For six weeks I’ve been describing things that are broken, like reporting systems, signals you’re capturing but not acting on, the gaps in your org chart and the processes you’re running blind.

All of that was diagnosis, but this is what the commercial system should actually look like.

Commercial signal moves through five stages: Capture, Structure, Interpret, Synthesize, Report. Every commercial decision your company makes is the output of that chain, whether or not anyone has ever drawn it. Most companies have all five stages present in some form. Almost none of them have the stages connected.

That’s the key item for this post. The stages are rarely the problem. The problem is what happens between them.

The five stages

Capture is the moment a signal enters your company. A rep types a note. A prospect opens an email three times in a day. A support ticket gets filed with an angry tone. Usage drops on a key account. Capture is mostly a solved problem, and your stack is likely already capturing far more than anyone reads.

Structure is turning captured signals into something a system can act on. Free text becomes a field. A transcript becomes a tagged set of topics. Behavior becomes a score. Structure is what makes signals queryable and actionable instead of just stored.

Interpret is asking the structured signal a question. What is different this week? Which accounts moved? What does this pattern mean? Interpretation is where a human or a model reads the data and produces a judgment.

Synthesize is combining judgments across sources into one view. The pipeline, usage and support interpretations become a single picture of the account. Synthesis is the stage that usually lacks ownership.

Report is presenting that picture to whoever decides something. The forecast call. The QBR. The board deck. Report is the only stage most companies have formalized, which is exactly why it’s the stage everyone mistakes for the system itself.

The four handoffs

Signal doesn’t usually die inside a stage. It dies crossing between them. There are four crossings, and each one can fail in a specific, recognizable way.

Capture to Structure

This is the most common failure and the least visible. Signal gets captured and never becomes anything a system can read.

A rep can write a good call note with lots of detail in it: the buyer mentioned a competitor, the timeline moved, procurement got involved earlier than expected. If that note goes into a text field, it’s captured, but it’s now permanently invisible. No dashboard will ever surface it, no forecast will ever weight it, and the only person who will ever read it again is the rep, if they scroll back.

I worked with a creative AI platform where this was happening at volume. Hundreds of calls a quarter, notes taken on nearly all of them, real signal about pricing resistance and packaging confusion sitting in free text. Nobody had ever structured any of it. The signal wasn’t missing. It just wasn’t in a form anything could use.

The tell: your reps are diligent about notes and your systems know nothing.

Structure to Interpret

Signal is structured but nobody asks it a question.

This is the stage where dashboards do the most damage. A dashboard displays structured data. It does not interpret it. Someone still has to look at the chart and decide what it means, and in most companies nobody has that job. The dashboard gets built, it gets checked for a few weeks, and then it becomes wallpaper.

You can spot this in the language people use. When a leader says “we have the data,” they usually mean they have structure. Having the data and having a read on the data are different things, and the gap between them is where most analytics investment fails.

The tell: you can produce a chart for any question and you can’t remember the last time a chart changed a decision.

Interpret to Synthesize

Someone reads the signal, but that interpretation never gets combined with anyone else’s.

This is the failure I see most often in PE-backed companies, and it’s the one that maps directly onto the org chart problem from Week 5. The CSM interprets usage. The AE interprets buyer engagement. Finance interprets billing. Marketing interprets intent. All four reads are correct, but all four are partial. Nobody puts them next to each other until something has already gone wrong.

Synthesis is nobody’s job in most orgs. RevOps owns systems. Sales leadership owns the number. Finance owns reporting. Synthesis falls between all of them and so it happens by accident, in hallway conversations and in Slack, or it doesn’t happen at all.

The tell: after a surprise churn, four different people can each explain what they saw coming, and none of them saw the whole thing.

Synthesize to Report

The synthesis exists and the board deck doesn’t reflect it.

This one is subtle because the report is usually accurate. The numbers can tie out in the slides, but what’s missing is everything that produced the numbers. The board may see that net revenue retention fell four points. What the board doesn’t see is the read from six weeks ago, the one where three people separately noticed the same pattern in three different systems.

The board will ask the obvious question: what happened. And the room answers with a story constructed after the fact, because the actual signal chain was never carried into the room.

The tell: your board deck is a description of outcomes with no visibility upstream.

What a connected chain looks like

Years ago I was the commercial leader at a global media company that didn’t have a CRM. Pipeline lived in spreadsheets, forecasts came from asking people, and the board got a number every quarter that nobody could trace back to anything.

I helped build the chain from zero, which is a strange advantage, because you have to make every stage explicit.

Capture came first, and it was deliberately narrow. A small set of things we agreed actually predicted revenue: deal stage movement, buyer engagement, renewal timing, and a short list of qualitative flags reps could set in one click.

Structure meant those flags were fields, not prose. A rep couldn’t write “client seems nervous about budget.” They had to pick a flag. That constraint annoyed people for about a month and then it became the reason the whole thing worked.

Interpret was a standing weekly read, owned by a named person, not a dashboard anyone could ignore. The question was always the same: what moved, and what does it mean.

Synthesize was a cross-functional effort. Once a week, at the executive meeting, the commercial read, the delivery read and the finance read went into one document. It was a single view of the business, with disagreements left visible rather than smoothed over.

Report then became easy, which surprised me at the time. The board deck stopped being a construction project. It was the synthesis, formatted.

Forecast accuracy roughly doubled and revenue grew 25 percent over the period. It was as simple as connecting five stages that had never been connected, in a company that had no tooling at all.

We didn’t have better software. We had a chain.

Why this is the post everything else hangs on

You now have the spine of Commercial Signal Architecture. Five stages, four handoffs, and a specific failure mode at each crossing.

What you don’t have yet is the map. These five stages don’t run in a single line through your company. They run through four different layers of it, your systems, your people, your processes and your reporting, and each layer fails at the handoffs differently.

The four handoffs tell you what breaks. The grid will tell you where yours breaks.

That’s the next post.

If you read the four handoffs and knew immediately which one is yours, that instinct is usually right, and it’s also usually incomplete. Working out where your chain actually breaks, and in which layer, is specific to your business. I work with a small number of commercial leaders and operators each quarter on exactly that.

CSA Playbook runs weekly through Week 12. Next week: the full map, layers by stages.

Matt Young is a senior commercial operator with exits from Blackstone, Bain Capital, and Thompson Street Capital. He advises PE firms and portfolio companies on revenue architecture through Value Edge LLC.

I write about AI, so I also use AI for editing, the way I’d use a research assistant. Every argument, framework, judgment and client example here is mine and drawn from real work. I’m telling you this because that’s what I tell all the companies I work with: know what’s underneath the signal.

Every revenue org runs the same seven processes: forecast, pipeline review, renewal and expansion, QBR, win/loss, board prep, cross-functional handoffs. These are the load-bearing rituals of the business, the meetings that end up deciding headcount, quota, board confidence, even valuation multiple. And nearly all of them run the same way. Someone pulls last week’s numbers, someone presents what already happened, the room reacts to it, everyone moves on until next time.

