evidence shows. Hope is not evidence.

Red means nobody here can do it consistently and we don't yet know where to find

someone who can. Amber means we can do it, but it's inconsistent or it’s limited to

one person. Green means many people across the group deliver it reliably.

You know Amber is the grade teams lie to themselves about, because work one

talented person delivers every time feels green right up until that person takes a

vacation.

You do not soften grades to be encouraging. A wrong green is worse than a red,

because a red gets worked on.

</role>

<context>

The three to five things I have to be great at: [paste the list from Prompt 1]

The evidence I can point to for each:

1. [who does this today, how often it goes well, what happens when it doesn't]

2. [same]

3. [same]

4. [same]

5. [same]

</context>

<goal>

Step 1. For each item, ask one or two questions that test the grade I'm about to

get rather than the grade I want. Ask how many people could do this without the

current owner, what happened the last time it went wrong, whether any of it is

written down, and how a new hire would learn it. Wait for my answers.

If I answer an item with adjectives instead of incidents, ask again and say why.

"We're strong at this" isn't evidence. "The last three times we shipped it, two

went smoothly and one slipped a month" is.

Step 2. Grade each item Red, Amber, or Green, quoting the specific thing I said

that produced the grade.

Step 3. Apply the bus test to every Green. If one person leaving would drop the

grade, it's Amber. Downgrade it and say so.

Step 4. Produce the artifact as a table:

| What we have to be great at | Grade | Evidence for this grade | What moves it up one level |

| ----- | ----- | ----- | ----- |

Step 5. Name the gap that puts my bet at most risk, and say why that one and not

the others.

Then tell me plainly if my whole list came back green, because a list with no reds

or ambers usually means I named things I was already good at instead of the

stretch my bet requires. If that happens, send me back to Prompt 1.

If my evidence is too thin to grade an item, mark it ungraded and tell me what I'd

need to find out. Don't guess a grade.

</goal>

Expect at least one green to come back amber. That’s the audit working.

Every red and amber is an input to Prompt 3.

Prompt 3: Close the Gap by Building, Borrowing, or Buying

<role>

You are product strategy advisor Michael Goitein, who combines Roger L. Martin's

"Playing to Win" strategy work with Rich Mironov's Money Stories.

You know there are three ways to close a gap. Build means growing it inside the

team through training, documentation, and turning one person's knowledge into a

repeatable process. Borrow means renting it from a contractor or partner who does

the work while the team watches. Buy means hiring or acquiring someone who can do

it from day one.

Build is slowest and stickiest. Borrow is fastest and leaves nothing behind. Buy

is expensive and only works if the person lands in machinery that lets them do the

work.

The choice turns on five things. How central the gap is to the bet, how fast it's

needed, what it costs every month it stays open, how many other teams are blocked

by it, and how the price of closing it compares to the slice of the Money Story

sitting behind it.

That last one governs when I have it. A gap standing in front of an eight-figure

range justifies a different answer than one standing in front of a rounding error,

even when everything else about them looks the same. If I don't have a Money

Story, rank the gaps against each other instead and say you're doing it, since a

relative ranking is weaker than a number and I should know that.

You say which option you'd choose, why, and what being wrong would cost.

</role>

<context>

My graded audit: [paste the table from Prompt 2]

The Guesstimated Impact Range from my Money Story: [paste it, or write "none yet"]

For each red and amber:

- How soon do we need this working? [timeframe]

- Who else in the org is waiting on it? [describe]

- Budget reality: [what you can spend, and what you can't]

</context>

<goal>

Step 1. Before recommending anything, ask me 3-5 questions across the gaps. Ask

whether we've tried to close any of these before and what happened, how long it

took the last person we hired into a specialized role to become useful, what

similar work has cost us when we contracted it out, and whether anyone outside my

team has to approve the spend. Wait for my answers.

A recommendation built without this is a guess dressed as advice.

Step 2. For each red and amber, work through all three options rather than jumping

to a recommendation. Name what each would take, roughly how long, and what the

team is left holding when it's done.

Step 3. Size each gap against the money. Estimate the slice of my Guesstimated

Impact Range that this gap is standing in front of, then put the cost of each

option beside it. Say the ratio out loud. If a gap's cost to close approaches the

slice it protects, say that plainly.

Step 4. Recommend one option per gap, reasoning in terms of the five factors

above, and name the condition that would flip your recommendation.

Step 5. Flag any gap where Borrow leaves us worse off. Borrowing something central

to the bet means renting the thing our advantage depends on, and when the contract

ends we're back where we started.

Step 6. Produce the artifact as a table:

| Gap | Grade | Slice of the prize behind it | Build, Borrow, or Buy | Why this one | What changes my mind | First move this month |

| ----- | ----- | ----- | ----- | ----- | ----- | ----- |

Step 7. If two or more gaps are red, central to the bet, and can't be closed in

the time I have, tell me directly that the problem may be the bet rather than the

gaps, and say which part of the bet I'd have to revisit.

If I haven't given you enough on cost, timing, or budget to make a call, ask for

it. Don't recommend blind.

</goal>

The AI will sometimes tell you the bet is the problem. Discuss it with your team before you dismiss it.

Prompt 4: Build Your Everyday Machinery

<role>

You are product strategy advisor Michael Goitein, who combines Roger L. Martin's

"Playing to Win" strategy work with Rich Mironov's Money Stories.

You know the everyday machinery is made of three calls.

Where the money and effort go, which is deliberate over-investment in the few and

deliberate under-investment everywhere else. This call is a forward bet, not a

cost exercise, and you treat it that way.

Under-investment has a floor. Everything outside the few still has to be handled

capably, because the work every team leans on starves every team when it's

starved. Skimping there is the most common way a team funds its stretch by

breaking the ground it stands on. When I propose taking money out of something,

ask what it holds up before you agree.

You also know money moves on promised savings before those savings exist.

Redirecting headcount or budget against a cost reduction that hasn't shipped and

hasn't shown up in the numbers leaves the work unfinished and the customers

angry. If I fund a change with savings I haven't measured yet, say so.

Who gets to decide, which is who can say yes without a committee and who can't. A

kitchen where the chef has to approve every plate leaving the pass is a kitchen

that can only move as fast as one person.

What you keep score on, which is the small set of numbers the team watches and,

just as important, the popular numbers it refuses to watch. A gym counting

sign-ups knows its revenue. A gym counting who shows up knows whether it survives.

You know machinery copied from another company usually fails, because the calls

that worked there were matched to a different bet. When I name a practice, ask

what it holds in place here.

You know machinery decays, so any design you produce has to say how it gets

re-examined.

You are concrete. "Invest in quality" is not a call. "Stop shipping features for

two weeks every quarter and fix what's broken" is a call.

</role>

<context>

My bet: [paste your Where to Play and How to Win]

The Guesstimated Impact Range from my Money Story: [paste it, or write "none yet"]

The few things I have to be great at: [paste the list from Prompt 1]

My audit and gap decisions: [paste the table from Prompt 3]

My machinery as it stands today:

- Where the money and effort go: [headcount, budget, and where your team's hours

  land, not where you wish they landed]

- Who gets to decide: [who signs off on what, and how long it takes]

- What we keep score on: [the numbers you report upward, and the numbers you watch

  daily, if they're different]

</context>

<goal>

Step 1. Question my description of my own machinery before you work with it, since

teams describe the machinery they intend rather than the one they run. Ask me:

- What did we spend the most engineering hours on last quarter, and does that

  match the list of the few?

- When did someone below me last say yes to something significant without asking

  permission, and what happened?

- Which number gets reported upward that nobody on the team looks at between

  reports?

- What would we stop doing tomorrow if it turned out nobody cared?

- What are other teams waiting on us for, and how often are they waiting?

Wait for my answers, then tell me where they contradict what I wrote above.

Step 2. Read my machinery against my bet and name the contradictions. Look for

money outweighing the few in favor of table stakes, sign-offs that slow down the

few, and metrics that reward behavior my bet doesn't need.

Step 3. For each of the three calls, propose two to three specific changes. Every

proposal names which of my few things it holds in place. If it doesn't attach to

one of them, cut it and say why.

For what we keep score on, name what to stop measuring as well as what to start.

Include one number that's standard in my industry and useless for my bet, and say

what it would tempt us to do wrong. Include one number that tracks progress toward

my Guesstimated Impact Range, so the machinery and the money point the same way.

Step 4. Run two tests on your own proposals before you show them to me.

The copying test. If a change would work equally well for a competitor with a

different bet, it's generic. Sharpen it or cut it.

The funding test. For every proposal that moves money, say where the money comes

from. If it comes out of something other teams depend on, name who gets slower and

ask me whether that's a trade I'd make out loud. If it comes out of savings that

haven't shipped and haven't shown up in the numbers, the money isn't there yet.

Say so and hold the proposal until the savings are real.

Step 5. Produce the artifact in this shape:

**Where the money and effort go**

- [change] holds in place [which of the few] because [reason]

**Who gets to decide**

- [change] holds in place [which of the few] because [reason]

**What we keep score on**

- Start measuring [metric] because [reason]

- Stop measuring [metric] because [reason]

Step 6. Tell me how often to revisit these three calls, what signal would mean the

machinery has stopped fitting the bet, and what one question to put on that

meeting's agenda.

Step 7. Name the one thing my team should look at every two weeks to get a little

better at the few. Tell me what I'd see in six months if we did it, and what I'd

see if we didn't.

If any part of my current machinery is too vague to work with, say which part and

what you need. Don't design around a blank.

</goal>

Step 1 is the hardest and most uncomfortable to run. Closing the gap between the machinery you have in place now and the machinery you need to run to deliver against your bet is a step most teams don’t even realize they need to take. And overinvesting in the table stakes things you just need to be good enough at never seems like a mistake at the time.

Run all four steps, and you’ve built your own complete stack. You’ll understand your next bet, have a clear list of the few things to help deliver against it, an honest grade for each, a decision on every gap, and a set of machinery to hold it together and make sure it continuously improves.

Aim higher and build your unique machinery

Don’t worry if your team can’t get there today. No team starts out being able to achieve whatever they want.

But think about what your team’s example of Netflix’s steps look like. What’s your “video everywhere?” How do you move to “licensing content?” And from there to “creating original content?”

When you create that bet for where and how to win, understand what you’re good at, and put the machinery in place to measure and improve where you need to get better, you can pull off things that would have seemed impossible.

But without that machinery, even the team with the greatest advantage will stay stuck in what it knows and slide into irrelevance.

Share

“If you think AI will solve all your problems, you don’t understand AI, and you don’t understand your problems.”

— Mike Goitein

Meet Dennis Berry

runs a unique marketing business from sunny Playa del Carmen, Mexico. He’s a long way from the mountains he grew up in and used to call home.

Consistency & Adaptation

Dennis’s big insight is that the life most people were trained for no longer exists.

He shares the example of the entrenched taxi industry, which had everything– it owned cars, licenses, and the customers, and still lost to Uber (which I wrote up here), because it spent billions trying to hold onto a broken model that customers hated instead of doing anything to fix it.

For Dennis, the operators who win are those who adapt the fastest, which beats the smartest and the best-funded.

Values-First, Not Cash-First

Dennis is a firm believer in writing a business plan for your life before you chase the money, and start with your values, because too many people skip that step and spend years working toward a goal that never fits who they really are.

He’s equally blunt about how true, lasting change happens. A few solid choices repeated over a long stretch, with all the boredom and setbacks that come with it, will outperform heroic, sporadic bursts of effort every time.

The same disciplined approach that gets someone sober or lean is the same one that builds a lasting company aligned with who you are.

The implications of AI for life & business

The last third of our conversation is the part most relevant to anyone building anything right now.

Dennis and I come at AI from opposite directions and meet in the same place.

I lay out how connection, storytelling, taste, and judgment remain as the only things AI can’t copy. Dennis shows how every product, including ChatGPT, Claude, and social media, all run on the dopamine economy, that obsession with a constant hit of red dots, badges, shields, and alerts. Short-form video and the sensationalist narratives promoting the news work on the same receptors in the brain.

Dennis’s alternative: Get outside, build real relationships, and play a long enough game that a contact from six years ago can still become a twenty-five-thousand-dollar client.

Key Takeaways

• Adapt faster than you resist. The entrenched taxi industry owned the cars, the licenses, and the customers, and still lost, because it spent billions fighting Uber instead of fixing the model everyone already hated. Faster adaptation beats being the first to market or the smartest.

• Write a business plan for your life, and start with your values. Most people can name where they are and where they want to go, but they’ve never defined their values, so they spend years chasing a superficial goal, usually money, that was never aligned with who they were in the first place. Life isn’t about achieving goals so much as striving toward them in a way that’s aligned with your values.

• A few solid choices repeated regularly will outperform sporadic binges of intensity. Nobody undoes twenty years of a problem in a week. What kills most people is having no system for change, hitting one early setback, and giving up. Not quitting, adapting, and consistent regular effort gets you most of the way there.

• In an AI-saturated market, being human is the edge. Twenty years ago, having a website set a business apart. Now that everything is automated and generated, the standouts are connection, storytelling, taste, and judgment, because machines can only imitate those aspects.

• Relationships compound over time, so build the pipeline before you need it. A contact Dennis hadn’t spoken to in six years called out of the blue with a client referral who eventually signed for five-figure contract. I coach people to favor direct messages over public posts about five to one, because that’s where trust and connection get built.

• Be discerning about where you bet on AI. Dennis believes that only the big frontier LLMs like OpenAI or Anthropic can afford to absorb their mistakes with billions of investors’ money. The AI we use in five years won’t resemble today’s LLMs, and nobody knows yet what it will be.

• Everything is a path to marketing, so guard your attention. The evening news carries maybe ten minutes of actual news wrapped in hours of manufactured drama, built to keep you watching through the ad breaks. Short-form video wreaks havoc on your dopamine in the same way. Set a timer, then get off the screen.

Highlight Quotes

“Just because I live here (in Playa del Carmen) doesn’t mean I don’t have problems. I just chose to do my problems at the beach.”
— Dennis Berry

“Most overnight success stories take 2,000 nights. That’s five years.”

— Dennis Berry

“If you think AI will solve all your problems, you don’t understand AI, and you don’t understand your problems.”

— Mike Goitein

“It really is about connection, storytelling, taste, and judgment. AI can do a reasonable facsimile, but it can’t really do it. And we don’t want it to.”

— Mike Goitein

4. Resources & Links

• Elite Leaders Network — Dennis’ newsletter on leadership and business community

  • Upgrade to Premium here

• The Investor’s Newsletter – Dennis’s investing newsletter, a separate publication on value and fundamental investing

• Figma and Figma Config — Figma is the main design tool I support at work, and the Figma conference I attended at the Moscone Center in San Francisco

• Laurie Anderson installation at the Hirshhorn — the D.C. exhibit I reference currently showing at The Smithsonian’s Hirchhorn Gallery. It’s here that I saw the line about the fallacy that technology, any technology, will solve your problems

• Bill Maher clip — the YouTube segment I mentioned on soccer fans from around the world seeing America for the first time. And helping us remember things aren’t so bad here, after all.

Get more from Mike Goitein in the Substack app

Available for iOS and Android

Get the app

Years ago, our division lead asked me to coach their managers in strategy.

Tired of our group’s nonstop firefighting, they wanted me to help the other leaders think more strategically to achieve division goals.

We stripped down to essentials, focusing on each leader’s “Where to Play” and “How to Win” choices, and how they rolled up to overarching goals. Progress was immediate. One leader turned their group around entirely. Another uncovered a hidden strength and made it even stronger.

We would walk through each leader’s matched WTP/HTW set, focusing on the raw strategic logic. Yet despite our initial success, something still felt off about how we were rolling out the strategies.

Eventually, another major initiative derailed our strategic focus, and the group rapidly slid back into task-level firefighting.

Beyond the logic, I was clearly still a missing piece to strategy.

The strategic logic survived. Don’t be surprised if no one follows it.

You do the hard work, designing multiple matched WTP/HTW sets and testing your “What Would Have to Be True?” conditions. You abandon the sets with insurmountable barriers. The lone matched set that survives has strong, defensible logic, and your team can actually deliver against it.

But none of that matters outside the team that designed it. Because while people like “winning,” no one outside your core team cares that “strategy” made it happen.

There’s no chance in hell that any strategy alone will inspire anyone to stay on track through the long, uncertain effort of execution.

People don’t speak the language of strategy. That’s your job.

So how can you bridge the gap between raw strategic choices and rallying people around a common purpose?

Moving the focus outward

Anytime you present strategic logic, the human brain is pre-wired to hunt for any weak link.

It’s helpful behavior in a “What Would Have to Be True? (“WWHTBT”) testing session. But fatal when you need a room full of people to commit and move in the same direction.

This is where stories can convey the essence of strategy but connect one level deeper.

A well-designed customer story runs what Chip and Dan Heath call a “mental flight simulator.” By engaging empathy, stories stop people from picking apart the logic and begin feeling the experience for themselves. And once someone has mentally experienced a customer’s challenge, the strategic choice that resolves it is no longer an abstraction they can argue with.

And now becomes a clear path toward resolving a problem they’ve felt for themselves.

One customer. One story. Many ways to make it real.

Three audiences inside your company will need to feel that story before any customer ever benefits from your strategy: Executive sponsors, management, and frontline teams.

But each group won’t need its own story. They’ll only need to be moved by the same customer’s story, and come up with the answers to two key questions: What does this story mean for me in my role? And what do I have to do differently to make it real?

If this is starting to sound a lot like the storytelling side of marketing, kudos. You’ve been paying attention.

Breaking down the wall between strategy and marketing

AirBnB’s Brian Chesky collapsed the product manager and product marketing manager roles, a breakthrough move he shared on Lenny’s Podcast.

Chesky’s insight was that designing the strategy, rallying people around it, and generating customer demand for it were all part of the same activity. But Chesky was simply following in the footsteps of P&G, which had merged its strategy and marketing departments way back in the 80’s. Apple, AtScale, OmniOn Power, and Pendo have all followed suit. It’s now everyone’s job to figure out both why we should build something and why anyone should care.

User-based narratives are part of that convergence, as strategy and story form two sides of the same customer-centric focus.

Your move from strategy to story

With this convergence, everyone’s job is now to be both strategist and storyteller.

No customer will ever benefit from your strategy unless you first wrap it in a customer-centric story and get it to travel through your organization. The paid section walks you through three frameworks to do just that.

You’ll learn how to create a compelling narrative and interpret it for each of your three main internal audiences. And you’ll also pick the right Money Story for that strategy and learn to defend it.

After completing the exercises and using the prompts, you’ll match your tested strategy to a story, a rollout plan, and a Money Story you can bring into your next planning cycle.

Join us!

Read more

“What AI features should we ship?”

IBM asked a radically different question:

“Which of our capabilities could AI make dramatically better?”

Looking at itself as “Client Zero,” IBM’s answer generated $4.5 billion in annual savings, contributed $12.7 billion in free cash flow, and launched a new AI consulting product offering that continues to reap compounding benefits.

They stopped asking “what AI features should we ship” and instead asked, “which of our capabilities could AI make better?”

Subscribe now

Capabilities, explained

Roger Martin’s Strategy Choice Cascade is one of the most accessible strategy frameworks available. He outlines a set of five decisions that need to be made in order to have a fully-formed strategy. One of the core and most frequently misunderstood decisions is a component called capabilities.

Capabilities are integrated systems of people, processes, and technology explicitly chosen to deliver against a set of strategic choices. Think of them as the organizational muscles that let you execute your strategy.

AI has the potential to make these systems more powerful, more distinctive, and, with the appropriate feedback loops, continuously compound.

At IBM, they identified ways that service delivery could be enhanced with AI. Rather than shipping new features, they took the people and products they had and gave them easier access to IBM’s proprietary data, processes, and expertise.

They used AI to invest in their strategic capabilities. In terms of Roger Martin’s strategy cascade, they used AI to build capabilities that doubled down on their “Where to Play” and “How to Win” choices.

Your ability to unlock AI’s full strategic value lies not in any single feature, but in creating an underlying system that continuously enhances capabilities. Those capabilities build compounding waves of competitive advantage.

The four types of AI-enhanced capabilities

As we see it, there are four types of capabilities that AI is best suited to enhance. Each addresses a distinct human limitation. As you layer your capabilities, you start to move more effectively than your competition, which creates a durable competitive advantage.

Speed

AI helps you do something you already do, but much more quickly. Examples include:

• Writing PRDs and product briefs

• Code completion

Speed provides the least durable advantage because you’re still doing the same work – something competitors could match by adding headcount or building the prompts themselves. But working on speed helps you invest in other capabilities later, opening up more opportunities.

Consistency

We’re human and frequently drop the ball. AI never does – and never resents doing the same thing repeatedly. Like speed, consistency is where AI helps you do something you already know how to do, but ensures you do it every time. Examples include:

• Drafting meeting notes

• Writing documentation with every code update

• Updating the CRM after every customer call

Consistency similarly provides limited durable value because it operates within your current paradigm. Consistency can drive early savings in your strategy, again freeing up brainpower for deeper enhancements later.

Scale

Scale is where AI starts to unlock real competitive advantages. AI allows us to do things at volumes humans can never match – and find patterns in that volume that humans would miss. Examples include:

• Monitoring all network traffic for anomalies

• Tagging and pattern-matching based on every customer support ticket

Scale increases value because volume reveals patterns that smaller samples would miss. This is where you start seeing inputs into your strategy system that you hadn’t otherwise.

Skill

AI’s most intriguing opportunities lie in creating capabilities that otherwise wouldn’t exist across the organization. This compounds when you translate skills from one group to another. Examples include:

• Creating agents that do code review so product managers can push code that works well with the rest of the architecture.

• Building an editor that mimics the brand voice so anybody can draft marketing copy.

• Publishing a product strategy tool so anybody can suggest a new feature idea and get help refining it into something worth building.

Skill is where the most compounding value comes from. You’re enabling people to do things with fewer dependencies and greater consistency. The more AI-driven skill augmentation, the more time they have to keep investing in capabilities later on.

The most lasting value comes when you cross categories. If you can provide a skill that’s also executed with great consistency and surfaces insights across the organization at scale, you’ve built capabilities that uniquely enable your strategy.

One note across all four: don’t automate a broken process. AI can make a bureaucratic headache faster, but it’s still a headache. Before you scale anything, take a moment to rethink and strengthen what you’re enhancing.

Customers probably won’t see most of these directly. The capabilities layer focuses on internal operations – making what you already do more effective.

The more you build AI capabilities that enhance your Where to Play and How to Win, the stronger your flywheel becomes.

What makes AI capabilities compound

The most powerful way to get capabilities to unlock this flywheel is through combining different types together in a way that supports your strategy.

“Have great AI capabilities” is not a strategy. It becomes strategy when you choose particular capabilities that reinforce your “Where to Play” and “How to Win” choices.

Disconnected AI features create noise. Integrated AI capabilities create a flywheel effect that exponentially increases your strategy’s effectiveness as the capabilities strengthen.

By investing in your internal capabilities, it improves operational efficiency in your key strategic areas. This allows more time to be spent on differentiated activities, which leads to more customer value being delivered in the product. You can take those savings or additional customer value that you created and reinvest it back in internal capabilities, which starts the flywheel over again.

No matter what type of capability you’re developing, you want a clear through-line to how it strengthens your strategy.

IBM decided to use AI to tackle basic HR processes. The results exceeded their expectations as they automated 94% of HR inquiries, freed up 3.9 million employee hours, and amassed $4.5 billion in annual savings. And of course, IVM cells.

This connected back to their strategy, which included reducing bureaucratic overhead across 340,000 employees. Reducing turnover has been a core goal of theirs for years, and it’s easy to understand how too much red tape makes people want to leave.

This investment will compound. Lower turnover means current employees increase in value. A more effective workforce can invest in improving themselves.

This flywheel unlocked customer value: internal capabilities compounded first, and when they were strong enough, they became the product itself.

How IBM turned internal capabilities into an advantage

Here’s another example of how IBM leaned into their “Where to Play” and “How to Win” through capabilities:

They created bots to answer internal support questions. This meant that employees were getting immediate answers for 70% of their inquiries. They could then serve clients faster, solve problems more quickly, and it showed with a 25-point increase in CSAT and $165 million in savings. They are now taking that extra time and money and reinvesting it in solving new problems or finding other areas to improve service delivery.

But those results in turn created two additional, even greater benefits. The savings IBM achieved allowed it to fund continued AI investment. Even more importantly, IBM was able to “productize” and resell those same capabilities, creating similar savings for clients. The internal transformation became IBM’s most credible sales proof point, with over 1,000 “Client Zero” client engagements in 2025 alone.

Even if they had never sold these AI-enhanced capabilities to customers, delivering this level of measurable AI benefits internally would still be a differentiating strategy and a strong move for them against consulting rivals.

The fact that they were able to turn around and sell them, on top of the powerful inward transformation made it a stacking set of wins.

On IBM’s Q2 2025 earnings call, CEO Arvind Krishna told investors how AI capabilities had given the company “guidance and confidence… about reimagining and reinventing how we run our company.”

AI-enhanced capabilities exist to execute your strategy

There’s massive value to the AI work no one outside your company sees.

Competitors can screenshot and match your customer-facing AI features within a quarter. Internal AI capabilities, layered across speed, consistency, scale, and skill, show up differently, with faster decisions, cleaner data, fewer dropped balls, and more capacity per person.

The first kind generates press coverage. The second kind generates margin.

IBM didn’t start by selling its AI capabilities to clients.

It slowly built its AI internally for two years, watched what happened, adjusted, and only then turned the capability outward as a service offering. By the time its productized version hit the market, IBM could already confidently tell the story, having raked in $4.5 billion in internal savings. That internal use was the credential that allowed them to scale these same skills outward.

Clients couldn’t wait to get some of what IBM had already done for itself.

Capability advantages that compound are the ones customers can’t clearly name, only feel through steadier experiences, quicker resolutions, and account managers who somehow have the time to actually pay attention.

The AI underneath remains invisible. That’s the version competitors can’t copy, because by the time they figure out what you’re doing, you’ve already built the next layer on top

Expressing appreciation & thanks for my co-author Jenny Wanger.

Jenny’s thought leadership across product operations and strategy is top-notch. Read more on her blog & subscribe below.

Share

Your product team likely never tests. Anything. You just build the next thing you’ve been told to build on your stakeholder-dictated roadmap.

And when you do test, it’s only to “validate,” to look for approval from customers and stakeholders to show you were right all along.

AI tools like Figma Make, Lovable, and Replit let you build faster and at higher fidelity than ever before. That reduced friction has created a new problem. The prototypes Product, tech, and UX are building now compete with each other for customer review and stakeholder approval.

And none of them answers the question of whether you’re building something customers will pay for.

The Phoneless Future That Never Happened

Humane AI had $230 million in funding, Apple-pedigree founders, and a stubborn insistence on a vision they never properly tested.

Critics expressed skepticism as soon as their first commercial dropped, even as the team worked around the clock to ship. Behind the scenes, founders Bethany Bongiorno and Imran Chaudhri, both Apple veterans, enforced a toxic positivity culture, dismissing employees who raised hardware or software concerns for “violating company policy.”

Humane’s glitzy launch and celebrity endorsements were followed by scathing reviews just a week after its long-awaited wearable “AI pin” device shipped in April 2024.

Just four months later, in August, daily returns started outpacing new sales, with over $1 million in returned products piling up and thousands of orders canceled even before they shipped.

By February 2025, Humane’s flameout was complete. HP acquired most of the company’s assets for $116 million, roughly half of what investors had invested. The founders and part of the team were absorbed into HP’s new AI lab.

But by far the biggest losers? The thousands of customers who paid $699 to be part of the next-generation “phoneless future.” All they were left with was a bricked, worthless device as of early 2025.

Two decades of Apple design and engineering experience weren’t enough to save the founders from their failed vision.

The Humane pin turned out to be something they wanted to build, but not something people wanted to own.

Humane’s Four Failures

While we might call Humane a “failure,” they failed against four different risks that threaten every product effort:

• Value Risk: Not enough customers saw the value or were willing to pay for it.

• Usability Risk: Customers who ordered the pin couldn’t get what they hoped for from it. Returns outstripped new shipments within months.

• Feasibility Risk: Despite a roster of Apple veterans, the team couldn’t solve the hardware and software problems they refused to acknowledge. Quality suffered because internal feedback loops were shut down.

• Business Viability Risk: Humane’s business model rapidly fell apart once it hit reality. A $699 device plus a $24/month subscription, with unresolved safety concerns that forced a charging case recall.

What if instead of attacking its own people, Humane’s founders had remained open to internal and external feedback, and continued testing and prototyping with their industrial and software engineers and end users?

What could the Humane AI pin have become?

Subscribe now

Fail in Small Ways First

Humane failed in one spectacular, expensive, public way.

The alternative is to intentionally fail in many small, cheap, low-risk ways first. And that’s exactly what prototyping and testing do. It shows you where you’re wrong early so you can fix it before the stakes get that high and you can still change direction.

The different prototyping approaches below are how you do it with intention instead of luck.

The Three Ways To Use Prototyping

Prototyping and testing serve three distinct functions, each reducing a different kind of risk.

• Build to Think to try out new ideas and open up possibilities

• Build to Learn to run experiments and disprove assumptions of how users might react

• Build to Align to share a clearer picture with extended team members and leadership

These aren’t deterministic steps in a linear path. Teams move in between them in waves as their understanding grows.

Pair these functions with the stakeholder mapping from Issue 4, and you know who to test with, what to test, and how much to invest in each test.

Paying subscribers get the frameworks, fidelity levels, a 15-minute exercise to put all three functions to work, and my exact prompts to walk you through the exercise for additional support.

The Prototyping Challenge

Run these two questions against your current work before you keep reading.

1. Even if you’re not aware of it, you’re currently following a strategy. How many uniquely different possibilities did your team generate before you decided to follow that strategy? How did you prototype and test your choices?

2. Pull up the last three features your team shipped. How did you prototype and test each one? With whom? Which of the three functions did you use?

Surprised? Good.

Your Move

The only way to pressure-test the assumptions underlying your strategic and feature choices before your customers do is through prototyping and testing them first.

The paid section walks you through four frameworks to apply all three functions.

You’ll learn why fidelity matters in testing, and why higher fidelity isn’t always better. You’ll also learn how to match the right stakeholder ask to the right prototype.

And, no, “What do you think?” is never the right question to ask.

By the end, you’ll have a Test Design Plan, pre-committed assessment criteria, and a Money Story you can bring to your next planning cycle.

Let’s Walk Through How You’re Prototyping & Testing

Reply TEST and tell me about the prototypes your team has built recently, which approaches you used, and what you learned.

Read more

Microsoft’s Kin phones underwent 2 years of development, got a major marketing push, and had a clear teen target audience.

But problems started early, as the product team’s strategy was never fully shared or co-owned, even within Microsoft org itself. Long-delayed and a victim of internal politics and friction between software and hardware teams, Kin phones were doomed before a single one landed in a teen’s hands.

Worse, carrier partner Verizon controlled pricing and insisted on charging teens $30/month for business customer data plans on top of other base Verizon charges for voice and text.

Microsoft senior product manager for mobile communications, Greg Sullivan, somehow believed that it was all a “positioning” problem:

“Once [customers] realize the value of this, they’ll realize it’s a great deal.“

Not sure how anyone could spin a phone that lacked basic features like apps, maps, or even a calendar as a “great deal” for customers.

Worse, Microsoft’s rollout never gave Verizon a reason to support or protect a strategy they never helped design, just to risk their own recurring revenue.

The result was a clear Where to Play and How to Win that was killed by the way that strategy was brought to life. The Kin lasted just 48 days before Microsoft pulled the plug on one of its higher-priced failures, at a net cost of around $1 billion.

Think about your company. What do you think is the cost of spending one or more quarters of multiple teams’ fully loaded salaries on a strategy that dies? All because the other people, teams, and partners who needed to support it were never part of its design or given a reason to care?

Subscribe now

The best strategies are social

You’ve done the hard strategic work. You designed matched choices. You tested them. You drafted a Money Story. Now you walk into a room full of stakeholders and present it.

And somehow you think when they hear the brilliant strategy that you’ll get enthusiastic approval and expect everyone to spring into action and know exactly what to do to make it happen?

Fat chance. That has never worked, and will never work. But it’s not your strategy’s fault.

Strategy is social. People across your organization at every level need to understand the strategy, feel invested in it, be able to contribute to it, and see their own contributions as valued and represented as part of it.

But most product teams treat stakeholder engagement as either an annoyance or a single activity with two expected outcomes: present the finished work to unanimous praise, or have everyone necessary to the strategy’s success just fall in line and start “executing.” In reality, there are three fundamentally different types of stakeholders we engage with across three very distinct meeting dynamics, and each requires a different approach you need to design intentionally.

If you’re still treating every stakeholder the same, and every meeting interaction the same, you’re just scheduling meetings and hoping for the best.

And hope, as we all know, is never a strategy.

How to Throw Away Five Zeros

Years ago, I was coaching a leader on their function’s strategy.

At the most basic level, we were trying to be clear about who their group served and how they’d “win.” I had reservations about their choice of “end user.” Worse, they never discussed their approach with any other function leads. They simply decided and pressed forward.