All seven are retrospective. They surface what the signal already knew a week ago, a month ago, sometimes a quarter ago, and none of them were built to catch it the moment it actually appeared.

Forecast. Built off stage-weighted pipeline and rep sentiment, both of which lag by definition. The deal that’s about to slip usually showed the signal three weeks earlier. A champion went quiet. A multi-thread contact stopped responding. A competitor’s name showed up in a call transcript that nobody read closely enough. The forecast call catches the slip the week it finally becomes visible. The signal was there the week it started.

Pipeline review. Most of these are a tour of stage and age, when what they should be is a tour of movement and decay. Stage tells you where a deal is supposed to sit. It doesn’t tell you whether anything happened to it this week, and that’s usually the more useful question.

Renewal and expansion. Usage dips. Support ticket tone shifts. A champion changes roles. All of it visible weeks before the renewal conversation ever happens, and all of it sitting in tools nobody’s cross-referencing until the CSM flags a risk two weeks out, by which point it’s a save motion instead of a signal response.

QBR. A retrospective wearing a strategy costume. Last quarter’s numbers, a few slides on what worked, and a forward-looking section that’s mostly educated guesswork because it isn’t actually built on anything that moved in real time.

Win/loss. Usually conducted well after the deal closes, from memory, filtered through whoever’s willing to fill out the survey that week. Whatever actually happened in the deal, live, started degrading the moment it closed, and nobody was in the room to capture it while it was still sharp.

Board prep. A synthesis exercise done under time pressure, built from whatever numbers happen to be assembled that particular week. The board ends up with a snapshot of the business rather than a signal of where it’s heading, and most boards can tell the difference even when nobody says so out loud.

Cross-functional handoffs. Sales to CS, marketing to sales, product to everyone. Each handoff loses information by design, because the function on the receiving end only gets the summary, never the signal underneath it.

Same structural flaw in every one of the seven: built to report on the past, not to route what’s happening right now to the person who could actually act on it.

This is the gap between “we have the data” and “we’re signal-driven.” Most PE-backed companies think they’ve closed it because they bought a BI tool or stood up a dashboard. A dashboard is just a faster way to look at last week. It’s not a faster way to catch this week, which is the part that actually matters.

I saw this most clearly rebuilding the revenue org at a Bain Capital-backed ad intelligence business. The pipeline reviews there looked rigorous on paper, stage by stage, on schedule, nothing obviously wrong with them. What was missing wasn’t rigor. It was routing. The signals that mattered, a stalled multi-thread here, a shift in buyer language there, were already being captured somewhere in the stack. Nobody had built the path from where a signal landed to the person who needed to see it, and by the time anyone did, the deal was already at risk.

The fix isn’t another dashboard, and it isn’t more meetings. It’s re-architecting what these seven processes are actually built on: moving each one from reviewing what happened to responding to what’s moving.

Here’s what that looked like for one of the seven, at that same company.

The pipeline review used to open with a stage-and-age view, same columns every week, same ritual of reps explaining why a given deal was still “in progress.” None of it was wrong exactly. It just wasn’t signal. Stage tells you where a deal is supposed to be. It doesn’t tell you whether anything actually happened to it this week.

So we rebuilt the review around movement instead. Every deal got scored against a small set of signals that mattered more than stage ever did: had the multi-thread contact gone quiet for more than ten days, had the deal sat in the same stage for more than two weeks with no logged activity, had buyer language in the last call shifted toward hesitation or a competitor’s name. Deals that tripped two or more of those got surfaced automatically ahead of the Friday review instead of waiting to be noticed in a CRM view someone might remember to check.

The review itself changed shape as a result. Instead of walking every deal in stage order, the first ten minutes went to the flagged deals, the ones with real signal against them, and the rest of the pipeline got a fast pass. Reps stopped narrating stage and started explaining what had actually moved, or hadn’t, and why.

Same pipeline, same reps, same CRM. What changed was what the process was built to notice.

This is Week 6 of a 12-part series on Commercial Signal Architecture. Next week: the five stages every commercial signal moves through, and where most of it dies in transit.

Matt Young is a senior commercial operator with exits from Blackstone, Bain Capital, and Thompson Street Capital. He advises PE firms and portfolio companies on revenue architecture through Value Edge LLC.

I write about AI, so I also use AI for editing, the way I'd use a research assistant. Every argument, framework, judgment and client example here is mine and drawn from real work. I'm telling you this because that's what I tell all the companies I work with: know what's underneath the signal.

Ask your CEO what the leading indicators of next quarter’s revenue are. Not the lagging ones — not last quarter’s bookings or current pipeline coverage. The leading ones. The signals that, right now, tell you whether the quarter is building or softening before it shows up in the number.

If the answer takes more than thirty seconds and involves a spreadsheet, the signal isn’t reaching the decision.

That’s the test. And in most PE-backed commercial organizations, most CEOs fail it — not because they aren’t capable, but because the signal never made it to them in a usable form. It got lost somewhere in the org chart.

If you ran last week’s audit and scored lower than you expected, this is probably why. Most commercial leaders look at a signal gap and reach for a tool. A new integration. A better dashboard. A cleaner data model. Sometimes that’s right. More often, it isn’t.

The signal breakdown in most PE-backed businesses isn’t in the software. It’s in the org chart. It’s in the structural way each function relates to commercial signal — what they capture, what they ignore, what they hand off, and what dies on their desk.

This week: the human layer. Every function in the commercial org, its role in the signal chain, and the specific gap that breaks it.

The AE

The account executive is the richest signal source in your commercial organization. Every call contains objections, competitor mentions, stakeholder dynamics, budget signals, and timing cues. Most of it never leaves the conversation.

The structural gap isn’t effort. AEs work hard. The gap is that signal capture was never part of the job design. AEs were hired to move deals forward, not to structure intelligence. So they take notes that serve them — enough to prep for the next call — and nothing that serves the system.

The consequence: your most valuable market intelligence is trapped in individual call notes, personal CRM fields filled inconsistently, and the institutional memory of reps who will eventually leave.

When an AE closes a deal, the signal that explains why it closed — the real reason, not the CRM dropdown — walks out the door with them every Friday.

The CSM

Customer success managers sit on top of the richest retention signal in the business. Product adoption rates. Stakeholder sentiment. Renewal risk. Expansion readiness. They feel all of it, in real time, before it shows up in any dashboard.

The structural gap: CSMs were built to manage relationships, not to route signal. Their job is to make customers successful. The intelligence they carry about account health lives in their heads, in their email threads, and in their platform notes — not in a structured form that feeds a renewal model or triggers a commercial motion.

At a Bain Capital-backed business I worked with, the CSM team knew — genuinely knew — which accounts were at risk months before the health scores reflected it. They’d seen the stakeholder change. They’d felt the engagement drop. They’d noticed the champion going quiet. None of that signal was structured. None of it was routed. The health scores stayed green while the accounts went cold.

That’s not a CSM failure. That’s architecture. You can’t expect humans to work around broken structure and simultaneously blame them for not logging it correctly.