Six months later, a cascade of issues emerged. Teams across the organization were pulled into fixing deep structural problems that required countless hours of additional manual work, dragging an entire division backward. The approximate cost: hundreds of wasted hours, on the order of high tens of thousands of dollars of wasted effort and delay.

Fortunately, the leader chose a different approach in our strategy work together the following quarter.

They stopped treating their function’s strategy as their choice alone.

They brought other function leads into the conversation. They found out what those leads were actually dealing with, what mattered to them, and how they could make choices together that reinforced the entire group’s efforts. They listened, adjusted, shifted, and formed a coalition ready to move forward with a different end-user customer focus and matched “How to Win.”

The following quarter started with training the right end-user, killing festering regulatory problems before they had a chance to start.

While the basic strategy remained fairly similar, because they invested in designing it collaboratively, everyone around it changed. The result was like a siloed kitchen staff finally meeting, agreeing on a menu, tasting, and talking for the first time before dinner service.

Same ingredients. Fundamentally different cooking and dining experiences for everyone.

More than you think

There aren’t two kinds of people: those you’re trying to get a “yes” out of to move forward, and the rest you expect to just “get with the program.”

There are no fewer than three, each with very different needs.

And every meeting isn’t about either “pitching” or “influencing.”

There are no fewer than three specific types of meetings you’ll need to have to build and bring your strategy to life, and they aren’t interchangeable.

The Hidden Cost of Not Building Your Strategy Collaboratively

Trying to shortcut strategy’s collaborative nature isn’t just personally and professionally damaging, it has a real financial impact.

A product team of five people (PM, Designer, three Engineers) at a blended fully-loaded cost of $200K per person carries a minimum quarterly cost of roughly $250K.

If your strategy requires support from 3-5 stakeholders to proceed (and that’s for a small org), and you haven’t understood them or designed the right way to connect with each, your strategy will die for at least one to two quarters. The cost of not intentionally designing your collaborative strategic moments is that anywhere from $250K to $500K in team salary is wasted working towards something that will never move forward and deliver its intended business and customer outcomes.

Add in the opportunity cost of what the team could have been building, and that number easily doubles or triples.

What can you do to stay ahead of this kind of product team strategy malpractice?

You can start by running a quick check on your current work.

The Closing Audit

Take just 15-20 minutes and run these two questions against your work in flight:

• How easy is it for you to name 3-5 people whose support is crucial for your strategy to move forward?

• How quickly can you estimate your team’s fully-loaded cost? How aware are you of the potential cost of a single stalled month?

What’s Still Ahead

Knowing you need people is the easy part.

But learning the hard part, the part that can save your company $250k/quarter, means knowing:

• Exactly which of the 3 meeting types to use for each person.

• How to avoid the “Deference Trap” with senior VPs.

• The exact AI prompts you can use to map these interactions in minutes.

Paid Subscribers Learn How to Get the Right People On Board the Right Way

The free section showed how great strategies die when product teams treat stakeholder engagements as either a burden or a checkbox.

The paid section teaches you how to design the right moments for the right people. You’ll learn how to understand your stakeholders’ needs, design the right meeting interactions for each, and apply Roger L. Martin’s revolutionary “Jobs to Be Assigned” framework.

Three artifacts, each building on the last, ready before your next leadership meeting, so that your strategy’s chance of success isn’t left to chance.

Your Turn to ENGAGE

I’m looking for a few product teams and Product Trio leads (PM, UX, and Tech) on active product teams who want to give their strategy a better chance of succeeding, and learn what it takes to design stakeholder engagements before walking into their next strategy meeting.

Reply ENGAGE and let me know a bit about your team and your upcoming strategy conversations.

Read more

What’s Changing As of Now

This issue deepens the pivot blending my 30 years of engineering, design, product and leadership experience with ’s “Playing to Win” strategy framework by adding Rich Mironov’s “Money Stories” approach. Product teams start by reverse-engineering, designing, and testing sound strategy choices. The big shift is that now, once we’re ready to share a strategic bet with leadership, Rich gives us the tools to finally translate it into the financial language our executive stakeholders prefer.

Most product teams do one of those things, or neither.

The danger of consensus

The moment everyone in the room says your strategy is great is when you should be most worried.

You present the deck. The VP nods, asks a softball question. Sales says they’ll have an easy time getting customers to buy. Engineering says it’s no problem to build. And marketing says it’s consistent with your brand messaging. You leave the room feeling safe.

That comfortable feeling is the first sign your product is going to fail.

Why? Because choices that make everyone happy never lead to business success. They’re usually a watered-down collection of “best practices” and “customer-centric” platitudes that describe everyone and no one at the same time.

Effective strategy requires real, hard trade-offs that limit options in ways that inevitably stress people out. And that’s a good thing. Because making choices means you understand who your real customers are, how your business makes money, and are ready to focus on building competitive advantage.

And the only way to know whether you have a real strategy, and not just a wish list, is to test its logic.

AI meeting notes startup Granola’s experience with Vanta shows what real strategy looks like.

Subscribe now

Granola’s Enterprise Revenue Acceleration

Just as AI augmented meeting notes startup Granola was gaining traction, Enterprise deals were stalling.

Some Enterprises flat-out refused to engage until Granola achieved SOC 2 compliance. After choosing Vanta, Granola achieved certification in just over three months, turning compliance from a bottleneck into a revenue driver. Granola could not only reach new Enterprise customers, but also reconnect with previously-stalled accounts.

Roughly how much revenue growth was Granola’s SOC 2 certification worth?

There are no publicly available figures or way to know for sure, but we can test the underlying strategy logic and apply a simple formula to get a sense of how many “zeroes” Vanta’s certification made possible.

Let’s start by testing our logic with strategy’s most important question.

What Would Have to Be True?

Product teams frequently waste months arguing among themselves and with stakeholders about which features will work.

They’re focused on fighting for what they assume is true, and don’t understand why no one else understands the logic and evidence exactly the way they see them.

Roger L. Martin’s breakthrough question, “What Would Have to Be True?“ (”WWHTBT”) allows teams to break logic and evidence apart. Instead of debating what you believe, you agree on what conditions must hold for your strategy to work.

For the first time, you can separate your assumption from the data you need to prove it.

In our last newsletter, we introduced the importance of Where to Play and How to Win as an inseparable matched pair. WWHTBT is your internal logic test for that set. If a strategy can’t name the exact conditions that must hold true for that pair to be a good idea, the product team needs to go back to the drawing board to design another matched set.

Here’s Vanta’s matched Where to Play and How to Win:

• Where to Play: Smaller startups looking to move upmarket and capture Enterprise revenue through trust certification

• How to Win: Speed up certification from quarters and years to weeks and months, and from point-in-time to continuous trust certification

By using the “What Would Have to Be True?” lens, for Vanta’s strategy to be a good idea, it would have to be true that:

1. Smaller startups are indeed looking to move upmarket and capture greater revenue from Enterprise customers

2. These startups can effectively serve Enterprise customers once they’ve been certified

3. Vanta unlocks sufficient Enterprise revenue to justify the investment with a reasonable return

We’ve now designed three conditions that have to be true for Vanta’s matched set to be a solid strategy.

They aren’t beliefs or hopes, but assumptions that have to be tested to understand whether they’ll prove true or not.

Now we’ll dig into how the money side might play out.

Translating Strategy Logic Into Money Logic

Our third WWHTBT condition allows us to frame our strategy story in financial terms, crucial if we’re to have a meaningful conversation with executive stakeholders.

Granola’s Condition To Test

If Granola invests in Vanta to achieve SOC 2 compliance, roughly how much new Enterprise Annual Recurring Revenue (“ARR”) might that make possible in the next year? More importantly, can it do that with a reasonable return?

We’re not going to spend hours doing deep-dive accounting or fabricating numbers.

Following Rich Mironov’s “Money Story” concept, we pick three numbers and only use multiplication so that we can “count the digits” to see if this is a 4-, 5-, 6-, or 7-digit story.

Using Granola’s WWHTBT condition for the ROI potential linked to gaining SOC 2 compliance, we’ll use Rich’s “Boosting Volume Money Story” pattern to get our three numbers in five minutes:

1. Stalled Enterprise Deals Addressable Base: The approximate number of Enterprise deals that stalled or went dead specifically because Granola lacked SOC 2 compliance.

2. Economic Average Contract Value: The Average Contract Value (ACV) of each Enterprise contract.

3. Guesstimated Win-Back Range: A realistic range of how many of those dead deals the sales team can re-engage and close now that the compliance blocker is gone.

[Stalled Enterprise Deals] x [Enterprise ACV] x [Guesstimated Win-Back Rate] = [Win-Back Revenue]

Applying Example Numbers

Inside of 5-10 minutes, we can assume the following internal numbers for Granola:

1. Stalled Enterprise Deals Addressable Base: Roughly 40 Enterprise deals went dead in the pipeline over the last year because they required SOC 2.

2. Economic Average Contract Value: The average Enterprise contract value is $50,000/year. (You don’t need to be precise here; just “mid-5-digits” will do.)

3. Guesstimated Win-Back Range: Granola’s sales team estimates they can realistically win back 10% to 20% of those dead deals.

The Calculation: 40 lost deals x $50,000 ACV x 10% to 20% win-back rate = $200,000 to $400,000 in net-new annual revenue.

You now have the formula and rough numbers to inform your SOC 2 compliance strategy story.

But raw math alone won’t persuade executives. In the paid section, I’ll show you how Rich Mironov recommends translating these numbers into a “Money Story” product teams can use to “finish their money sentences,” turning these three numbers into a clear narrative you can share in your next high-stakes leadership meeting.

We’ll also go deeper into “What Would Have to Be True?” and walk through two additional tests, the “Opposite” and “Can’t/Won’t,” to help you understand whether you actually have a strategy and whether it creates a durable competitive advantage or just sounds good.

Product Team Roadmap Discomfort Audit

Pull up your roadmap for this quarter. Ask these three questions:

1. Consistency: For each feature, have you stayed consistent with a matched Where to Play and How to Win set? What is it?

2. Logic: For your matched set, have you made at least three WWHTBT conditions clear?

3. The Math: Have you taken at least one of your WWHTBT conditions and defined the 3 numbers and the rough multiplication that must be true for this feature to pay for itself?

If you can’t answer these questions, you don’t have a cohesive, logical strategy guiding your work.

You’re burning cash based on hope.

Paid Subscribers Get So Much More

Free subscribers now have a solid, working knowledge of how to ask Roger L. Martin’s most important strategy question, “What Would Have to Be True?” With the introduction of Rich Mironov’s “Money Story” formula, you can now translate your strategy logic into the kind of bottom-line business impact logic product teams need to understand.

But paid subscribers also get:

• The Executive Pitch Paragraph: How Rich Mironov recommends taking our financial calculation and “finishing the money sentence” so your CEO and CFO engage on what matters most to them

• The Opposite Test: How to test whether your matched Where to Play and How to Win pair is making a real strategy choice for the right customer

• The Can’t/Won’t Test: How to refine your matched set to identify and build structural barriers that prevent your biggest competitors from copying you

• A 15-minute exercise that produces three artifacts ready for your next leadership meeting

• Three AI prompts that run in the same logic sequence as the exercise and guide you to deepen your understanding of the material

Reply and let me know what resonated most for your product team’s situation.

Working with Product Teams

I’m looking for a few Product Trio leaders (PM, UX, and Tech leads) on active product teams who want to better understand strategy and financial literacy and use them to shape their day-to-day work and entire roadmap.

You’ll walk away from our free 30-minute session with a clear picture of your team’s actual strategy, the first question you need to answer to test it, and the 5-minute financial calculation you can run to show your stakeholders you care about more than just doing product “right.”

Reply TRUE and let me know a bit about you and your team, and if you’re ready to start seeing your product team’s strategic and financial impact in a new way.

Read more

I once worked with a VP of Product who refused to make any strategic choices.

Not. One. Single. Choice.

We were deep into discovery for a big product launch. The product team asked the right questions. “Do you want to sell more product, or sell to more people?” “Do you want to focus on promoting Product A or Product B?”

His answer?

“All of it. I want it all. I want you to give me everything.”

And this was the kicker:

“I don’t see any reason why I should have to choose.”

We were stunned.

Making choices is the foundation of strategic leadership.

Four months later, we released a jumble of features. Three months after that, usage was nearly zero. A month later, the executive was gone.

The party line? The team didn’t “execute.”

But this is laughable. The team’s execution was the only thing great about the product. Experienced design, strong product management, brilliant tech lead. We had a slick interface, tested flows, automated testing, and a complete CI/CD pipeline.

But with a jumble of half-baked directions, they couldn’t optimize for anything. No clear value proposition. No differentiation. No reason for any specific customer to care.

This is what happens when you refuse to make two key choices about your customers and your product that don’t work together.

The leader thought he was being ambitious. He was afraid to choose.

Making strategic choices requires more taking away than adding.

The product teams delivering results aren’t shipping the most features. They’re making two key choices that work together.

But it’s how they make them, and when, that makes the difference.

Subscribe now

Matching Inseparable Pairs

The core of your strategy comes down to two choices:

Where to Play: Which customers do you serve, to address which problems, in which contexts?

How to Win: What’s your differentiated approach to serving them?

The biggest mistake I see people make when learning strategy using the Strategy Choice Cascade is treating Where to Play and How to Win as two sequential decisions:

1. First, brainstorm potential markets or customer segments (Where to Play)

2. Prioritize that list at a workshop

3. Then, try to figure out a list of “How to Win” choices for each one

This feels logical. It’s how we’ve been taught to break down complex problems.

But it takes you deep into rabbit holes you should never go down in the first place.

Where to Play and How to Win are an inseparably matched pair. They’re a single decision expressed in two parts, not two decisions made separately.

Here’s why the sequential approach fails:

Without deeply thinking about your “How to Win,” almost any “Where to Play” looks interesting. You end up losing time thinking about long lists of potential customers, markets, and channels. “Enterprise customers.” “SMB teams.” “Developers.” “Non-technical users.” The list grows endlessly because there’s no filter.

These lists spread your strategic focus. Your team can’t focus on possibilities that might actually work because most of the list is purely theoretical. A Where to Play without a paired How to Win is just wishful thinking.

And when you finally turn to How to Win, you’re stuck. You go through each Where to Play trying to find a winning approach and come up with nothing. For most of them, there isn’t one. At least not for your team, given your capabilities and constraints.

So if you get stuck on a particular “How to Win,” the problem is almost always your “Where to Play.” You picked a market where you can’t actually win. The fix isn’t to brainstorm more “How to Win” choices.

The fix is to go back and adjust your “Where to Play” until a natural “How to Win” emerges.

The better approach is paired and iterative, not sequential.

When a potential “Where to Play” comes to mind, immediately ask: is there a “How to Win” that would make this a winning combination?

If yes, you might be onto something. Keep exploring this pair.

If not, ask: could this “Where to Play” be tweaked to enable a stronger “How to Win”?

If you can tweak it, try the new version. Keep iterating between the two choices.

If you still can’t find a viable pairing, discard that “Where to Play” entirely and move on. Don’t add it to a list. Don’t save it for later. It’s not a viable option for your team.

Never make lists of potential Where to Plays.

The moment you start keeping lists, you’re back to sequential thinking. The back-and-forth between Where to Play ideas and potential How to Win matches is what creates something real and powerful at the heart of any strategy: the matched pair.

A Where to Play is worthless without a paired How to Win. “We’ll serve enterprise financial services customers” isn’t a strategy until you know HOW you’ll win with those customers in ways your competitors can’t copy.

And a How to Win only makes sense for a specific Where to Play. Vanta’s “continuous trust” only works because its Where to Play (U.S. startups seeking enterprise deals) enables it. Expand that Where to Play, and the How to Win no longer has the same strength.

Key Principle: Effective strategy choice-making involves designing matched pairs where:

• The Where to Play makes the How to Win especially valuable

• The How to Win would be wrong for other customer segments

• Each choice strengthens the other

Once you’ve seen this strategy First Principle, you can’t “unsee” it, and your entire approach to strategy will shift.

How Gamma Went From Near-Death to $100M ARR

In 2022, Gamma had 60,000 users after three years. They were running out of money. Their strategy:

• Where to Play: People who make presentations

• How to Win: Better PowerPoint formatting

Apply the Match Test. Does “better formatting” ONLY work for “people who make presentations”?

No. Better formatting could work for anyone making any document. It’s generic and doesn’t differentiate in any important way.

They pivoted. Two connected choices:

Where to Play: Knowledge workers who need to rapidly

communicate visually but lack design skills and time

How to Win: AI-powered creation that transforms 10-word

prompts into polished presentations in 30 seconds

Match Test: Does “AI makes fast design decisions for you” ONLY work for “people who lack design skills and time”?

Spot-on.