RevOps

RevOps is the function best positioned to own signal routing. In most organizations, it doesn’t — not because RevOps lacks capability, but because it has been positioned as a reporting function rather than a routing function.

RevOps builds the dashboards. RevOps runs the forecast calls. RevOps manages CRM hygiene. All of that is downstream work — taking whatever signal the commercial team produced and making it legible. Almost none of it is upstream work: defining what signal should be captured, where it should flow, what should trigger what.

The structural gap: RevOps gets involved after the signal has already been lost. By the time a deal shows up in a pipeline review with incomplete fields and inconsistent staging, the signal that would have made it legible is gone. RevOps is left cleaning up the output of a broken capture process rather than designing the process itself.

The reframe worth making: the highest-leverage thing a RevOps leader can do isn’t a cleaner dashboard. It’s moving upstream — owning signal design before signal capture happens, not after. That shift turns RevOps from a reporting function into the connective tissue of the commercial organization. The right question for your RevOps lead isn’t “how accurate is our forecast?” It’s “what happens to a signal between the moment it’s created and the moment it reaches a decision?”

Sales Leadership

Sales leaders consume signal more than anyone in the commercial organization. Pipeline reviews. Forecast calls. Win/loss conversations. Deal coaching sessions. The entire job is signal interpretation.

The structural gap: most sales leaders are interpreting signal that’s already been filtered twice — once by the rep who decided what to log, and once by the system that decided what to surface. By the time a number lands in a forecast call, it has passed through enough human judgment and system limitation that it may bear only a loose relationship to commercial reality.

The tell is the override. When a sales leader adjusts the called number based on gut feel — because they know Rep A sandbangs and Rep B is optimistic and this particular deal has a CFO who’s slow to sign — they are compensating for a broken signal chain with institutional knowledge. That knowledge isn’t scalable, isn’t transferable, and disappears when they do.

Signal architecture replaces the override with a system. Not to remove judgment — judgment still matters — but to give judgment something accurate to work with.

Marketing

Marketing’s relationship with signal is asymmetric in a specific way: they produce a lot of it and consume almost none of it from downstream.

Marketing generates intent data, engagement rates, content consumption patterns, MQL flows, and campaign attribution. Most of that signal travels in one direction — into sales — and very little commercial signal travels back. Which MQLs actually converted and why. Which content influenced late-stage deals. Which segments are showing up in closed-won versus closed-lost.

The structural gap: marketing optimizes for the metrics it can see, which are almost entirely top-of-funnel. The signal that would tell them whether their ICP targeting is accurate, whether their messaging is landing in qualified conversations, whether their pipeline contribution is real or inflated — that signal lives downstream and rarely makes it back.

At a creative AI platform I worked with, marketing was generating significant MQL volume against a broad ICP. Sales was qualifying out the majority of it. The gap between what marketing thought it was producing and what sales was actually working was never surfaced systematically — it lived in the friction of the handoff, not in a dashboard anyone owned. The ICP tightening that eventually fixed the pipeline required pulling that downstream signal back upstream, which nobody had designed for.

Finance

Finance sees the output of the commercial signal chain more clearly than anyone — revenue, churn, expansion, contraction, all of it lands in the P&L — and has almost no visibility into the signals that explain it.

The structural gap: finance produces the revenue bridge but doesn’t own the commercial narrative behind it. When NRR misses, finance can tell you by how much and in which segment. It cannot tell you whether the miss was a retention problem, a pricing problem, an ICP problem, or a competitive problem — because those signals live in systems finance doesn’t touch.

The consequence shows up in board meetings. Finance presents the number. Commercial leadership is asked to explain it. If the signal infrastructure isn’t there, the explanation is assembled on the fly from institutional memory and rep anecdotes. The board hears narrative, not signal. Confidence erodes.

When commercial and finance are reading from the same signal chain — when the revenue bridge and the commercial commentary are built from the same underlying data — the dynamic in the room changes completely.

The CEO

The CEO is the last stop on the signal chain and the most exposed to its failures.

The structural gap is the most consequential one in this list: the CEO sees metrics, not the signals behind them. NRR of 94%. Pipeline coverage of 3.2x. Win rate of 22%. These numbers arrive in a board deck, stripped of the signal chain that produced them. Whether the NRR reflects a structural retention problem or a one-time cohort issue — the CEO often cannot tell from the metric alone.

This is not a CEO failure. It is an architecture failure. When the signal chain is broken, the metric is all that survives the journey to the top. And a metric without signal context is just a number that requires explanation rather than one that enables decision.

The Pattern Across All of It

Every function in the commercial org has a structural relationship with signal that was never designed — it evolved. AEs capture what helps them sell. CSMs capture what helps them retain. RevOps reports what it can see. Marketing optimizes what it can measure. Finance reconciles what lands. The CEO reads what survives.

Nobody designed the chain. Nobody owns the handoffs. Nobody is accountable for what happens to a signal between the moment it’s created and the moment it reaches a decision.

That’s the human layer gap. And no software fixes it, because it isn’t a software problem. It’s an accountability problem — and accountability problems require structural solutions.

The restructuring question — what this actually looks like at a $20M business versus a $100M one, and how to sequence it without a data team or a six-month implementation — is where this series goes in Month 3. The architecture comes first.

If your org chart looks familiar in these descriptions — if you recognize your CSM team, your RevOps function, or your sales leader in any of these gaps — the signal chain work starts with the humans, not the tools. I work with a small number of commercial leaders and operators each quarter on exactly this. No deck. Just the diagnostic and the work that follows it. You know where to find me.

CSA Playbook runs weekly through Week 12. Next week: seven revenue processes that are running blind.

Matt Young is a senior commercial operator with exits from Blackstone, Bain Capital, and Thompson Street Capital. He advises PE firms and portfolio companies on revenue architecture through Value Edge LLC.

If you’ve been following this series, you know the framework. You know where revenue decisions break down — not in the boardroom, not in the QBR, but upstream, in the gaps between capture and synthesis.

This week isn’t theory. It’s a diagnostic.

Below are ten yes or no questions. Run this yourself, run it with your RevOps lead and run it before your next board prep. The score at the bottom tells you where your signal chain is breaking — and how badly.

The Audit

Answer yes or no. Be honest. “Mostly” counts as no.

Question 1

Do you have a single customer ID that ties together your CRM, your product data, and your billing system?

Not three systems that sort of overlap. One ID. One record. One source of truth that a revenue leader can query without asking two people to reconcile a spreadsheet first.

If your answer is no, you are not running a signal architecture. You are running three separate opinions about the same customer.

Most commercial teams don’t fail at analysis. They fail at unification. The signal exists, it’s just living in three places that don’t talk.

Question 2

Is your MQL definition in writing, agreed upon by both marketing and sales, and reviewed in the last 12 months?

This is not the definition that exists in someone’s head. Not the one from the last go-to-market refresh that sales never actually bought into. The written version, with sign-off, current as of this fiscal year.