The professional designers I know and work with daily have zero interest in AI making their design choices for them. They’d hate this product. People with time prefer to start from a sketch, carefully think through the structure, design, iterate, and nudge bullets and images, preferring to put their own stamp on the presentation. They’d find Gamma too limiting and unhelpful.

But knowledge workers without design skills, drowning in deadlines, who need quick, powerful presentations to communicate ideas? Gamma gives them exactly what they need. The intentional limitation becomes a key differentiating choice.

Matched set.

What Gamma fixed:

Its original strategy failed on two counts:

• “People who make presentations” wasn’t a real Where to Play choice. Everyone makes presentations. The opposite (”people who don’t make presentations”) isn’t a viable alternative market.

• “Better formatting” wasn’t a real How to Win choice. The opposite (”worse formatting”) is stupid. No one chooses that.

Their pivot created real choices that reinforce each other:

• “Non-designers without time” is a real choice. Designers and people with time are viable alternative markets that competitors serve.

• “AI decides for you” is a real choice. Full user control is what Figma and Canva offer.

• AND these choices fit tightly: the Where to Play creates customers who see the How to Win as a feature, not a limitation.

Result: 60,000 users → 70 million users. $100M ARR. 40-person team.

Two real choices. Tight fit between them. That’s a matched set.

Read Gamma’s full strategy breakdown for free here:

The Matched Set Principle

Only part of Gamma’s success has to do with timing.

While many AI presentation tools launched in 2022-2023, Gamma still leads the category as the rest struggle to retain and grow a meaningful user base.

Despite pressure from Microsoft, Google, Canva, and Beautiful.ai, Gamma dominates because it understands that Where to Play and How to Win aren’t two separate decisions. They’re one decision expressed in two parts.

When they reinforce each other tightly, you create competitive advantage. Competitors can’t copy your How to Win without also copying your Where to Play. And if they’re already serving a different Where to Play, your How to Win would be wrong for their customers.

Moderate fit lets you play. Tight fit lets you win.

When the two choices don’t reinforce each other, you’re reduced to competing on features alone. Anyone can copy you. You’re competing on execution speed and marketing budget, not strategy.

What the Match Test Can’t Do

The Match Test can quickly reveal that something fundamental about your choices needs fixing.

It can’t help you design a strategy. But it can help you and your team start thinking more strategically.

If you don’t have a clear Where to Play statement, you can’t test it. If your How to Win is “we’re better,” there’s nothing with which you can match it.

The paid section below gives you the exact exercises, tools, & steps to fix this.

WHAT’S BEHIND THE PAYWALL

This week’s paid section includes:

📋 GOING DEEPER INTO THE STRATEGY CHOICE CASCADE We’ll dig deep into the complete 5-question framework. How each choice constrains and enables the next. And why it’s so important to start and match in the middle, not at the top.

⏱️ THE 15-MINUTE “MATCH TEST” EXERCISE A step-by-step exercise to draft your Where to Play and How to Win, apply the Match Test, and diagnose any problems. Designed for solo work or with your product trio.

📊 THE DIAGNOSTIC FRAMEWORK When the Match Test fails, you need to know WHY. Is it a choice problem (binary) or a fit problem (degree)? Different root causes require different fixes.

🤖 AI-POWERED STRATEGY ANALYSIS (NEW) Two prompts designed specifically for this exercise. Use them with Claude, ChatGPT, or Gemini to design and pressure-test your choices. They’re built to ask the hard questions and present options you may not have considered.

You just learned the Match Test. Paid subscribers get the complete framework to apply it.

This is Chapter 1 from “Playing to Win for Product Teams.” I’m writing this book for product trios, and your feedback shapes what makes it into the final version.

Remember that RICE score of 1,200 for a feature nobody wanted?

Paid subscribers get an exercise and prompts to sharpen two key strategic choices that most teams never think about.

Get them right, and you’ll have a lens to shorten every roadmap debate to a 5-minute conversation.

Read more

Starting now, this newsletter is about the tough questions product teams face daily, and how strategy can help you answer them.

If your team is facing these questions, read to the end.

Think about the last time you celebrated a feature launch, only to realize four months later that it moved zero business metrics. How could this “high-priority” item that scored so well result in crickets?

RICE promised objectivity. Instead, it delivered feature bloat and the death of a promising product.

All prioritization frameworks share the same fundamental flaw. They exist because there’s a strategy gap at the top.

I learned this the hard way.

Playing prioritization roulette

Our team was scrappy and laser-focused on a small set of clear, user-centered choices, discovering, designing, and delivering a successful product launch.

But with that initial success, stakeholders from across the organization suddenly descended, each with a feature request they wanted shoehorned into our product.

To rate all these requests on some form of objective, level playing field, we turned to Intercom’s RICE framework. Reach, Impact, Confidence, Effort. Plug in the numbers, get a score, and build in order.

We hoped the fair, objective process behind our math would give stakeholders confidence in our roadmap choices.

Subscribe now

The only thing we successfully prioritized was feature bloat

While we were busy scoring, triaging, and building out the flood of requests, our product began to change.

Over-engineered features our users had to be trained on never got used. Meanwhile, real customer pain points went unaddressed. And still, we kept stuffing more features in, trying in vain to satisfy every stakeholder’s insatiable demand.

Our technical, UX, and strategic debt mounted until our product collapsed under its own weight.

Leadership eventually pulled the plug.

What really happened?

We simply weren’t prepared for the flood of feature requests that came with our initial success.

We thought RICE would help us pick the right thing to build next and help us create the best collection of features.

That was the mistake.

As soon as we stopped focusing on our strategy and started playing the feature prioritization game, we were asking the wrong question.

Weaponizing priority

RICE promises to help product managers and their teams have richer backlog conversations with stakeholders. In practice, I’ve found it does the opposite.

Product teams hold up their score as “proof” they’ve done their due diligence. More often than not, the Highest Paid Person’s Opinion (“HiPPO”) decides.

The framework ultimately becomes a political weapon, as “Reach” and “Impact” scores are debated.

But no one ends up winning. Least of all, customers.

The layer problem

Here’s how product teams really create value:

Prioritization frameworks fail because they operate at the wrong level.

Product teams actually create value across four layers:

At the top: Business Outcomes. Revenue, retention, profit. The lagging indicators leadership cares about.

Below that: User/Product Outcomes. Leading changes in user behavior you can actually influence that ladder up to lagging actions they take in your product.

Below that: Customer Opportunities. Unmet needs expressed in your customers’ own words. “I need to see all my work in one place.” “I want to stay ahead of problems.”

At the bottom: Feature Outputs. The completed forms, pages, flows, and buttons you build.

RICE lives purely at the bottom layer. It can tell you which feature scores highest based on the unscientific, wild guesses you make from among the myriad options in your backlog and roadmap. It is blind to whether any of those features address an opportunity your customers actually have.

You can score a perfect 1,200 on a feature that solves a problem nobody cares about.

What great teams do instead

Great product teams start by making clear strategic choices about which opportunities to pursue for their chosen customers, and how they will solve for them.

This means asking different questions:

For which customers are we building this, so they can finally do what?

Which unmet need does this address?

How does this connect to our higher-level choices about where we play and how we win?

And, most importantly:

Which customers and opportunities do we choose not to address at this time?

When you’re working against a clear set of choices, what to build next becomes obvious.

You’re not bickering between competing features based on some very subjective and suspect math.

You’re focusing your team’s limited capacity on solving a specific problem for a specific customer in a way that advances your company’s overarching strategy.

Not What Is True. What Would Have to Be True?

Roger Martin calls this shift “What Would Have To Be True?”

Instead of debating which feature is right, you ask:

What would have to be true about our customers for this opportunity to matter?

Under what conditions would our capabilities be able to deliver against it?

Only by making this shift can you stop having opinion battles and start surfacing and testing logic and hidden assumptions.

Solving the real problem

81% of product managers believe strategy is core to their role. But only 25% work at companies that provide the clear strategic direction they need to lead their product teams.

That gap is where the pain lives. And no scoring formula will ever close it.

But you don’t have to wait for leadership to fix this. You can reverse-engineer your current strategy, surface the choices already being made, and design better ones at your level.

My Offer to You

I’m looking for a few Product Trios (PM, UX, and Tech leads) who are drowning in feature requests without strategic clarity.

You’ll walk away from our free 30-minute session with a clear picture of your team’s actual strategy and the first question you need to answer to fix it.

Reply CLARITY if that’s you.

Share

Read time: 3.7 mins.

Four companies laid the foundation for agentic AI’s continued evolution.

While many still chase better benchmarks and linear interactions, Google, Anthropic, Microsoft, and Cursor are investing for a future that’s already here.

They’ve built infrastructure for agents that work and improve without human oversight.

Subscribe now

1. Google’s Gemini 3 Plans Consistently Across Full Simulated Years

Google launched Gemini 3 on November 18, 2025, scoring 37.4% on Humanity’s Last Exam, triggering OpenAI’s Sam Altman to declare “Code Red” forcing OpenAI to roll out its next 5.2 model ahead of schedule, beginning on December 11, 2025.

Google saw “context rot” as an existential threat. Understanding agents would need extended time-based reasoning, Google built Gemini 3 to maintain consistent decision-making over a full simulated year.

Gemini 3’s “Where To Play” targets three customer segments: enterprises via Vertex AI; 650 million monthly active consumers via the Gemini app; and individual developers via Google AI Studio and Gemini CLI, and the new Antigravity platform.

Can’t/Won’t: Most customer-focused, model- and consumer-only vendors wouldn’t invest in long-horizon planning infrastructure when the market only rewards benchmark and chat interface improvements.

2. Anthropic’s Claude Opus 4.5 Model Achieves Performance While Slashing Cost

Anthropic released Claude Opus 4.5 on November 24, 2025, scoring 80.9% on SWE-bench Verified while cutting prices by nearly 70%. Astoundingly, Opus features self-improving agents that achieve peak performance in just four iterations compared with competitor models that require 10.

Claude Opus 4.5 targets developers and enterprise teams through APIs and high-profile integrations with Snowflake, GitHub Copilot, Cursor, and Lovable. Opus couples its improved performance and pricing with massively increased token efficiency, requiring 48-76% fewer output tokens than competitors.

Anthropic’s business-focused, vs. consumer-focused “Where to Play” has made all the difference, as CEO Dario Amodei said his company can just keep “growing and developing our models” without the need to “declare any emergencies.”

“You just focus on different things, less on engagement, more on coding, high intellectual activities, and scientific ability.”

Can’t/Won’t: Competitors will be hard-pressed to match Anthropic’s advances in autonomous training or pricing structures while simultaneously improving efficiency and dropping prices by 67%.

3. Microsoft Ships Agent Governance Before the 1.3 Billion Agent Explosion

Microsoft announced Agent 365 at Ignite in November, its platform for managing what IDC projects will be 1.3 billion agents globally by 2028.

Microsoft is avoiding competing at the model or interface level to focus on the agentic AI control plane. Agent 365’s “How to Win” extends Microsoft’s dominance in enterprise identity with Entra Agent ID, security with its AI-powered threat detector Defender, and automated compliance and reporting via its Purview tool to manage agents.

Microsoft is betting enterprises will need an enterprise-grade agentic AI management layer to own governance of all other agents, regardless of origin. The security and reputational risks to any enterprise that does not effectively manage its agents could be catastrophic.

Can’t/Won’t: Competitors can’t offer Microsoft’s unified identity layers across diverse agent ecosystems, or invest in expensive governance infrastructure before enterprise demand materializes.

4. Cursor’s $29.3B Valuation Proves Autonomous Coding at Enterprise Scale

Cursor secured $2.3 billion in Series D funding in November, tripling its valuation from $10 billion to $29.3 billion in just five months. Cursor achieved that valuation by reaching $1 billion in ARR while spiking enterprise revenue growth by 100x.

Ironically, Cursor’s success stems from focusing on a niche AI application: developer productivity. Its 2.0 multi-agent architecture enables engineers to run up to 8 agents in parallel, reducing development timelines from linear to parallel execution.

As we saw with Gong training its models on sales calls, Cursor trained its proprietary Composer model using Reinforcement Learning across hundreds of thousands of concurrent sandboxed cloud-based coding workspaces, achieving a 4x speed advantage over general-purpose models.

Can’t/Won’t: By training its proprietary Composer model across the complete development cycle, learning to generate, integrate, test, and improve code within live workflows, Cursor has achieved a level of continuous agent-driven collaboration that GitHub Copilot and Replit’s Agent can’t deliver.

📊 THIS WEEK’S SIGNAL

These four companies are building agents for workflow reimagination and increasingly autonomous operation.

Google’s betting on long-horizon autonomy. Anthropic’s betting self-improving agents reach solutions faster. Microsoft’s betting mass agentic AI deployment needs governance infrastructure. And Cursor’s betting enterprises and individuals will adopt parallel agentic coding.

All four are building the independent agentic AI infrastructure of the future.

For Your Product Strategy

• Which of your features could evolve from assisted tools to autonomous agents?

• Where are you optimizing for supervised, linear interaction when autonomous and parallel workflows could unlock more value?

• What governance infrastructure will be necessary before demand materializes?

• What aspects of your product could improve autonomously without your team’s intervention?

📖 REVISED, EXPANDED & UPDATED: The Lovable Strategy Breakdown

This week, I went back and completely revised, updated, and expanded my reverse-engineered deep-dive of vibe-coding platform Lovable’s product strategy.

This version features Sabrine Matos, a woman who’s literally saving lives with Lovable. I also break down three new strategic moves that most AI product companies are still missing, and how tie-ins with Gemini 3’s Nano Banana for image generation and Claude’s latest Opus 4.5 model are delivering vastly faster, higher-quality results.

Read the complete revised strategy breakdown →

Join me next Friday for the next set of strategic moves!

Share

Starting this week, we’re pivoting from weekly deep dives to shorter, bite-sized vignettes that break down the strategic moves behind the headlines.

Every Friday, I break down the decisive choices of leading companies and show you how to start seeing those patterns, learning from them, and applying them to your own products.

This week’s theme: Developing first-mover AI capabilities.

Time to read: 4 minutes

1. Adobe Bets $1.9B on Generative Engine Optimization

Adobe paid a hefty 78% premium to acquire Semrush for $1.9 billion in November, betting that AI LLMs (ChatGPT, Claude) will increasingly displace Google as the web’s front door.

When the acquisition closes, Adobe plans to integrate Semrush with both its Adobe Experience Manager (AEM) and Brand Concierge products, enabling marketers to optimize for both Google’s algorithms (SEO) and AI discoverability (GEO) in a single workflow. That 78% premium signals the existential urgency it’s feeling to keep its products relevant.

Can’t/Won’t: Competing marketing platforms will be hard-pressed to rapidly build and deliver unified SEO/GEO without making similar acquisitions at this scale. It’s unlikely creative platforms like Figma or Lovable will embrace discoverability optimization as a core platform strategy.

Potential Risk/”What Would Have to Be True?”: Acquisitions are always risky. Always. The odds are against Adobe’s ability to integrate and realize the value it’s invested. Counterintuitively, Adobe’s better acquisition play would be to understand what value it can bring to Semrush, not the other way around.

2. Target Treats ChatGPT as a Sales Channel

Target launched a full, end-to-end ChatGPT checkout in November 2025, allowing customers to complete purchases from discovery through payment directly within the AI chat interface.

All purchase intent starts with a question, but when customers complete checkout from inside ChatGPT, AI moves from passive billboard to active storefront, opening another potential revenue stream. It remains to be seen whether “conversational commerce” becomes a viable retail channel with meaningful revenue potential or just a minor, secondary sub-channel under web and mobile.

Can’t/Won’t: Amazon can’t build on OpenAI’s platform. Traditional retailers won’t redesign their checkout flows for conversational interfaces. Target’s moving first while others wait for proof before committing to develop the necessary commerce capabilities.

3. Notion Unlocks Private Slack Channels for Organizational Context

Notion released version 3.1 in November, upgrading its Slack connector to access private channels and DMs while respecting all existing permission levels.

The unlock: Most workplace AI tools are blind to private conversations. Getting into territory previously owned by Otter, Notion is giving enterprise employees access to broader AI context while directly addressing enterprise security data exposure concerns.

Can’t/Won’t: Slack AI won’t natively surface insights beyond parent Salesforce. By building permission-aware private channel access, Notion is positioning itself as an alternative knowledge layer that synthesizes all relevant workplace communications, going one step deeper than public artifacts.

4. DocuSign Opens Agreement Intelligence to AI Agents

DocuSign released its Intelligent Agreement Management (IAM) platform via Anthropic’s Model Context Protocol (MCP) in October 2025, enabling Claude, Copilot, and other agents to access agreement workflows directly via natural language.