If the answer is no, you don’t have a lead scoring model. You have a handoff dispute waiting to happen.

At a Bain Capital-backed SaaS business I worked with, MQL definition had drifted so far from what sales actually qualified that pipeline coverage looked healthy until it didn’t. The fix wasn’t a new scoring model — it was a single conversation with a written output and a committed owner.

Question 3

Are more than 60% of your CRM contact and account records complete on the fields your team actually uses to prioritize?

Not all fields. The fields that drive action: segment, ARR tier, product line, renewal date, last meaningful engagement.

Below 60% field completion, your pipeline reports are telling you what your reps entered, not what’s true.

The consequence isn’t a data quality problem — it’s a forecast problem. When the fields that drive prioritization are incomplete, your pipeline reviews become a conversation about what reps think is happening, not what is. That gap shows up in the number at the end of the quarter. This number is almost always worse than people expect. Pull it before you answer.

Question 4

When a deal goes dark, do you know within 72 hours — from a system signal, not a rep update?

Behavioral silence is a signal. Declining login frequency is a signal. No response to the last three outreach touches is a signal. The question is whether your stack surfaces it automatically or whether you find out on the Friday call when the rep says “they’ve gone quiet.”

When you find out late, you didn’t lose the deal in that moment — you lost it two weeks ago when the silence started and nobody saw it. The 72-hour window is the difference between a save play and a post-mortem.

This is also where AI can earn its keep. Automated silence detection, behavioral pattern recognition, multi-signal churn indicators — the capability exists and it works. But it requires clean, unified behavioral data as the input. Build the input layer first. Then deploy the model. That’s the sequence.

Question 5

Is your renewal motion triggered by customer health signals, or by calendar dates?

This is the question that separates companies that retain revenue from companies that recover it (sometimes).

Calendar-triggered renewal means you’re reacting to time. Signal-triggered renewal means you’re reacting to behavior. One of those is a retention strategy. The other is a hope.

If your CSM team’s 90-day renewal playbook starts on day 90, you are already late for a meaningful percentage of your book.

AI-powered health scoring makes signal-triggered renewal systematic rather than heroic. The companies running it well aren’t doing more CS work — they’re doing it earlier, on the right accounts, because the model tells them where to look. Getting there requires the underlying signal to be structured and unified first. That’s not a caveat. That’s the build sequence.

Question 6

Do you know, right now, which of your accounts in the top two revenue tiers have had zero meaningful engagement in the last 30 days?

Not “we have health scores.” Not “we do EBRs quarterly.” Do you know, today, which logos are quiet — and does someone own that list with a specific action attached?

At Motorsport Network, we had no CRM when I arrived. None. We built the signal infrastructure from zero. The first thing that list told us — once we could actually see it — was that a third of our top accounts hadn’t had a meaningful commercial touch in over 60 days. That’s not a sales problem. That’s a visibility problem.

Question 7

Can you produce a revenue bridge — showing new logo, expansion, contraction, and churn — without a multi-day data pull?

Revenue bridges are not finance documents. They are commercial accountability documents. If you can’t build yours in a day, your signal infrastructure has a synthesis problem.

The downstream cost is real: when commercial leaders can’t produce the bridge independently, they walk into board meetings reactive. Finance sets the frame, asks the questions, and the CRO spends the meeting explaining history instead of defending forward. That’s a credibility problem that compounds.

Question 8

Is your ICP definition specific enough to disqualify — or does it describe almost every company you’d ever want to sell to?

An ICP that doesn’t disqualify isn’t a profile. It’s a wish list.

At an AI-native fintech I worked with, the ICP rebuild was the first thing we did before touching pipeline. We went from a universe of roughly 2,000 firms to a scored, tiered list of 350. Pipeline tripled within two quarters — not because we worked harder, but because signal was concentrated on accounts where the pattern matched.

Specificity is not a constraint on growth. It is the precondition for it.

Question 9

Do your top three leading indicators of revenue conversion have an owner, a tracking cadence, and a defined intervention threshold?

Leading indicators without owners are decorations. If you track pipeline velocity, product activation rate, and multi-threaded engagement but nobody is accountable for moving them, you have a dashboard, not a system.

Question 10

Does your board or PE sponsor receive signal-based revenue commentary — or activity-based commentary?

There is a difference between “we ran 47 discovery calls and added $2.3M to pipeline” and “conversion rate on our top ICP segment dropped 8 points and here’s what we’re doing about it.”

One describes effort. One describes the business. The further you are from the second version, the more your reporting is working against you.

What the audit doesn’t tell you

The ten questions above tell you where the gap is. They don’t tell you why it exists or how to close it — because that part is specific to you.

The fix for a broken customer ID is not the same at a $15M SaaS business as it is at a $150M multi-product platform. The right renewal signal threshold depends on your churn pattern, your contract structure, your CS capacity, and how your product is actually used. The ICP rebuild at an AI-native fintech looks nothing like the same exercise at a B2B media company. This is where generic frameworks stop being useful.

If your score is 6 or below, the gaps in your audit aren’t a configuration problem. They’re a structural problem — and structural problems need to be diagnosed before they’re fixed.

The AI layer gets its own dedicated post later in this series — specifically, which nodes in your commercial stack AI should touch first, and where most companies are deploying it at the wrong layer entirely. The short version: if your audit score is below 6, AI is not your next move. Architecture is.

CSA Playbook runs weekly through Week 12. Next week: your tech stack probably isn’t the problem — your signal routing is.

Matt Young is the founder of AI Value Edge and a senior commercial operator with two decades of experience scaling revenue inside PE-backed and sponsor-supported companies. He publishes AI Value Edge — The Growth Multiple on Substack.

In early April, I wrote about OpenAI’s pitch to PE firms: a 17.5% guaranteed return structure designed to lock in portfolio-wide AI adoption from the GP level down. My argument then was that the mandate would arrive before the architecture was ready — and that operators would be left holding the gap.

Six weeks later, Anthropic did something more significant.

On May 4th, Anthropic announced a $1.5 billion joint venture with Blackstone, Hellman & Friedman, and Goldman Sachs — also backed by Apollo Global Management, General Atlantic, Leonard Green, GIC, and Sequoia. The structure is explicit: embed Anthropic engineers directly inside mid-sized portfolio companies to redesign workflows and integrate Claude into core operations. Not advise or license. Embed.

Blackstone’s Jon Gray said: “We believe it can help break down one of the most significant bottlenecks to enterprise AI adoption by expanding the number of highly skilled implementation partners.” Goldman’s Marc Nachmann added: “Having the model alone doesn’t change your workflows or how you operate. You need people who can combine the technology with what’s actually happening in the business and implement those changes.”

Neither of them is talking about revenue. They’re talking about implementation. That distinction matters more than it might seem.

What This JV Is Actually Saying

Anthropic CFO Krishna Rao framed the supply-side logic: “Enterprise demand for Claude is significantly outpacing any single delivery model.”