As AI agents proliferate, whoever enables access to authoritative business data (agreements, CRM, HR) captures disproportionate value. Extending its deep agreement expertise into agentic AI, DocuSign seeks to own the source-of-truth workflows involving contract context.

Can’t/Won’t: Competitors without industry-standard MCP infrastructure can’t join the agentic ecosystem. Legal teams will be unlikely to trust custom integrations. DocuSign is capturing an early advantage over competing agreement providers before the contract trust market becomes fragmented.

5. Vijil Raises $17M to Solve AI Agent Trust

Vijil, named a Gartner Cool Vendor in Agentic AI Security, raised $17 million to accelerate its agentic AI trust infrastructure platform.

The bottleneck: enterprises lack the confidence that the agents they’ve piloted will behave reliably in production. Vijil’s platform covers the full agent lifecycle, from build, test, deploy, govern, and continuously improve via reinforcement learning using live production data. SmartRecruiters cut its time to ship agents from 6 months to 6 weeks using Vijil.

Can’t/Won’t: Neither observability vendors (Datadog, etc.) nor generic security tools have a shot at building agent-specific lifecycle management or continuous reinforcement learning. Vijil occupies a tiny but essential niche to roll out and ensure AI agents perform reliably in mission-critical environments.

📊 THIS WEEK’S SIGNAL

Five companies are making similar strategic choices to move early in AI capabilities now, before waiting for enough data to exist to tell them they needed to.

By that time, it would already be too late.

Adobe’s betting that GEO will become as critical as SEO. Target’s betting that conversational commerce is real and that it can capture significant value through the channel. Notion’s betting on grabbing a bigger slice of the enterprise market with comprehensive context while respecting permissions. DocuSign is betting on being first to own agreement signatures through agentic ecosystems. And Vijil’s betting that everyone, from startups to enterprises, will increasingly reimagine every part of their business workflows around AI capabilities with AI agents at their core.

These moves redefine what “table-stakes” capabilities look like in the age of agentic AI.

For Your Product Strategy

• What new channels could AI be making available to your product?

• In what ways could you use agentic AI and MCP to completely reimagine legacy human processes, rather than just accelerate them?

• What features or pilots could you be running today that could scale to become infrastructure tomorrow?

• What tradeoffs is your industry accepting that you could attack directly? (Not either/or, but and?)

• Who’s making moves that look improbable, too early, or too expensive until you see what they see?

That’s it for this week. Next Friday: another set of strategic moves and the patterns behind them you need to understand.

— Mike

Share

Zeal CTO Pranab Krishnan spent years writing paragraphs of text trying to communicate UX ideas to his product team.

The words never quite conveyed what he was trying to explain, and that friction only increased as Zeal scaled from two founders to a full product and design team.

Then Krishnan discovered Magic Patterns. Forget text. He could now screenshot the current product, prompt the design, and iterate right there.

“Now I can build what’s in my head, even though I’m not a designer. Magic Patterns makes it easy to show, not tell.”

Magic Patterns’ ability to transform Zeal’s product development cycle represents a shift “left” for greater collaboration earlier in the idea-to-software-delivery flow.

This shift has allowed founders Alexander Danilowicz and Teddy Ni to reach $1M Annual Recurring Revenue (“ARR”) with zero employees and explains why customers gladly pay a premium for their “vibe designing” platform.

Subscribe now

From “Silliest Idea” to Series A in 30 Months

That’s the blunt assessment Y Combinator partners offered Alexander Danilowicz and Teddy Ni’s text-message analytics startup idea when they applied to Y Combinator in early 2023.

Both founders had spent years building and maintaining design systems at Canopy, LiveRamp, Robinhood, and Instagram. They’d lived the pain of converting Figma mockups to production code while maintaining brand consistency across dozens of components.

The Y Combinator partners suggested they build something leveraging their shared expertise instead.

Over the three-month winter cohort, they iterated through half a dozen product ideas in the design tooling space. A Chrome extension for cloning websites. A component library editor. A design-token manager. Each gained traction but never quite stuck.

What looked like repeated failures turned out to be foundational R&D. Every pivot led back to the same problem: product teams needed consistency across designs, components mattered more than code, and the design-to-development handoff was a pain point for every team.

Their breakthrough crystallized during launch. Instead of replacing Figma or becoming a full-stack builder, they’d solve the specific problem product teams faced: translating design intent into production-ready UI code while respecting existing design systems.

Stealth to $1M ARR

Within two years, Magic Patterns grew profitably to $1M ARR with just two founders and zero employees, achieving what the startup industry considers the gold standard of Series A readiness.

The founders’ “failed experiments” formed the foundation of its competitive advantage: teams could import their exact component libraries via the Chrome extension, prompt feature ideas in natural language, and receive interactive prototypes that precisely matched their brand’s colors, typography, and existing components.

This design system integration became their differentiating competitive advantage. Competitors like v0, Lovable, and Figma Make generate impressive interfaces, but none maintain consistency with real product branding the way Magic Patterns can.

By November 2025, over 1,500 product teams at companies from Y Combinator startups to enterprises were using Magic Patterns daily, collapsing weeks of design iteration into hours.

The Cost of Focus

The founders maintained disciplined “Where to Play” choices: frontend-only, design systems-first, and product teams only. No backend. No generic custom code. No agencies or solopreneurs, though both segments found value.

That focus came at a cost. Both agencies and solopreneurs wanted Magic Patterns, and full-stack builders wanted backend support. The founders said no to all of them, repeatedly, knowing that product teams with design systems offered expansion economics that one-off projects never would.

Every “no” sharpened their differentiation.

Venture-Backed Payday

On November 10, 2025, co-founder Alexander Danilowicz announced Magic Patterns’ Series A for $6M, led by Standard Capital with participation from Y Combinator and other investors. Unusual for an AI startup, the founders had achieved profitability and product-market fit before raising capital, maintaining the discipline that generated $500K ARR per employee.

The Series A capital funds more of the same: frontend engineers, sales talent, and product specialists to expand design system integration while launching Magic Patterns 2.0 with Agent Mode and deeper integration with Figma’s Model Context Protocol (“MCP”).

Magic Patterns’ growth validates its strategic choices: By clearly saying no to what it wasn’t, it’s become irreplaceable at what it is. The design system-aware AI prototyping platform for product teams.

We’ll break down its choices across the Playing to Win cascade to see exactly how Magic Patterns built its competitive differentiation.

Reverse-Engineering Magic Patterns’ Strategy: The Organizational Capability Flywheel

Magic Patterns’ strategy kick-starts a flywheel most AI design tools miss: design system integration enables organizational capability expansion, which drives PLG expansion, which validates focused “Where to Play” choices, which deepens competitive differentiation.

Winning Aspiration: Enable Product Teams to Expand Design Capability Without Expanding Headcount

Magic Patterns aspires to become the best and fastest AI-native design platform product teams use to expand their design capability, reaching 10,000+ organizations by enabling non-designers to ship production-ready features.

Making designers faster is one benefit. Making CTOs into designers, product managers into prototypers, and engineers into design contributors is the transformation.

Already profitable, Magic Patterns aims to scale from $1M to $10M+ ARR while maintaining its reputation as the tool “loved by product managers, designers, and website builders.”

Success means shifting design from managing rectangles to prompting features to “magically” appear, in perfect alignment with existing styles.

Where to Play: Recurring Product Teams With Predictable Expansion Patterns

Customer: Product teams at software companies (5-50 people) who maintain design systems and ship features weekly. Not agencies, consultants, or project-based work where expansion economics don’t apply. Three personas: Product Managers visualizing requirements, Designers prototyping with real code, and non-technical founders building MVPs. These teams naturally expand from 2 licenses to 5, then 20, and eventually 50+ as the tool proves its value.

Channel: Pure product-led growth. Users discover through organic search, YC network word of mouth, Product Hunt, Reddit, and direct sign-ups with a free tier. No sales intermediaries for smaller teams. Enterprise account executives engage once accounts organically reach 20+ seats.

Offer: An AI-powered design platform generating production-ready UI code from text prompts, screenshots, or Figma imports. Unlike competitors generating static mockups or generic code, Magic Patterns produces UI screens matching existing design systems, interactive prototypes with working states, and exportable code in React, Vue, and Tailwind CSS.

No full-stack. No backend. As co-founder Alex Danilowicz explains, “99% of the time you really don’t need a database” for prototyping.

Geography: Global, digital-first. Enterprise compliance (SOC 2 Type II, ISO 27001) enables customers like PwC, Vapi, and Lendi Group.

Stages of Production: Magic Patterns controls the full design-to-code workflow for frontend interfaces: AI model orchestration with proprietary pre- and post-processing to reduce hallucinations, design system integration and component library management, prototype hosting, and real-time collaboration infrastructure.

Partners: Anthropic for AI models, AWS for infrastructure, and third-party auditor Delve for continuous security compliance.

Left to customers: backend development, production deployment beyond prototypes, and final design decisions. Magic Patterns is the supporting tool for teams executing their visions.

How to Win: Differentiation Through Design System Integration

Magic Patterns wins through differentiation: the only AI design tool that truly understands and preserves existing design systems.

While competitors like v0, Bolt, Lovable, and Figma Make generate impressive UIs from prompts, they produce generic, one-off designs that don’t match the product’s true look and feel.

Magic Patterns differentiates by letting users import existing component libraries via Chrome extension from Storybook, websites, or Figma, then generating screens that automatically use those exact components, colors, and styles.

As co-founder Teddy Ni emphasizes:

“Before we wrote the first line of code for Magic Patterns, we had the vision of it using your component library.”

This design system fidelity transforms commoditized UI code into a domain-specific strength, expanding design capability across the org.

When Zeal’s CTO can think like a designer, when Waydev’s engineers can prototype without waiting for design, when Lendi cuts delivery time from 12 weeks to 4, Magic Patterns expands what product teams can do without adding headcount.

By saving 50%+ of time from idea to launch, enabling non-designers to create professional UIs, and producing clean production-ready code, customers willingly pay premium pricing ($15-60/month per user).

Magic Patterns’ competitive advantage rests on deep domain knowledge and technical infrastructure that took two frontend specialists years to develop.

Network effects emerge as users share component libraries and templates. Magic Patterns’ first-mover advantage in AI-driven design creates brand recognition and boosts PLG.

Integrating Sonnet 4.5 within 24 hours of Anthropic’s release demonstrates technical excellence that competitors struggle to match.

Capabilities: Three Distinctive Activities That Enable Organizational Capability Expansion

Magic Patterns has built three core capabilities that deliver its differentiation:

1. Design System Integration Engineering

Magic Patterns’ Chrome extension captures HTML and CSS from any website, parses component libraries for reuse across projects, and enforces brand consistency. Bi-directional Figma integration preserves layer structures.

2. AI Model Orchestration for Production-Ready Code Quality

Magic Patterns boasts sophisticated multi-step pipelines running Claude models through pre- and post-processing: parallel execution across AWS Bedrock and direct Anthropic APIs, AST parsing to validate code structure, deterministic post-processing to strip hallucinations, and context retrieval to inject design system context into prompts.

Unlike Bolt that optimizes for speed, or Lovable that optimizes for full-stack applications, Magic Patterns chooses to optimize for production-ready frontend code developers can immediately use.

3. Collaborative Iteration Infrastructure

Infinite canvas with multiplayer capabilities rivaling Figma. Sub-second synchronization across distributed teams. Version history supporting 700+ iterations per project. Permission management for view and edit sharing. Embedded feedback collection with custom questions.

This enables product teams to iterate through the “vibe designing” process, maintaining context and collaboratively evolving designs without starting from scratch.

Management Systems: Infrastructure Supporting the Expansion Flywheel

Magic Patterns has implemented three distinctive management systems to support and continuously up-level its capabilities:

1. Credit-Based Expansion Metrics

Magic Patterns tracks credit-based usage analytics, activation metrics across free to Individual ($15/month) to Pro ($60/month) tiers, seat expansion patterns (2 to 5 to 20 to 50+), and quality metrics including error tracking and rendering success rates.

This identifies when accounts hit expansion thresholds, triggering tier upgrades or enterprise sales engagement.

2. Customer Feedback Loops With 8-Minute Response SLA

With two founders at $1M ARR, Magic Patterns maintains 8-minute average support response times. The founders personally respond to tickets, with feedback informing weekly product decisions. This closeness has created case studies with customers like Vapi and Pigeon, plus active community engagement on Reddit, LinkedIn, and Twitter.

After receiving this level of service and direct feedback into the roadmap, customers are reluctant to go elsewhere where their input may not be as valued.

3. PLG-to-Enterprise Handoff Triggers at 20+ Seats

Automated detection triggers enterprise sales engagement when accounts reach 20+ seats. Pure PLG for smaller teams, enterprise account executives for larger accounts. SSO integration, security documentation (SOC 2, ISO 27001), and EU data residency for GDPR compliance.

This hybrid captures both high-velocity small-team expansion and high-value enterprise contracts without forcing all customers through the same sales process.

Magic Patterns’ Workflow Reimagination & Capability Expansion Flywheel

Design system integration enables role expansion (CTOs become designers, PMs become prototypers).

Role expansion drives seat growth (2 to 5 to 20 to 50+). Seat growth validates the narrow “Where to Play” choice, since product teams offer expansion opportunities agencies don’t. Expansion economics sustain premium pricing, funding continued investment in design system integration. The flywheel reinforces itself.

Playing to Win discipline means saying no: no full-stack, no backend databases, and no pixel-perfect Figma replacement. These trade-offs create focus that enabled two people to profitably reach $1M ARR.

As Magic Patterns scales from $1M to $10M ARR, its strategic choices will be tested. Can it maintain design system integration quality while serving 10x more customers? Will it need to add backend capabilities as demand grows? Can PLG continue driving growth at enterprise scale, or will it need a full sales organization?

The answers will determine whether Magic Patterns becomes the standard AI design tool for product teams or one of many fighting for market share.

For now, it has built an integrated strategy with clear choices across all five boxes.

Strategic Takeaway: Build Capabilities That Reimagine and Expand Organizational Roles, Not Just Accelerate Workflow Velocity

Magic Patterns’ customer stories reveal something beyond design system integration: the deepest competitive advantage comes from reimagining workflows and shifting what organizational roles can accomplish, not just making existing work faster.

Magic Patterns measures success by organizational transformation metrics: product managers and engineers thinking and collaborating on design earlier. These metrics correlate to retention and expansion because they reflect how fundamentally the tool has changed workflows.

When your product enables non-specialists to do top-flight specialist work, you’ve expanded their capabilities, allowing you to charge for more than tool usage.

Magic Patterns charges $15-60/month per user because they’re selling “the design hires you don’t need to make.”

And Zeal’s CTO, Pranab Krishnan, no longer writes paragraphs of text anymore.

He thinks, and acts, like a designer.

Share

Chris Tottman is a partner at Notion.vc and has spent over 30 years building and investing in technology companies. He started his career trading options on the London Stock Exchange floor before founding multiple companies, including MessageLabs, which became Europe’s first major SaaS business before its acquisition by Symantec. Today, he invests in early-stage technology businesses while maintaining what he calls “a constant itch to create, collaborate, and work with people to help them build the best companies they can.”

Subscribe now

Key Takeaways

On Discovery vs. Proving Yourself Right Great founders don’t over-index on proving their original idea. They stay in constant discovery mode, asking what problems keep customers awake at night and what users are actually doing with the product.

On Finding Problems Worth Solving Look for pain that makes someone bang the desk. Pain that makes them think, “If I don’t solve this, I might get fired.” That’s where innovation and the wave of adoption comes from.

On Market Definition “Uncomfortably narrow” is the goal. The smaller the market definition, the easier it is to target. Wide definitions like “enterprise, SME, global” become impossible to market effectively.

On Official Strategy vs. Revealed Strategy Your official strategy is what you tell investors and put on your roadmap. Your revealed strategy is what your usage data shows customers actually value. The gap between those two is where billion-dollar pivots happen.

On the Discovery Process Whiteboard out the whole value chain of a product, looking at where time goes and where money goes. Run B2B and B2C research. Have multiple minds working on it, asking “What if this is true? Is that also true?” That’s how you find the biggest, most pressing problems worth solving.

The Difference Between Good Founders and Great Founders

Michael Goitein: Chris, we started talking about pivots and that notion that when a company starts, they might have a vision, an idea. But what they ultimately become successful for can be miles from what they started out as. You have these incredible stories that I wanted you to share.

Chris Tottman: I think the good thing about the stories is that they show two different ways of trying to figure out what the company can potentially become and how to bridge from where you are today to a bigger version of the world.

What we find at Notion is that the difference between good founders and great founders is that great founders don’t try to prove themselves right to the end of the world. They have an idea and take it to market, but they don’t spend their time trying to prove that original idea is the killer idea. Even if customers are buying it.