Read that carefully. The constraint isn’t model access. It isn’t budget. It’s the ability to put the technology to work inside a real business.

The smartest capital allocators on the planet just underwrote $1.5 billion to solve an implementation problem — not a technology problem. For anyone who has been watching PE’s AI narrative over the past 18 months, that’s a meaningful signal. The conversation has shifted. The question is no longer “which AI” — it’s “who can make AI actually work inside the business.”

That’s a different question, and it has a different answer.

What the JV Solves — and What It Doesn’t

The forward-deployed engineer model is seriously powerful. It’s the same logic Palantir used to entrench itself inside government and defense — send smart people in, learn the workflows, build deep, be hard to remove. Anthropic is applying that playbook to the mid-market PE stack, and it will work.

What it solves: workflow integration. Process redesign. Technical implementation at the operational layer.

What it doesn’t solve: the commercial signal layer.

Workflow efficiency and revenue signal are not the same problem. A board can have every internal process redesigned around AI — procurement, finance, HR, operations — and still be blind to what the pipeline is actually telling them. The engineers who show up will make the business run better. They won’t tell you which deals are quietly dying, which customers are 90 days from churning, or where pricing is leaving money on the table.

Those aren’t engineering questions. They’re commercial judgment questions — and they sit at the intersection of operator experience and signal architecture. No workflow redesign produces them. No implementation team is scoped to answer them.

That layer is what’s still missing from this equation. And for a GP trying to understand why EBITDA isn’t moving after the AI investment, it’s the layer that matters most.

The Sequencing Problem

The Anthropic/Blackstone model embeds engineers to redesign workflows around AI. That work happens at the infrastructure and operational layer — and it’s valuable.

But there’s a layer above it. One that determines which commercial signals — from which data sources, in what combination, with what prioritization logic — should drive which decisions, on what timeline, with what escalation path to the board.

You can’t build that layer without operational context. But operational context alone doesn’t produce it. It requires someone who has sat inside a P&L, watched what actually moves a revenue number, and designed the connective architecture between the data that exists and the decisions that need to get made.

The JV gets you layer one. Layer two is a different engagement — and a different kind of operator.

For PE-backed companies with a four-year hold period and ambitious EBITDA targets, layer two is where the growth multiple lives.

Why This Matters Now

The firms behind this JV — Blackstone, H&F, Goldman, Apollo, GA — collectively own hundreds of portfolio companies. As the venture scales, those companies will get engineers. Workflows will get redesigned. AI will get integrated.

And then the board will ask the question this JV isn’t built to answer: the efficiency work is underway — why aren’t the commercial outcomes moving?

I’ve been inside that room. The tools are live. The workflows are cleaner. The revenue line hasn’t moved. What’s missing isn’t more implementation. It’s the architecture that converts operational AI into commercial signal.

The implementation layer is being built. The commercial signal layer still needs an operator.

Matt Young is the founder of AI Value Edge and a senior commercial operator with two decades of experience scaling revenue inside PE-backed and sponsor-supported companies. He publishes AI Value Edge — The Growth Multiple on Substack.

At a Bain Capital-backed intelligence business, we ran a full audit of the commercial stack the year before it exited. CRM, conversation intelligence, a customer success platform, BI tools. Every category had AI in it already. Lead scoring. Call summarization. Churn prediction. Automated dashboards.

By any vendor’s definition, we were “AI-enabled.” Every renewal conversation with every vendor said so.

None of it was preventing the thing that was actually costing us deals.

Every category had AI. None of it was routing.

That’s the distinction this series keeps coming back to, and it’s the one most stack audits miss. Lead scoring tells you a deal is at risk. It doesn’t tell the AE that the prospect’s tone shifted on a call four days ago. Call summarization gives you a clean transcript. It doesn’t flag that pricing objection to the person who owns packaging. Churn prediction gives CS a red flag in a dashboard nobody opens until the QBR.

AI was doing real work in every one of these tools. It was just doing the work at the wrong end of the pipe — scoring, summarizing, predicting — and stopping exactly where it mattered most: deciding who needs to see this, right now, and putting it in front of them.

Why this keeps happening

Vendors build AI into the part of the workflow they own, not the part that crosses functional lines. A CRM company builds AI for the CRM. A conversation intelligence company builds AI for calls. Nobody’s incentivized to build the connective layer, because the connective layer isn’t a product — it’s an architecture decision specific to your business.

That’s the same five-stage problem from Week 1: Capture, Structure, Interpret, Synthesize, Report. Most AI in commercial stacks today lives entirely inside Structure and Interpret — turning raw signal into something legible. Almost none of it touches Synthesize, which is where a structured signal becomes a routed action: this person, this trigger, this moment.

What changed wasn’t a new tool — it was where the routing decision got made

We didn’t replace the stack at the Bain Cap intelligence business. CRM stayed. Conversation intelligence stayed. What changed was the layer that decided which structured signal went to which human, on what trigger and with what urgency. That layer didn’t come from a vendor.

The difference showed up as speed, not features. A pricing objection that used to surface at the next QBR — if it surfaced at all — started reaching the person who owned packaging within days. A churn signal that used to sit in a CS dashboard until renewal season started triggering action while there was still time to act on it. Same tools. Same data. The only thing that moved was how fast a structured signal found the person who could do something with it.

Most companies have no idea how their AI signals are turning into decisions. That’s usually the first sign the architecture, not the tooling, is the gap.

The specific routing logic — what triggered what, who owned which handoff, how it was built — depends entirely on your stack, your team structure, and where in your five stages signal is currently dying. That’s the diagnostic Week 2 started and Week 4 finishes.

That’s also the part I work through directly with portfolio companies. The principle in this series is universal. The mechanics are specific to you.

What this means for week 4

If Week 2 was the signal inventory, Week 3 was the diagnosis — AI is everywhere in your stack and nowhere in your routing. Week 4 is the audit: a framework for scoring your own stack against the five stages, so you can see exactly where your signal is dying before you spend another dollar on a tool that's AI-enabled in the wrong place.

At a creator AI platform where I advised the commercial leadership team, every sales call was recorded. Conversation intelligence was live — transcripts, recordings, searchable libraries. By any audit of “do we have the tooling,” the answer was yes.

But nobody was reading them for signal.

Buried in those calls was exactly the information that mattered most: which product in the portfolio prospects actually leaned into during the pitch, which pricing language landed and which made people go quiet, and where in the conversation deals started slowing down. That information existed. It was captured, word for word, in a searchable database. And it was having zero effect on how the business packaged, priced, or sold.

Once we started structuring that signal — pulling patterns across calls instead of treating each one as a one-off — pricing and packaging got sharper, because we were building it around language and reactions visible across dozens of conversations instead of guessing. Deal velocity improved, because the friction points showing up call after call stopped being a surprise in week six of a sales cycle and started getting addressed in week one.

The tool was never the gap. The gap was that nobody had built the bridge from “we recorded it” to “we acted on it.”

The four categories of raw signal

Every commercial organization is sitting on four buckets of raw signal. Most are heavily captured and almost entirely unread. Here’s the map.

1. Customer Health Signals

These are the signals that tell you whether a customer relationship is strengthening or eroding — usually long before it shows up in a renewal conversation.

• Executive sponsor goes dark (no response across multiple channels)

• Champion changes jobs or is no longer in meetings

• Usage or login frequency drops without explanation

• Support ticket volume or severity spikes

• NPS or CSAT score declines between touchpoints

• Renewal conversation gets pushed or rescheduled by the customer

The version of this at the creator AI platform: a prospect’s tone shifting mid-call when a specific product was mentioned. That’s a health signal about fit — and it was sitting in a transcript, not a CRM field, which is exactly why it never made it anywhere.

2. Pipeline Signals

These are signals about deal health that exist well before a deal is marked “at risk” in the CRM.

• Deal sits in the same stage for longer than the historical average

• Multiple stakeholders stop attending calls

• Pricing objections repeat across multiple conversations

• Champion can’t get internal budget approval scheduled

• Competitive mentions increase late in the cycle

• Deal slips immediately after being marked “commit”

The pricing objection pattern is the one that hit hardest. The same friction point was showing up call after call — not because every prospect was identical, but because the packaging genuinely didn’t match how people wanted to buy. Each rep experienced it as “this one deal has a pricing issue.” Nobody saw it as a pattern until the calls were read as a set.

3. Market Signals

These are signals from outside your customer base that indicate where demand is shifting.

• Competitor pricing or packaging changes

• New entrants targeting your ICP

• Analyst or press coverage shifting category language

• Win/loss patterns shifting by segment or vertical

• Inbound inquiry mix changing (different titles, different company sizes)

• Category-defining terminology changing in calls and RFPs

Market signal is the category most companies treat as someone else’s job — usually marketing’s, or nobody’s. But it’s often sitting in the same conversation data as everything else. A prospect using a competitor’s term for your category, unprompted, is market signal. It almost never gets flagged as such.

4. Product / Usage Signals

These are signals from how customers actually use what they’ve bought — the clearest leading indicator of both expansion and churn.

• Feature adoption stalls after onboarding

• Usage concentrated in a narrow set of features (low expansion surface)

• Seat utilization climbing without a corresponding upsell conversation

• New use cases emerging that weren’t part of the original sale

• API or integration usage signaling deeper technical dependency

• Usage patterns diverging sharply between similar accounts

This is the category most disconnected from sales and CS conversations entirely. The richest version of “what does this customer actually need next” often lives in product data that the commercial team never sees — and the richest version of “why are they about to leave” often lives in a transcript that the product team never sees.

Why these signals die before anyone sees them

It’s not that companies lack the data. Walk through the list above and most organizations are capturing 15 of these 20 signals in some tool, somewhere. Calls are recorded. CRM fields exist. Usage data is logged. Support tickets are ticketed.

The reason they die is structural, and it’s the same reason from Week 1: signal has to survive five stages — Capture, Structure, Interpret, Synthesize, Report — and almost everything stops at Capture.

A recorded call is captured signal. It becomes useful only once someone or something structures it — extracts the pricing objection, tags the product mentioned, flags the tone shift. Most organizations have no one whose job is structuring. CS owns health signals but doesn’t talk to product. Sales owns pipeline signals but doesn’t talk to marketing. Product owns usage signals but rarely sits in on sales calls. Each function captures its slice and stops.

What changed wasn’t a mystery — but it wasn’t free, either

The shift from “signal exists” to “signal gets acted on” involved real changes: who saw what, when, and what triggered a response. That’s not a secret. It’s also not something that fits in a Substack post — the specifics depend entirely on your stack, your team structure, and where your signal is actually dying.

That’s the part I work through directly with portfolio companies. The principle in this series is universal. The mechanics are specific to you.

Marc Benioff said the quiet part out loud last year. Salesforce cut its customer support headcount nearly in half — from 9,000 to 5,000 — because AI agents made the structure redundant. Most people read that as a story about job cuts, but it wasn’t. It was a story about a company that realized the operating model underneath the headcount no longer made sense. The jobs didn’t disappear because the work went away. They disappeared because the layer built to do that work — the buffer between signal and action — had been eliminated.

Most commercial organizations are sitting on the same realization and haven’t made it yet.

The Model That Made Sense When the Work Was Human

The baton-pass model has organized B2B commercial teams for twenty-five years. Marketing generates leads. SDRs qualify them. AEs close them. CSMs retain them. Each team owns its section of the track, passes the baton, and measures its own leg.

It wasn’t a bad design. It was the right design for a specific constraint: the person who knew the customer best — the AE — didn’t have the capacity to work raw volume. So you built a layer to filter it for them. SDRs existed to protect AE time. The handoff existed because the AE couldn’t see everything.

That constraint is gone, and the structure built around it has become the liability.

What I Watched Happen at the Handoff

I’ve run commercial organizations across seven M&A transactions, multiple PE-backed SaaS businesses, and a $1B+ media P&L. The place revenue reliably dies — not because the product is wrong, not because the market isn’t there — is at the handoff.

At one company, we had strong top-of-funnel. Marketing was generating volume, scoring looked right, the SDR team was working the queue. But conversion from SDR-touched to AE-advanced was collapsing. When we pulled the data, the problem wasn’t lead quality and it wasn’t rep effort. It was simpler and more structural than that: by the time a lead got scored, queued, assigned, and touched by an SDR, it arrived at the AE stripped of context and cold.

The SDR hadn’t done anything wrong. The handoff had done exactly what it was designed to do — filter and pass. But the AE is the one who knows the customer. The AE is the one who can read a signal and know whether it’s noise or an opening. Every step between the signal and the AE was a step away from the person best positioned to act on it.

We collapsed the qualification layer and put signal directly in front of AEs. Conversion recovered.

Nothing about the product or the market changed. The logic of who owned the signal did.

The Numbers Make It Visible

MQL-to-SQL conversion fell 24% between 2024 and 2026, with the biggest drop at the handoff stage itself — sales teams rejecting more marketing-passed leads as low-intent. More than half of all MQLs still go uncontacted past the 24-hour mark.

That’s not a lead quality problem. It’s a structural one. Signal is arriving faster than the handoff process was designed to move — because it was designed for human throughput, not machine-generated volume.

The org chart is starting to reflect this whether leaders redesign it or not. AE headcount grew 32% last year. SDR headcount grew 3%. That’s not SDRs being automated out of existence — it’s the market redistributing ownership. More of the commercial work is consolidating around the people who have customer context, and less of it is living in the filtering layer between signal and those people.

The Question Nobody Is Asking

Every commercial leader I talk to is asking: how do we use AI in our funnel? The right question is different: does a funnel built around protecting AE time still make sense when AI can do that job?

That’s not a technology question. It’s an organizational one — and it’s harder, because the answers touch comp plans, role definitions, and headcount ratios that nobody wants to redesign mid-growth-plan.