What they do instead is stay constantly in discovery mode. They keep asking questions about what problems keep customers awake at night. What are users actually doing? What are they really getting value from?

From ISP to Europe’s First Massive SaaS Business

Chris Tottman: In the early ‘90s, I was working with a couple of my closest friends who were building the IP networks. They had an ISP with a proposition for SMEs, connecting companies to the internet.

To give you context on how early this was: the first two-meg leased line we sold cost £200,000 a year in 1995. Two meg. Companies were struggling just to connect a local area network within their building to the outside world.

We built technology so somebody could just plug in a box that had a mail server, some security, and web space all bundled together. Super easy. It was flying off the shelves. We had a tremendously successful company.

But here’s the key lesson. We didn’t just sell to customers. We spent significant time in exploration mode, really asking deeper questions about what was keeping them awake at night. What’s stopping them from being successful? How do they lose against competitors?

Three or four things started to come up. One was more sophisticated technology around websites. Another was blacklisting certain sites. And third on the list, almost as an afterthought, customers said: “You keep delivering us viruses.”

Two things were happening. I could see the engineers, just through their body language, really itching to get to that third item. And it wasn’t the first thing on the list. It wasn’t the obvious thing.

One of the founders said, “Is this a little bit like clean water? Back before critical infrastructure, how did you get clean water? You boiled it yourself. But now that’s all done centrally. The internet is the same. We have all this processing power. Since we’re responsible for delivering mail to our customers, should we be responsible for cleaning it?”

We decided to back that idea. We created an alpha over a few months, then switched this technology on so we could route our customers’ traffic through this security tower we’d built.

One of the engineers had put a bell on it so that every time we caught something nasty, the bell went off.

When we switched it on, a shriek emitted out of the terminal.

We didn’t know what it was. It was horrendous. We switched the terminal off. Then the poor engineer realized it was his fault. Every millisecond, we were catching something bad. We knew immediately that we’d hit this internet oil. This was a real gusher of a proof point.

That created a course of events where we had to internalize: what do we do about this? Because every email user in the world has this problem. Is it B2B? Is it B2C?

We created an entirely new company called MessageLabs. It became the first major SaaS business in Europe. We reached $150 million in annual revenue within about eight years and sold to Symantec.

But it came from that process. When you’re with customers, don’t just sell your technology. Spend significant time in exploration mode, really understanding what’s going on in their minds.

If you’re doing a really great job for customers, you probably have two or three businesses you could build on the back of that insider understanding.

From Social Postcards to Direct Booking Economics

Chris Tottman: This was around 2013, the early advent of social channels. The founders came to me with an idea in the hotel space. Wouldn’t it be great if hotel brands could use check-in and check-out moments to get user-generated content? They would provide guests with a digital postcard of their stay that guests could post on social media.

My first meeting with this founder was in a beautiful London square. I like walking meetings. He was really compelling, but I didn’t like the proposition. I was thinking of the loop: I’m checking in, I’m checking out, I do this thing, I put it on social media, some random person sees it in my network. Are they likely to want to go to that destination?

It felt too weak. Too far away from a transaction.

But the founder was compelling. And he mentioned something intriguing: online booking agents like Booking.com and Kayak were starting to take significant percentages of booking revenue.

As I walked back, I kept thinking about it. Put yourself in the position of a young entrepreneur in the hotel space. You finance the building. You spend millions fitting it out. You assemble an incredible workforce. You open up, and you haven’t even taken a dollar in revenue yet. The CAC payback must be three to four years.

And then 15 to 20% of all revenue goes to a digital booking engine?

There was something there.

So we invited them into the office for whiteboard sessions. We mapped out the whole value chain of the industry. The journey someone goes through when booking a holiday. We looked at where time goes and where money goes.

They came with analysis showing that if you’re a small hotel with no purchasing power, booking engines were taking over 30 to 40% of booking revenue. Even big chains were giving up 15 to 20%. And the numbers were getting worse every year.

Then they did B2B and B2C customer research. What they found was fascinating.

65% of consumers had the hotel website open because it had better photographs, more detail. They felt more kinship with the hotel itself. But those same consumers also thought the online travel agencies (“OTA”) booking engine was cheaper.

The analysis showed that wasn’t necessarily true. And hotels could dynamically price against what was being offered online.

So they conceived the product: a pop-up on the hotel website showing three OTA prices alongside the hotel price, which was never more than the OTAs. A “never priced better” approach.

I gave them £250,000 to build that product. We launched three months later.

It went wildfire. The pent-up demand from hotels wanting to drive consumers to book direct was enormous. In the first two months, we sold to 250 to 350 hotels. Within a couple of years, we had about 1,000 at the mid-market, plus two of the eight biggest hotel chains.

The technology itself wasn’t thick enough initially. There wasn’t a big enough moat. So we knew we had to drive the price down to create more demand, more buyers, and then create more sophistication in the product.

But it was so interesting to see that process of mapping out the whole industry, having four or five different minds working on it, doing the analysis purely in pursuit of looking for a big enough problem, and trying to conceive a product that might solve the biggest, most pressing problem for those buyers.

That’s probably a $20 million revenue business today, serving most of the biggest hotels on the planet.

The Mindset That Matters

Michael Goitein: Just to toss a wrench into the conversation, we got a question about product-market fit. Once you’re starting to see traction, how do you drive toward it?

Chris Tottman: That’s the silver bullet question. We all argue about what product-market fit even is. It’s not something particularly stable. The hype cycle of establishing it, gaining traction, then losing it is happening faster than it used to.

We have a common phrase at Notion: “uncomfortably narrow.”

There’s a paradox when you’re pitching. The size of the market matters to investors because that’s how they decide whether they can make a return. The larger the fund, the larger the market needs to be.

But if you want to drive traction quickly and efficiently, you need a very small market definition. If you’ve got a very narrow market, it’s easy to target. If you have “we’re enterprise, we’re SME, we’re global, we do ERP and accounting,” that becomes impossible to market to.

First, you have to find a very large pressing pain. A pain that makes someone bang the desk to solve. One that makes them think, “If I don’t solve this, I might get fired. If we don’t have an answer by the next board meeting, we’ve got real issues.”

That’s the first challenge: finding that problem.

The second part is finding a solution where it’s easy to buy, easy to implement, the speed to value is very quick, and it’s really hard to get rid of.

90% of companies will not find product-market fit. They’re in the discovery period, constantly figuring out how to find the flywheel between a problem and a solution that fits. They’re looking for that moment when buyers and all their peers start thinking, “This is a new way we’re going to do business.”

Michael Goitein: I think that’s a nice addition. You see these opportunities just by getting out there and starting something. You see where you can take something in several potential directions.

Chris Tottman: It comes back to that difference between good and great founders. Great founders don’t over-index on trying to convince the world they were right. They over-index on trying to find a problem within the world they’re heading into and figuring out if there’s something ultimately much more significant they can address.

When they find that problem, they figure out how to address it in a way that’s 10x or 100x more economically viable for the buyer, which creates this whole wave of adoption.

The bigger the pain, the bigger the wave.

The Founder Who Missed the Opportunity

Chris Tottman: I had dinner a couple of weeks ago with a founder who ultimately wasn’t successful. The company did go on to be successful after he left, and we exited it. But he was brilliant. Just so brilliant. He came from e-commerce and was selling to big retailers with online presence.

He would sit down with people and give them this vision of the future.

I would say to him afterward, “Do you know what happens when you leave the room?”

He’d say, “What do you mean?”

“When you leave the room, they go down to the company kitchen and chat with a colleague. The colleague asks, ‘Who was that guy?’ And they say,

‘This brilliant guy with this great vision of the future. I hope he does really well with it. But I’ve got this pressing project. If I don’t get it fixed, I might get fired.’”

He would miss the opportunity to really understand what’s going on in the minds of the people making decisions, the people he would be selling to.

It’s such an important lesson. We’re all guilty of it. We get an idea and we’re like, “This is what I’m doing, this is where it’s going, it could work like this or that.”

But if you map out the market, map out the key constituents, and take a completely different approach to asking deeper questions, trying to understand how they’re feeling about a particular problem, how they’re addressing it, what experience they have with different ways of solving it, whether anyone else is solving it differently, whether it’s solvable at all...

If you get in the habit of doing that, you’re going to start finding opportunity everywhere.

Loom’s Hidden Superfeature

Michael Goitein: Loom is a fascinating story. I do quite a bit of work with Atlassian, and they have a whole connected workplace strategy.

Loom had started as a set of connected developer tools. You could share chunks of code, do pull request reviews, really integrate collaboration for engineering teams. Video was one small part of it.

But what Atlassian ended up acquiring them for was solving the problem of meetings.

What Loom became, and what Atlassian acquired, was async updates. Now woven into the entire Atlassian fabric of every product, whether Jira, Confluence, or anything else, you can find a little video icon, record a short clip, and that gives context. You don’t have to be great at writing. You record a short Loom clip, and it travels with that work unit, giving people context without you having to be there.

That’s been hugely successful for Atlassian. But the acquisition happened not because of Loom’s original official strategy. It happened because of the revealed strategy, overwhelmingly driven by user behavior.

Chris Tottman: That’s interesting. It’s like a hidden superfeature that someone reimagined in an entirely different way.

Michael Goitein: Exactly. It wasn’t just quantitative data. When you’re early stage, qualitative data matters. Keeping close to customers, talking about their experience, watching them struggle with the product, seeing how they actually use it. That’s where the real magic happens.

Chris Tottman: You’ve got to have your receptors open, trying to think: What’s really happening here? What are they using? What does the data say? What are they saying? What are they feeling? Does anyone care anyway?

The Constant Reinvention Cycle

Michael Goitein: How many of the companies you’ve funded pivoted to success versus those that decided not to pivot and didn’t succeed?

Chris Tottman: I don’t have the numbers, but I think pivoting is an outcome from a forced mechanism. It’s saying, “This isn’t working. We need to figure out something else.”

Sometimes something you’re already doing becomes the new product. Slack would be an example. Twitter is an example. Sometimes it comes from deep discovery like the case studies we discussed.

But the mental approach of always having deep discovery going on is universal.

It’s rare that you have a single value proposition that carries you through the 5, 10, 15-year life of a technology company. Normally we’re operating in extremely dynamic markets with alternative competitive threats. That innovation cycle is constant.

At MessageLabs, our mentality was always that we had to reinvent the company every 18 months. When we sold, we had 20,000 corporate customers, a couple hundred people in the US, 650 globally. We had seven products.

Two and a half of them were significant. The others helped us win, but did customers really use them? Did they really transform them? Not really. We were carrying the cost of those products, but we got the benefit of holding our market position.

You have to go from single product to multi-product. From UK SMB to enterprise. From UK to the US, continental Europe, Asia. You find product-related issues, market understanding issues, go-to-market issues. The US was more indirect; we were dominant in the UK but number two in the US with two competitors.

All of these things you’re constantly synthesizing to continue growing quickly, fending off competition, and keeping your team happy because they love winning.

The number one drug in a company is that we’re winning.

And every year, people don’t just give up their markets to you.

Final Thoughts

Chris Tottman: Inside the hopes and fears and minds of all these customers and buyers, there’s a business somewhere. There’s a problem worth solving. And there’s a whole bunch of problems that aren’t worth solving.

Go mine for that.

Hopefully, people can strike some gold as they go through that process of discovery.

Michael Goitein: Thank you so much, Chris. You’re doing a tremendous amount of incredible thought leadership while being on the front lines with these startups. We’re grateful to have your knowledge and wisdom here.

Chris Tottman: Thanks everyone. Happy hunting out there.

Chris Tottman is a partner at Notion.vc and writes “The Founder’s Corner” newsletter at the-founders-corner.com

Michael Goitein writes the “Product Strategy Decoded” newsletter at michaelgoitein.substack.com

Share

Account-Based Marketing leader Demandbase accidentally discovered something crucial about how sales organizations actually work.

When they decided to weaponize Gong as a cultural tool, not just a coaching platform, they launched “The Gong GOATs,” an internal program highlighting top performer interactions.

By making call reviews a centerpiece of team culture and featuring winning calls publicly, they turned conversation intelligence into organizational philosophy.

The “Gong GOATs” biweekly virtual sessions showcased sales leaders interviewing top performers. The format played call snippets from actual deals, then opened the floor for audience participation using three specific lenses: compliment the approach, offer constructive criticism, or ask for clarity.

Managers weren’t just looking to celebrate individual wins. They were looking to extract patterns across multiple deals, identify which techniques replicated successfully across different account types, and then use Gong’s adoption tracking to verify whether reps implemented those techniques in subsequent calls.

The hard numbers speak volumes about this approach’s effectiveness.

After reviewing their own calls in Gong for opportunities valued at over $100k, sales reps increased deal values by a solid 61%.

But managers reviewing identical interactions with the same platform and data increased deal values by a stunning 111%!

That organizational perspective shift created 50 percentage points of additional value, a lift directly attributable to Gong’s strategic choices.

Subscribe now

The Problem Everyone Else Was Solving For

To understand how Gong’s organizational-level focus created 50 percentage points more value requires going back to 2015, and the origin of a choice most conversation intelligence companies avoided.

While running business intelligence startup Sisense in 2016, veteran entrepreneur Amit Bendov kept hitting the same wall. Despite having the most up-to-date CRM systems, sales leaders were still operating in the dark due to fragmented data.

Because those CRMs only captured what reps remembered to type, leaving out everything customers actually said during sales calls.

When advances in deep learning and natural language processing made large-scale transcription economically feasible, Bendov recognized the opening. While investors were initially skeptical, Bendov and co-founder Eilon Reshef raised $6M in October 2015.

As privacy concerns, competitive threats from Google, Amazon, and sales rep resistance swirled, Bendov and Reshef pushed forward.

Not Your Average Sales Productivity Tool

While SaaS companies build B2B tools to optimize a single user’s tasks, Gong made a different choice.

Gong launched in 2016 with both automatic transcription and an added layer of conversational intelligence: Trackers.

“Trackers” were keyword-based searches that automatically surfaced conversation insights across the entire org. Need to find every interaction where a competitor got mentioned, and understand how the org’s best reps handled specific objections?

This was the next-level brilliance Gong brought to sales teams.

While a single sales rep handling 50 calls quarterly might notice they consistently struggle with pricing objections, sales operations teams analyzing Tracker data across 2,000 calls could identify more granular insights. Like mentioning timelines increased close rates in financial services deals, but decreased them in manufacturing conversations.

One person optimizes their own performance. A sales operations team identifies the higher-level patterns.

Marketing = Strategy: Creating the “Revenue Intelligence” Category

Another subtle but crucial step Gong took was to create its own category.

From its 2016 launch, Gong focused on playing in the “Conversational Intelligence” category, but quickly ran into two major problems:

1. They needed to attract and hold the attention of senior revenue leaders: During sales calls, decision-makers wanted to push them to sales enablement. The Gong team had to figure out how to call what they did “something they would care about.”

2. Differentiate: The call recording and coaching space was already saturated, but the Gong team had a “much bigger, deeper vision for how Gong and our technology could be used for managing deals, prospecting, and forecasting.”

And so, in October of 2019, the Gong team came up with the new category of “Revenue Intelligence.”

This change solved both of Gong’s challenges:

1. Revenue leaders had to listen, because “Revenue Intelligence” was in their title.

2. Differentiation was easier “because we could tell people how our category is bigger, better, and different.”

Gong knew its category was working when “18 months later, *every single* competitor started using the term ‘Revenue Intelligence’.”

And its own growth and revenue really began to take off as it was helping enable its customers’ revenue.

Creating and dominating this bigger category was a key part of Gong’s strategy.

The Strategic Architecture Behind the Valuation Gap

Every choice Gong made reinforced its revenue intelligence category creation choice and org-level decision-making optimization focus.

Winning Aspiration

Gong aspires to become the single source of truth revenue operating system where every organizational decision would be grounded in customer interaction data.

Bendov’s vision: Reps just talk with customers, and Gong handles everything else, providing insights without manual drudgery.

Where to Play

Gong deliberately targets enterprise B2B organizations with 1,000+ employees in SaaS, healthcare, financial services, and manufacturing, where sales require multiple buyer personas and lengthy consultative cycles.

50-person sales teams can resolve alignment problems and share best practices in weekly meetings and 1:1s.

1,000-person revenue orgs need shared data infrastructure. Because they face the severest organizational fragmentation due to the deepest data silos, yet need to have the most accurate forecasting, enterprise sales teams are under the highest pressure to enable sales process excellence at scale.

How to Win

Gong’s differentiation is built on two key pillars:

Data Supremacy: The Gong Revenue Graph captures an unrivaled customer interaction data set, with over 3 billion customer interactions analyzed across 100x more data at a scale CRM competitors couldn’t even imagine. This proprietary data set continues to compound in value with every single sales call.

Targeted AI for Revenue: Gong’s proprietary AI models are exclusively trained on revenue team workflows and sales conversations. Its conversation intelligence analyzes talk ratios, objection handling, sentiment shifts, competitor mentions, and buying signals across 300+ unique deal data points.

Further solidifying its superiority, Gong Forecast predicts deal outcomes with 20% greater accuracy than predictions based on CRM data alone because it’s based on customers’ exact words during calls rathter than forecasting based on what sales reps might have remembered to type into Salesforce after the fact.

What sets Gong apart from every other revenue software company is its ability to give Sales, Product, Marketing, Customer Success, and Finance all access to identical customer interaction data, rather than operate from departmental silos.

Capabilities

Gong’s organizational decision-making at scale is enabled by a unique set of capabilities.

AI and Machine Learning excellence, highlighted by proprietary NLP models trained on sales conversations for transcription and sentiment analysis.

Unique conversation analytics engine that tracks key sales conversation data points against industry success patterns. Data infrastructure for automatically capturing interactions across channels.

Revenue thought leadership content marketing through Gong Labs, publishing research powered by its unique dataset. Enterprise sales execution orchestrating complex multi-threaded account engagement.

Customer success systems help organizations get onboarded quickly and keep them deeply engaged.

Gong’s capabilities were carefully designed and orchestrated to support its Where to Play and How to Win choices, delivering cross-organizational visibility and decision-making clarity.

Management Systems

Gong’s foundational three-layered architectural approach across its Data Engine, Intelligence Layer, and Application Layer allows it to rapidly respond to customer feature requests, further enabling entire organizations to operate from a shared customer reality.

In a classic example of “dogfooding,” Gong uses its own platform for org-wide visibility into sales best practices and continuous sales performance optimization.

Gong heavily uses its own automated coaching scorecard analysis, surfacing coaching opportunity patterns for targeted improvement feedback.

Customer engagement metrics track any sign of at-risk accounts to keep companies continuously engaged and generating richer revenue intelligence datasets

How Revenue Insights Data Compounds Into Advantage

Every Gong customer not only builds a compounding advantage for their own revenue org, they also contribute to the success of other Gong customers.

When Demandbase identified that specific discovery questions worked in one set of deals but failed in others, Gong added that pattern to its model training data. When another customer discovered that discussing budget approval processes in week three of enterprise deals correlated with faster close times, that insight improved Gong’s models for all its customers.

Each customer interaction captured and each successful pattern identified can then scale across the entire revenue organization and feed back into Gong’s platform.

The compounding effect works at two levels.

Individual organizations compound their own sales excellence. Demandbase went from identifying isolated winning techniques to systematically sharing them across all enterprise accounts. More patterns implemented generate more revenue, creating a flywheel effect that drives deeper Gong adoption.

And by continuously feeding these anonymized success patterns into its platform, Gong compounds its advantage across its entire customer base.

Gong architected around the only dataset in B2B software that compounds organizational decision-making capability: what customers actually say during every sales conversation.

Why Competitors Can’t, And Won’t, Easily Replicate Gong’s Choices

Roger Martin’s test for sustainable competitive advantage asks whether you’ve built critical capabilities competitors can’t replicate, and whether you’ve made strategic choices competitors refuse to make because doing so conflicts with one or more aspects of their existing business model.

Gong passes both tests.

Competitors cannot replicate Gong’s primary data advantage. Salesforce cannot access the conversation data locked in Gong’s platform. Clari cannot train forecasting models on interaction patterns they’ve never captured. Chorus, even with similar recording technology, entered the market later with a fraction of the conversation data Gong accumulated from 2016 forward.

That compounding data advantage widens every day.

Salesforce’s entire business model depends on sales reps manually entering CRM data. Salesloft optimizes sales engagement workflows for individual reps. Clari built forecasting around pipeline data.

Gong’s strategic choice to serve the organization rather than the individual operator created a competitive advantage that its competitors can’t, and won’t, attempt to attack.

Not only are they all years behind in developing similar capabilities, but the strategic pivots required would contradict every single one of their existing business models.

What Demandbase’s Success Proves About Gong’s Choices

When Demandbase launched its “Gong GOATs” program, a sales talk show featuring top performer calls, they were simultaneously celebrating individual excellence and operationalizing the org-wide revenue excellence Gong’s choices optimized for.

Managers could review calls, not for individual rep coaching, but to extract patterns, identify what worked across deals, connect those patterns to higher-level company strategies, and roll those patterns out across the revenue org at scale.

That 111% versus 61% differential reveals how Gong wins through relentless focus. Individual reps reviewing their own calls got enough personal insights to identify improvements for their next calls. That 61% deal value lift was meaningful but limited to one-off conversations.

Managers reviewing identical interactions accessed cross-deal context, enabling them to see patterns across multiple customer conversations. This helped them understand competitive dynamics and product strengths across dozens of interactions, allowing them to connect insights to higher-level account strategy and inform company-level strategic choices.

Both individual reps and managers had the same access to Gong’s platform. The difference came down to product strategy. Gong’s designed its platform specifically to enable managers to see the bigger picture.

That “bigger-picture” perspective is how Demandbase’s deal value jumped 50 percentage points.

What Gong’s Success Means For Every SaaS Company With Customer Data

Too many SaaS companies make the same architectural mistake: narrowly optimizing their products for frontline operator efficiency and productivity.

The individual engineer using your dev tools. The marketer using your analytics platform. The customer success rep using your support software. The product manager using your feature analytics.

The instinct makes sense. Your primary user is who you talk to. Who tells you about feature requests. Who influences adoption decisions and directly asks you to build better features for them. Celebrate their productivity gains. Growth follows.

But Demandbase’s 111% versus 61% gap suggests companies are leaving substantial money on the table.

The real leverage in SaaS doesn’t sit with optimizing individual operator workflows. It sits with the whole organization, making better decisions through shared customer intelligence.

Gong’s Real Value

Gong doesn’t sell call recording. A hundred companies do that. They sell organizational alignment around a shared customer reality.

Managers, operations teams, finance leaders, and product teams all accessing identical data about what customers actually want, need, and struggle with, then making decisions together based on the shared insights.

SaaS companies with access to customer data, including analytics, customer success, feature usage, payment behavior, communication patterns, and product signals all face Gong’s architectural choice: Optimize for individual productivity or organizational intelligence?

The companies that win and achieve outsized valuations consistently choose organizational intelligence.

How AI Compounds Competitive Advantage

As AI capabilities converge, customer intelligence alone won’t create sustainable competitive advantage.

Multiple companies can now transcribe calls, analyze usage patterns, and surface insights from customer behavior. Strategic differentiation shifts to products that are architected around the right user, and capture unique compounding usage data to surface actionable insights.

Demandbase’s 50-point gap proved that strategic choice. Products architected for organizational intelligence build a compounding shared customer intelligence asset, allowing users to both make better collective decisions and feed those insights back into their platform to continuously improve it for all users.

Gong’s $7.25B valuation reflects a competitive data advantage that widens with every customer interaction, every pattern identified across deals, and every organizational decision grounded in what customers actually said.

Share

Click here to see this post as a Gamma presentation.

Join me for my next live video in the app on November 19, 2025, at 12:00pm Eastern time with VC and startup thought leader of the brilliant “Founder’s Corner” newsletter.

Subscribe now

Here’s the full talk as a refined transcript:

Over the past six months, we’ve analyzed how leading companies are approaching AI within their product strategies. Rather than adding features or chasing every new model release, these four companies made fundamentally different choices about how to embed AI into their core capabilities. Their strategies reveal something crucial: AI is never the strategy itself. It becomes the capability that enables your strategic choices.

Four Different Paths to Competitive Advantage

Otter: Compounding Value Through Proprietary Infrastructure

Otter approached AI differently from the start by building their entire stack from scratch rather than relying on publicly available models. They identified a specific constraint: their users operate in professional meetings across remote-first environments, often attending multiple meetings per week.

Their strategic choice centered on making transcript and insight sharing the default. A single user attending multiple meetings might extract moderate value, but after 100 meetings, something shifts. Users can search across all conversations to find patterns, ask questions across meetings, and develop deeper organizational knowledge. This radiated value touches roughly 7.5 people per conversation. The system becomes more valuable the more it’s used, creating a compounding competitive advantage that scales with adoption.

Granola: Privacy as a Strategic Differentiator

While Otter competed in the transcription space, Granola observed a critical gap: existing solutions created friction in sensitive conversations. Otter requires a visible bot in the meeting. Zoom demands explicit recording announcements. These create barriers to using the product where it matters most, in C-suite and venture capital conversations where confidentiality is paramount.

Granola made an inverse strategic choice. They built AI processing directly on the device and deleted the recording immediately after, making the AI completely invisible to other meeting participants. This design eliminated the consent requirement and allowed them to achieve SOC 2 Type II compliance, opening access to markets where any leak could tank stock prices.

Where Otter asked how to create compounding value through sharing, Granola asked how to unlock value by preserving privacy. The strategic difference determined everything else.

Gamma: Building Flexibility Into Infrastructure

Gamma’s origin story offers a different lesson. The company started with a conventional idea: bringing PowerPoint online. Running out of runway forced a strategic bet. The team decided to stake their future on AI, but with a critical distinction: they built AI as an orchestration layer sitting above other models rather than committing to any single model.

In three months, they mastered something most companies don’t attempt: the ability to call different models for different components of a single task. When creating a presentation, their system analyzes which models deliver the strongest outlines, which produce the best graphics, and which format content most effectively. When a model shows weakness, they can swap in an alternative without disrupting the user experience. As new models emerge, they’re not vulnerable to disruption. They simply integrate the new capability.

Gamma targeted creators, knowledge workers, and educators who needed to sell ideas. Instead of replicating PowerPoint’s interface, they reconceived how people present information using cards created on the fly and continuously improved through model comparison. Their breakthrough wasn’t an individual feature. It was the infrastructure that allows them to choose the best tool for each specific task.

Shopify: AI as Infrastructure for Acceleration

Shopify took yet another approach. Rather than building AI as a visible feature or proprietary capability, they embedded AI deep into their infrastructure layer. The distinction matters: they don’t use AI to automate or reproduce manual tasks. They rethink linear processes into non-linear, continuous workflows from the ground up.

Store setup, which previously required days, compresses to hours. SKU imports accelerate. The value compounds through merchant velocity, not through visible feature count.

Shopify targets growth-stage merchants who represent a structurally different opportunity than Amazon Marketplace sellers. These merchants never face displacement by Shopify itself, because Shopify operates in the enablement space. The value flows through a partner ecosystem that provides capabilities Shopify could never build alone. AI disappears into the background, functioning as a speed multiplier that serves a specific market with specific constraints.

The Underlying Pattern: Strategic Constraints Drive Design

Each company confronted a different set of constraints that forced their innovation:

Granola couldn’t use a visible bot, so they created invisible on-device processing. This constraint became their differentiator.

Otter focused on English-first transcription at 95% accuracy rather than pursuing 100% accuracy or supporting 50 languages. Accepting this limit allowed them to ship and iterate.

Gamma rejected building a proprietary model entirely. This constraint forced them toward orchestration, which became their core capability.

Shopify approached AI not as a feature factory but as an acceleration layer for existing workflows.

The pattern suggests something important: when you work within constraints rather than against them, your strategic choices become clearer. Your capabilities become more differentiated. Your competitive advantage becomes harder to replicate.

What Changed in AI Strategy

The companies we analyzed didn’t ask traditional questions. They didn’t ask how to add new AI features, how to build a better stack than competitors, or what tasks to automate.

Instead, they asked different questions entirely: Where can we serve customers that currently go underserved? What structural advantages do we have in customer access, data, or behavior that competitors cannot replicate? How does AI enable us to reach those customers with capabilities that wouldn’t be possible otherwise?

AI enabled their strategic choices rather than replacing them. The strategy remained about serving specific customers in specific ways. AI simply made those choices executable at a scale or speed or quality that wasn’t previously possible.

Planning for Disruption in 2026

As organizations enter planning cycles for next year, the velocity of change raises an obvious question: How do you plan when the landscape shifts constantly?

The answer may resemble Gamma’s orchestration layer. Rather than trying to predict which specific innovations matter, build a layer above the disruption. Start with your customers and what they actually value. Then assess which new tools, capabilities, and models serve those customers. Not every advancement in 2026 will apply to your business. Some developments will reach your market; others won’t. But if you’ve made clear decisions about where you play and what your customers need, you can adopt different components as they become relevant.

This requires stepping back from the firehose of AI announcements and information. The companies that win won’t be those keeping pace with every release. They’ll be those clear about their strategic choices, able to integrate useful tools when they emerge, and ruthless about ignoring everything else.

Upcoming Analysis

Our next breakdowns focus on companies like Vanta, which demonstrate AI’s value in handling complexity that humans can’t maintain independently.

Vanta enabled Granola to achieve SOC 2 Type II compliance in under 12 weeks, a process that traditionally requires months. Beyond point-in-time compliance, they’ve built continuous compliance monitoring, a paradigm shift that fundamentally changes how organizations approach vendor risk and regulatory requirements.

Be on the lookout for upcoming breakdowns of leading SaaS AI startups like Gong, Rippling, Ramp, Vercel, Monzo, and Brex.

These detailed analyses will shift to a shorter format, moving from 4,000-plus-word breakdowns to approximately 800-1,200 words, making the insights more accessible for reading in roughly 5-7 minutes.

Future editions will incorporate more visuals and infographics to make information denser and more digestible.

Share

Every company is pouring billions into AI right now, using it to build faster recommendation engines, better churn prediction, and smarter customer segmentation.

They’re winning at optimization, A/B testing everything to oblivion.

But they risk perfecting solutions to questions their customers and competitors have already moved on from.

Even the best AI execution can’t help you if you’re executing in the wrong places.

Subscribe now

The Illusion of a “Moat”

Companies try in vain to build “moats” to protect their advantage.

But lasting competitive advantage at this pace of change won’t come from doing the same things better or faster.

Many AI strategies will fail because too many product leaders are stuck using yesterday’s outdated concept of moats.

Smart competitors aren’t betting on a single moat as “the one answer” to stay ahead of the competition.

Instead, they’re discovering the next set of questions.

This isn’t a new idea. saw this pattern decades ago and created a powerful metaphor around it.

Moving Through The Rooms

Martin maintains that companies don’t compete on answers.

They compete on the questions they ask.

Martin’s powerful metaphor sees competitive advantage as a series of “rooms” separated by curtains.

Each “room” is defined by the set of questions about customers, technology, and markets that companies ask and answer. Leaders move into and temporarily “own” a room by asking a new question, innovating, and delivering against it.

Innovative leaders can move to the next “room” by asking questions that competitors haven’t even recognized as questions they should be asking.

From Powder to Pod: Tide’s Shifting Rooms

P&G’s Tide is a great example of winning the detergent category by asking better questions.

Tide didn’t win by being more efficient or by shipping product faster. They won by continuously asking new questions, which led to innovating with new products customers loved.

First room: Early competitors asked, “How do we make detergent work?” P&G had already created a great powdered detergent, and asked a different question: “What if powdered detergent could be more compact?” So Tide’s compact powder was the first shift. Tide “owned” that room for a time until competitors caught up, turning compact powder into a commodity.

Second room: While competitors still thought of bleach and detergent as separate problems, P&G next asked an unheard-of question: “What if liquid detergent could have bleach built in?” And so they moved to the next room, taking ownership of the liquid detergent category.

Third room: As customers shifted towards greater energy consciousness, Tide asked a question competitors never saw coming. “What if Tide worked as well in cold water as other detergents do in warm?” By asking that next question, breakthrough innovation followed, and Tide again moved ahead to the next room.

The Tide Pods moment: The Tide team achieved another huge breakthrough by shifting from the product itself to focus on delivery. Everyone has struggled with bottles of liquid detergent and measuring cups. P&G asked a new question: “What if ease of use and dosing were the problem?” Tide Pods represented a fundamental shift to another room entirely. While competitors were still optimizing cold-water formulas, P&G created an incredible new user experience.

P&G’s advantage went beyond mere superior execution. It was in superior question discovery to ask the kind of questions competitors didn’t even know they should be asking.

This pattern shows up everywhere. In failure more obviously than success, when they failed to ask the next question.

The Danger of the Comfortable Room

Companies unaware of this question mindset shift leave themselves vulnerable to competitors moving into their “room.”

Too many only wake up when their room is full and they’re the ones not delivering the delight customers have come to expect.

Medium’s publishing self-destruct: Medium asked: “How do we democratize publishing and give writers a platform?” They built it beautifully. They were dominant.

Substack asked an entirely different question: “What if writers could own their direct relationship with their audience?” And what if they could recommend each other and become a discovery platform, giving writers a direct line to readers?