But the companies that ask it now will look structurally different from the ones that bolted AI onto a 2015 org chart and wondered why conversion didn’t improve. Faster handoffs inside a broken structure don’t fix the structure. They just surface the problem at higher volume.

The AE always knew the customer. The only question was whether the organization was designed to let them act on it.

Now it can be.

Matt Young is the founder of AI Value Edge and a senior commercial operator with two decades of experience scaling revenue inside PE-backed and sponsor-supported companies. He publishes AI Value Edge — The Growth Multiple on Substack.

This is Week 1 of a 12-week serialized framework. New installment every week.

When I joined Motorsport Network as CRO in 2019, there was no CRM. Not a broken one. Not an underused one. None. A ~$100M ARR business running on spreadsheets, tribal knowledge, and weekly calls where the forecast was whatever the senior rep said it was with enough conviction.

I built one and stood it up across 15 offices, unified the adtech stack, implemented KPIs that didn’t exist before. Within twelve months, forecast accuracy doubled. Revenue grew 25%. Not because we hired better salespeople. Not because the market got easier. Because for the first time, the business could actually see what was happening in its pipeline before the quarter closed.

That experience is what CSA is built on. The insight wasn’t that Motorsport needed a CRM. It was that a $100M business had been running blind — and the moment it could read its own signals, commercial performance changed materially and fast.

The difference between a reporting system and a revenue operating system

A reporting system tells you what happened. It takes outputs — ARR, NRR, churn rate, pipeline coverage — and displays them. It’s backward-looking by design. It answers the board’s questions after the quarter closes.

A revenue operating system tells you what’s about to happen. It reads signals — behavioral, conversational, financial, product-usage — before they become outcomes. It creates conditions to intervene when intervention is still possible. And critically: it doesn’t require new tools. It requires a different architecture layered over the tools you already have.

What running blind looks like at a $100M ARR portco

What happens when you build the operating system

Here’s what the Motorsport numbers looked like before and after the signal infrastructure was in place.

I saw the same dynamic at a different scale at a Bain Capital-backed ad intelligence business where I served as CCO. The total commercial result was 52% YoY revenue growth. But the part I can specifically attribute to rebuilding segmentation and pipeline architecture — getting the business aimed at the right accounts with the right signal about who was actually converting — was roughly 10 points of that growth. In a $30M+ EBITDA business, 10 points is not a rounding error.

And earlier this year, an AI-native fintech with strong product-market fit but a pipeline that wasn’t moving. The issue: effort distributed evenly across a prospect universe regardless of conversion probability. No ICP framework. No tiering. No signal about where to aim.

What Commercial Signal Architecture actually is

CSA is the operating model that connects raw commercial signals to board-level outcomes. It runs across four layers: 1) the systems that capture signals; 2) the humans who own them; 3) the processes that act on them; and 4) the reporting that makes them visible to leadership and investors.

It is not a tool recommendation. It is not a RevOps engagement. It is not a twelve-week project after which you hand over a deck and leave. It is the connective tissue between your commercial stack and your commercial results — and once it’s running, it changes how the company operates permanently, not just how it reports this quarter.

Every piece of commercial signal — a deal going quiet, a champion changing jobs, a support ticket spike, a seat utilization drop — has to survive five stages before it can influence a decision:

What this series covers

Over the next eleven weeks I’m publishing the complete CSA framework — built and deployed across Motorsport Network (GMF Capital), Vivvix (Bain Capital), Recurrent Ventures (Blackstone), and BrightRoll (acquired by Yahoo).

Each week covers one layer or one application: raw signals and how to read them, the systems layer and where AI actually belongs in it, the human layer and why signal breaks down in your org chart, process design, the board questions CSA is built to answer, and implementation sequencing for PE-backed companies without a data team. Specific throughout — named signal types, named gap patterns, real numbers where I can use them.

One question before next week

If the answer is “we saw it in the data after it happened,” you have a reporting system, not an operating system. That’s what this series is for.

A few years ago I was running commercial operations at a PE-backed company scaling aggressively through acquisition. Leadership brought in a well-known technology consultancy to evaluate our revenue tech stack — assess what we had, identify the gaps, and recommend the path forward.

The assessment was thorough and the recommendation pointed, almost entirely, toward outsourcing our core revenue tech infrastructure to a third-party platform — one the consultancy happened to implement and support as their primary business.

We followed the recommendation. Nine months later we were ripping it out.

The technology was good, but the solution had been designed around what they could sell us, not what our business actually needed. By the time we understood that, we’d lost time, money, and competitive ground we couldn’t get back.

That experience changed how I think about vendor engagements. Watching the AI transformation wave hit PE-backed companies right now, I’m seeing the exact same playbook — just with a larger number on the invoice and a more urgent pitch attached.

The FOMO is real. The emergency is not.

Every major AI vendor’s sales motion starts the same way: here’s what your competitors are doing, here’s the window closing, here’s what missing out costs you. For PE firms, this lands hard. You answer to LPs. You have hold periods. You have comp tied to outcomes. Showing AI progress feels urgent even when the underlying opportunity isn’t fully understood.

The irony is that the firms moving fastest on broad AI deployments are often accumulating the most technical and commercial debt. I’ve watched companies invest heavily in custom AI infrastructure their teams couldn’t maintain six months after the consultants left. I’ve watched PE-backed businesses deploy AI-powered commercial platforms on top of CRM data so degraded the tools produced worse outcomes than the spreadsheets they replaced.

An operating partner colleague of mine recently mentioned a food manufacturer that just built a state-of-the-art production facility and installed 300 AI-enabled cameras for safety monitoring — because that’s what everyone is doing right now. When the CIO went to deploy the AI safety software, not a single vendor had a compatible solution for their camera system. Hundreds of thousands of dollars in hardware. Zero usable AI capability. That’s technology debt in its purest form — not from moving too slow, but from moving too fast without asking the right questions first.

Being deliberate is not the same as being behind. The firms that win here will be the ones who defined the opportunity before the vendor did.

The fox-and-hen-house problem

There’s a growing category of AI transformation firms that will assess your company, design the solution, implement it, and provide the proprietary tools — all in one engagement. This isn’t malicious. It’s structural. Their business model produces a certain predictable answer.

The assessment always finds the opportunity that fits their implementation capability. The roadmap requires their platform. The business case justifies the engagement size.

Nobody got fired for hiring IBM. But I’ve sat across the table from portfolio company CEOs who spent seven figures on engagements that delivered a deck, a dependency, and a renewal conversation. I’ve inherited commercial tech stacks at PE-backed companies where three overlapping vendor relationships were all billing monthly — none of them talking to each other, none of them owned internally, all of them originated from a single “independent” assessment.

The problem shows up everywhere: back office, GTM, pricing, customer success, channel management. The fox-and-hen-house dynamic doesn’t care which department you’re in. What’s missing is an impartial voice asking the only question that matters: is this the right path, for this problem, at this company, right now?

Three paths. And most companies need more than one.

A single company will typically find multiple AI opportunities pointing to different paths simultaneously. The mistake isn’t choosing the wrong path. It’s letting a single vendor map all of them — when they only operate in one.