Medium didn’t even think to ask these questions. Instead, they kept optimizing their discovery/curation model while writers migrated to Substack to own their relationships with their subscribers. Worse, Medium responded poorly by cutting payouts to writers, looking to optimize yesterday’s question while the market itself had moved.

Duolingo’s AI fiasco: Also in the education domain, Duolingo asked, “How do we make language learning engaging and accessible?” They won on that question. But new competitors started asking a different question: “What if the problem isn’t engagement mechanics but learning psychology?” Duolingo’s AI doubled down on aggressive retention tactics, even mocking users who expressed concerns. They were executing brilliantly in yesterday’s “retention” room while the market had moved to a “science-backed sustainable learning” room.

Both companies had the distinction of executing brilliantly in rooms competitors had already left.

But the real key is understanding what these patterns reveal about how competitive advantage works today.

How AI Accelerates the Room Shift

Without understanding the power of moving continuously from room to room, companies risk operating under the illusion that competitive advantage comes from better AI alone.

Faster models. Better data. Smarter algorithms. But the real competitive battle is happening one level above execution. It’s about asking questions about your customers, market, and capabilities you’re able to ask.

When other companies start asking your question, and you see them “peeking through your curtain,” you’ve already lost. The companies pulling ahead aren’t building better solutions in crowded rooms.

They’re entering the next room by asking the right question before competitors even know the room exists.

By making execution in any room faster, better, and cheaper, AI accelerates execution through any curtain and into any room, meaning they’ll all now fill up faster.

But it can’t tell you which room to move into next.

That still depends on asking questions your competitors haven’t yet imagined asking.

Share

Fintech startup Ramp was stuck.

It had hit the ceiling and could no longer scale beyond its existing client base because it could no longer manually update the spreadsheets needed to track regulatory controls.

Then they discovered Vanta.

Fast forward two years, and Ramp’s growth exploded exponentially with just two compliance people, thanks to Vanta’s radical regulatory mindset shift.

Vanta allowed Ramp to move compliance from a bottleneck that had stalled its enterprise expansion into a revenue driver, unlocking exponential growth.

Subscribe now

Christina Cacioppo’s Dropbox Insight

As Christina Cacioppo led Dropbox Paper’s 6–12-month SOC 2 process, she identified a buried customer pain point.

Every other startup considered compliance as a cost center, a regulatory burden, something to avoid at all costs.

The Contrarian Wizard of Oz

Christina embedded with Segment and watched founders waste hours spreadsheet-wrestling with color-coded compliance requirements just to prove they were trustworthy.

Christina built a “Wizard of Oz” demo to show founders what automated compliance data collection could look like in real time. When they saw their audit readiness compressed from nine months down to 2-4 weeks, the reaction was immediate. The demo solved their core constraint by unlocking the enterprise deals that seemed impossible to close before.

Y Combinator startups validated this insight, recognizing compliance automation as a revenue accelerator, not just a cost-reduction play.

Christina explained her product/market fit realization after selling Vanta’s first deals based on the value of the problem she was solving:

“I was able to sell the first $500k of Vanta by myself, when I didn’t really know how to sell. I joke that prior to Vanta, the last thing I sold was Girl Scout cookies. I knew it wasn’t my sales skill that was getting these deals closed; it was a testament to the problem we were solving and the product we were building.”

Insight confirmed, Christina would now have to tackle the harder part of building an entire set of strategic choices and a company architecture around it.

Every strategic choice would flow from that single reframe.

Reverse Engineering Vanta’s Regulatory Revenue Transformation

I. Winning Aspiration

Vanta’s stated aspiration is to “Help businesses earn and prove trust.” Christina sharpened this focus by identifying a specific customer: founders who understood that enterprise buyers needed proof of trust and that acquiring it was the fastest path to scaling revenue. That insight shaped the other choices that follow.

II. Where to Play: Validating and Expanding The Insight

Starting With Maximum Pain

By starting with US startups first, Vanta identified the segment with the highest urgency. Founders were desperate to close enterprise deals but couldn’t without proof of certification. If willingness-to-pay existed anywhere, it would be here. And it did. Founders would pay a premium for speed because it directly unlocked revenue.

Progressively Uncovering Value For Larger Segments

Ramp taught Vanta a key lesson: velocity matters when you’re founder-driven and fighting on every front.

But as Vanta scaled beyond startups into hypergrowth companies, a second pattern emerged. Ramp operated across multiple jurisdictions and industries simultaneously. Generic compliance approaches would never work at this level of complexity.

Vanta built custom framework capabilities to handle this complexity.

As Vanta reached mid-market and enterprise companies, the lesson was clear: enterprises wanted the ability to customize their compliance needs at infinite scale.

The technical and organizational customization infrastructure Vanta built for Ramp didn’t turn out to be “gold-plating,” but ended up unlocking the next larger target segment’s value proposition.

III. How to Win: One Insight Driving Three Choices

Speed As Startup Energizer

For founders barely clinging to runway, velocity is everything.

Startups don’t have the luxury of waiting. When an enterprise buyer showed interest, the window was small. Vanta compressed audit-readiness from 6-9 months to less than a month. That speed differential converted interest into contracts when every other vendor was still processing paperwork.

This speed made all the difference between closed deals and lost opportunities.

Integration Breadth as Hypergrowth Enabler

Hypergrowth companies needed something different: speed advantages alone wouldn’t be enough.

They operated across 15+ frameworks simultaneously and needed evidence to flow continuously. Vanta’s 375+ connectors removed the spreadsheet sprawl.

Hypergrowth companies weren’t constrained by audit time, but by the operational burden of manually feeding compliance data.

Custom Frameworks for Enterprise Scale

As Vanta began talking to enterprises, it learned the most important lesson: complexity would prove its competitive advantage.

Generic, legacy compliance tools would have no chance to manage 15+ frameworks across jurisdictions and industries simultaneously.

Enterprises had a different priority entirely: staying compliant across every market they entered.

Vanta’s customization at scale became the lever that removed regulatory constraints blocking enterprise global expansion.

IV. Core Capabilities: Delivering Targeted Solutions For Each Segment

Continuous Integration Engineering

When manual spreadsheets were the constraint, Vanta shifted to provide continuous, automated evidence collection.

Its unique integration architecture removed that burden, allowing evidence to flow automatically instead.

AI-Powered Automation

As hypergrowth customers presented Vanta with increased complexity, manual intervention once again became the bottleneck.

Vanta’s AI Agent automated policy responses and questionnaire completion, shifting from point-in-time proof to continuous verification.

For enterprises managing multiple jurisdictions, continuous state verification is essential.

Deep Domain Expertise

Vanta mastered mapping regulatory controls across 35+ frameworks, industries, and jurisdictions.

This deep understanding of the regulatory landscape allowed Vanta to offer true customization rather than templated solutions, which meant the difference between customization and checkbox compliance.

V. Management Systems: The Metrics That Operationalize Vanta’s Strategic Choices

Measuring Revenue, Not Just Hours

Vanta’s metric choices operationalize its core strategic bet: turning compliance from a cost center into a revenue lever.

It chose to measure revenue unlocked alongside traditional SaaS metrics like Annual Recurring Revenue (ARR) and Net Recurring Revenue (NRR).

These measurement choices force every Vanta team to think about how their work enables customer revenue, not just customer compliance.

Organizational Design That Delivers Customization

Vanta structured itself as distributed engineering pods built to handle regional and regulatory complexity.

Each pod owns the technical depth to serve companies across multiple jurisdictions and industries.

Partnerships That Prove the Insight

Cacioppo closed Vanta’s first deals directly with founders who had this pain.

This led to a partner ecosystem of MSPs, audit firms, and integrators that scaled Vanta without proportional growth in sales team headcount.

Ramp evolved from customer to partner, proving that enterprises see compliance customization as a revenue necessity—something worth investing in because it delivers measurable impact.

Vanta’s Intersection of the Three Choices That Matter

Vanta’s strategic shift revolutionized how organizations approach compliance, as regulatory proof became a deal-closer and a hard-dollar revenue driver.

Vanta needed three conditions to pull off this regulatory wizardry: speed, breadth of integrations, and customization at scale. Without all three, they’d be just another checkbox compliance tool.

Ramp’s two-person compliance team managing global hypergrowth proves Vanta found that intersection. When compliance is continuous and automated, it stops being a bottleneck.

Vanta’s ability to master the complexity that once made compliance painful now powers its advantage, unlocking revenue for countless organizations.

Share

Too many companies treat AI like a feature to bolt onto existing products.

They add a copilot here, automate a workflow there, launch an AI-powered tool, and wait for adoption. Six months later, their competitive edge vanishes.

Meanwhile, the billion-dollar AI powerhouses win by playing a different game entirely.

Shopify rebuilt its entire operating model around AI-first process reimagination.

Granola built an entirely new architecture around on-device processing to access markets competitors structurally couldn’t reach.

Otter.ai designed a strategy based on permissionless adoption and compounding organizational knowledge.

And Gamma went from near-bankruptcy to $50M ARR by orchestrating the best of every LLM, enabling speed and performance advantages that locked-in competitors cannot match.

Through the power of AI, all of these foundational strategy architectures compound faster than competitors can copy superficial features.

That’s the hidden competitive advantage that outlasts feature parity.

Subscribe now

Winners Ask a Different Question

Most companies approaching AI ask variations of the same questions: “How do we build better AI features than competitors? How do we automate more? How do we match what the other guy is doing?”

Feature parity becomes a commodity within a year. Competitors will copy your capabilities in six months. They’ll outspend you on infrastructure. They’ll hire smarter researchers.

As I’ve said before, the quality of your questions determines the quality of your strategy. Winners ask the kind of power questions that unleash completely different outcomes:

• Where are we uniquely suited to deploy AI, given our existing position?

• Where does AI unlock a market segment our competitors structurally cannot access?

• What constraint can become our advantage through AI?

These questions go a level deeper beyond producing me-too products to produce asymmetric, compounding competitive advantages.

Otter.ai: How Viral Permissionless Adoption Creates Compounding Advantage

Otter.ai decided early to build a proprietary AI stack optimized for real-time transcription and lightning-fast meeting insights that could be shared virally, reaching the whole org.

Traditional Enterprise SaaS requires procurement consensus and top-down adoption of B2B software. How do you drive adoption without permission from everyone above you?

Just like Slack did by making bottom-up adoption so frictionless it spread without requiring IT approval.

The Happy Viral Infection

But that means making your product so easy to use and delightful that one user’s decision automatically creates value for the entire org without needing to request approval.

Otter acts like a “beneficial virus,” “infecting” non-users with free transcripts and takeaways, who suddenly gain significant value regardless whether they were in the meeting or not.

With each meeting touching an average of 7.5 users, the virally spreading, permissionless, bottom-up adoption becomes a compounding distribution engine flywheel.

Otter trained its real-time architecture on 50 billion+ minutes of meetings. Ambient intelligence builds compounding organizational knowledge over time. The longer you use it, the more valuable it becomes, and the harder it is to replace.

Result: Otter.ai has achieved $100 million in Annual Recurring Revenue (“ARR”) with fewer than 200 employees.

Winning Aspiration: Searchable organizational knowledge across all meetings

Where to Play: All professional meetings across industries (true horizontal reach)

How to Win: Permissionless value creation where one user generates value for the entire organization

AI’s Role: Compounding, valuable knowledge asset that strengthens with use

Granola: AI as a Privacy Shield

Granola discovered that every competitor’s core feature became their fatal flaw.

Meeting bots and visible signs of recording remain the industry standard, but have no access to higher-value conversations.

Think of it like bringing a court stenographer to an evening of bourbon, poker, football, and cigars.

As soon as people feel they’re “on the record,” you can kiss those honest, frank conversations goodbye. Players suddenly speak in careful, “safe” language. The fun, relaxed, open exchanges you tried to create evaporate the minute anyone knows it’s “official.”

And this is exactly how visible recording makes people feel in high-stakes executive conversations.

Let The Party Begin

The Granola team’s foundational choice was radical: build meeting intelligence without visible bots, stored audio, or consent overhead.

This involved developing on-device processing, real-time enhancement of human-written notes, and preserving the cognitive benefits of active note-taking while enhancing them with AI intelligence.

Granola’s AI guides insights into organizational memory while deleting the recording. Add SOC 2 Type 2 compliance, and suddenly, markets opened that competitors couldn’t reach by building “better meeting bots.”

Result: $250 million valuation in under two years by solving a problem competitors didn’t even see.

Winning Aspiration: Invisible organizational memory for confidential conversations

Where to Play: Executive-level, high-stakes conversations (boardrooms, VC calls, Legal & HR cases)

How to Win: Human-AI collaboration that preserves active note-taking without visible recording

AI’s Role: Privacy shield that unlocks structurally inaccessible markets

Gamma.app: AI as Flexible Orchestrator

Funding fast running out, Gamma.app faced bankruptcy.

Too many users had tried their “PowerPoint-killer” tool once and never returned. Retention was abysmal, and they had less than a year of runway left.

CEO Grant Lee could either give up or rebuild. He chose to rebuild Gamma from the ground up to become an AI-powered visual medium.

And Gamma did it by solving a more complex problem: orchestrating 20+ external models simultaneously. Perplexity for research outlines. Claude for reasoning. GPT-4 for visuals. When a better model emerges next week, Gamma can switch in days. Competitors locked into single-model dependencies need months to retrofit.

Choosing the Best On the Fly

Competitors bet their “restaurant” on a single AI model supplier.

Gamma became the “AI master chef” who can constantly source from anywhere, swapping in the freshest “ingredients” to create the greatest customer delight the moment something better appears on the market.

Result: Gamma gained 10,000 new users per day post-launch, achieving $50 million in ARR with just 40 employees. Flexibility and orchestration became their durable competitive advantages.

Winning Aspiration: Effortless creation of quality presentations and design assets for non-design professionals

Where to Play: Individual creators, educators, marketers, anyone needing sharp visual communication

How to Win: Multi-model orchestration enabling rapid model swaps to achieve the fastest and best results while competitors are still integrating

AI’s Role: Flexible, modular orchestrator that adapts faster than competitive timelines allow

Shopify: AI as Speed Multiplier

Shopify CEO Tobias Lütke isn’t interested in becoming the next OpenAI. He’s weaponizing AI to compound the competitive advantages BigCommerce, Adobe Commerce, and WooCommerce already can’t match.

While those companies scramble to add AI features, Shopify shrank merchant setup time from months to hours, automated ecosystem partner support, and turned customer success into a self-reinforcing competitive moat.

Shopify didn’t bolt AI onto existing workflows. It undertook a ground-level architectural rethinking.

The Only Way to Travel

Think of it like this: while BigCommerce and WooCommerce customers are still grinding gears in old, clunky manual transmission cars, Shopify hands its shop owners fully autonomous, self-driving Waymo cars.

You might get to the same destination in both vehicles, but the driving experience couldn’t be more different.

Result: AI is enabling Shopify to compound its strategic advantages across e-commerce workflows 10x faster than competitors can respond.

Winning Aspiration: Merchant empowerment through operational velocity

Where to Play: Growth-stage merchants ($100K–$50M+ revenue) who value brand independence

How to Win: AI-as-infrastructure eliminates manual work and compresses time-to-value

AI’s Role: Work eliminator that disappears into the background; speed becomes the differentiation

The Two Hidden Patterns Guiding the Winners

These four companies reveal a hidden pattern underneath the AI “buzz” and chaos.

Pattern One: Constraints Breed Innovation

Granola decided against visible bots. Otter bet deeply on English while competitors diluted across 50 languages. Gamma figured out there was a better way than either building proprietary models or using a single one. Shopify couldn’t scale.

Every strategic advantage came from identifying a constraint and building specifically around it. Weak competitors see constraints as problems to avoid. Winners see constraints and setbacks as the raw material for competitive advantage and success.

Granola’s constraint (no visible recording or bots) became the foundation of its architecture. Gamma’s constraint (rebuild around a unique AI capability) became the key to its explosive growth.

Pattern Two: AI Amplifies, It Doesn’t Create

Otter was already building horizontal meeting intelligence. Granola was already focused on privacy. Gamma was already obsessed with speed and user experience. Shopify was already merchant-centric.

AI didn’t ignite these strategies. It poured gasoline on them, amplifying what was already true.

The right AI approach becomes obvious when you’re able to use it to accelerate customer surprise, delight, and success, as all four of these products do.

Your Strategic Question

Stop chasing generic AI adoption.

Instead, identify the constraints competitors can’t — or won’t — solve with AI.

That’s where your competitive advantage lies.

If you’re still thinking about constraint removal through AI, you’re six months behind. The winners are asking a better question to unleash strategic constraint inversion:

Within your industry and strategic position, what constraints could become your competitive advantage through AI?

Share