After operating across PE-backed companies in media, SaaS, and distribution — and watching AI transformation efforts succeed and fail — I’ve come to believe every opportunity maps to one of three paths.

Custom and agentic. You build internal AI agents for specific high-value workflows. Highest upside, highest execution risk. Requires internal talent to own and maintain it post-engagement. Right for proprietary commercial processes — pricing logic, customer segmentation, competitive intelligence — where the IP matters and you have the bench to sustain it. At one PE-backed data and intelligence business I ran, we had the underlying data and the internal team to build custom signal architecture that became a genuine competitive moat. That worked. The same approach at a company without a data science function and a fragmented CRM would have been an expensive disaster.

Platform and off-the-shelf. You buy a fit-for-purpose tool — finance automation, CRM intelligence, sales enablement, revenue analytics. Faster time to value, lower build risk. The real risks are implementation quality and vendor durability. I’ve seen this path work exceptionally well when the functional problem is clearly defined and the tool has a proven ROI profile. I’ve also watched companies sign annual contracts for platforms their teams never adopted because no one owned the rollout once the vendor left. The tool isn’t the transformation. Adoption is.

AI-powered BPO. You outsource the function to a provider who has already made the AI investment. You capture 75–80% of the value with transferred execution risk. Most people think BPO means call centers and invoice processing. It doesn’t anymore. AI-native BPO is moving fast into GTM functions — long-tail customer coverage, inside sales, quote management, channel partner support, customer success for lower-tier accounts. For a large industrial or distribution business serving hundreds of thousands of customers across hundreds of locations, the bottom tier of the customer base may never get meaningful rep coverage. An AI-powered BPO covering that segment is cheaper, faster, and more scalable than hiring and training inside reps who will churn anyway. The question isn’t whether to outsource — it’s whether the function is a commodity or a differentiator.

The critical question isn’t which path is best in the abstract. It’s which path is right for this function, at this company, at this stage of maturity — and who is asking that question with no financial stake in the answer.

What good governance actually looks like

Separate the assessment from the implementation. Structurally different vendors, structurally different incentives. This is not a new idea — it’s how PE has always approached legal counsel, M&A diligence, and financial audits. AI made everyone forget the basics.

Consider going a little old school on the selection process. Bring an independent advisor in early — before any vendor conversation — to help scope the opportunity, define evaluation criteria, and run a structured process. Send out an RFP. Score presenters against specific criteria. Make the final decision defensible with a paper trail, not just a gut call after five vendor dinners. PE firms have historically been allergic to this kind of process — the logic being that 90 days of formal diligence costs you 90 days of value creation inside a four-year hold. That tradeoff made sense when you were picking an audit partner. It doesn’t hold when you’re making an AI infrastructure decision that will be nearly impossible to unwind if you get it wrong.

Define the revenue or EBITDA outcome before the engagement is designed, not after. At every company where I’ve run commercial transformation, the discipline of agreeing on the number upfront — and holding everyone in the room accountable to it — was the single biggest predictor of whether the work actually landed. Vendors who resist that conversation are telling you something important.

Map every opportunity to one of the three buckets before any vendor conversation begins. Then bring in fit-for-purpose partners for each. The firm you’d use for custom agentic development is not the firm you’d use for a BPO transition. Treating them as interchangeable is how you end up with a nine-figure opportunity map that requires one firm to execute the entire thing.

Put someone in the room whose only job is to keep the outcome honest — not the PE firm, not the vendor, not the portfolio company CEO who doesn’t want to push back on the owners. Someone who will say out loud: this isn’t the right path. It’s important that they have no financial interest in being wrong.

Consider the exposure when it goes wrong — because some of it will. PE firms actively directing vendor selection and pushing portfolio companies to move fast are quietly eroding the corporate veil. If an engagement fails to deliver and there’s no formal process showing the portfolio company CEO made an informed, independent decision, the sponsor owns the outcome. That’s not just a reputational problem. In an environment where LP scrutiny is rising and AI investment is accelerating, it’s a liability most firms aren’t thinking about. A structured selection process, led by an impartial advisor, isn’t bureaucracy. It’s protection.

The window isn’t closing

The firms that win the AI value creation game won’t be the ones who moved fastest. They’ll be the ones who asked the right questions before the vendor got comfortable.

Your leverage — to define scope, demand accountability, and pick the right path — disappears the moment urgency replaces judgment. The emergency your vendor is describing is real for them. It doesn’t have to be real for you.

Who in your portfolio engagements has no financial interest in being wrong?

Matt Young is the founder of AI Value Edge and a senior commercial operator with two decades of experience scaling revenue inside PE-backed and sponsor-supported companies. He publishes AI Value Edge — The Growth Multiple on Substack.

There’s a version of this story that pretty much every PE-backed services company tells itself.

We have proprietary expertise. We’re going to productize it. We’ll build a platform, scale delivery, reduce headcount dependency, expand margins. In three years we’ll look less like a services business and more like a SaaS company. The multiple will follow.

It’s a compelling narrative. And it fails, partially or completely, more often than anyone in the room wants to admit.

Not because the thesis is wrong. Because the GTM never gets rebuilt to match it.

The Asset Most Services Companies Don’t Know They Have

Every services business — consulting firm, research house, specialized agency — is sitting on something more valuable than its deliverables.

It’s sitting on a proprietary data asset, built one engagement at a time.

Every client interaction generates signal that doesn’t exist anywhere else. The pattern of questions a buyer asks before they’re ready to commit. The objections that surface in a specific vertical. The competitive intelligence that only emerges inside a live engagement. The analytical framework a senior practitioner applies, almost automatically, that took fifteen years to develop.

Most services businesses treat this as a byproduct of delivery. The best ones eventually recognize it as the core asset.

The transition from services to tech-enabled services is, at its core, the decision to stop treating that asset as a byproduct and start treating it as a product.

But what most companies miss: the asset only becomes a product if you designed the capture layer from the beginning. And most didn’t.

I’ve walked into more than one company where the most valuable competitive intelligence in the building lived in a senior partner’s memory and nowhere else. When that person left, the asset walked out with them. No system. No structure. No institutional signal — just institutional memory that was never made durable.

Three Stages. Three Different GTM Motions. Most Companies Run One.

The mistake I’ve watched repeatedly — from the revenue seat, inside multiple PE-backed companies navigating this exact transition — is treating the services-to-tech evolution as a single GTM motion with increasing automation layered on top.

It isn’t. It’s three distinct stages, each with a different buyer, a different sale, and a different pricing model. Companies that try to run all three simultaneously with the same team, the same motion, and the same compensation structure reliably stall.

Stage One: The Expertise Sale.

This is the core services business. High-touch, relationship-driven, sold by senior practitioners. The buyer is a C-suite executive or a senior decision-maker who is purchasing judgment, not output. Pricing is retainer or project-based. The engagement is bespoke.

What most companies don’t do at this stage — and should — is treat every engagement as a structured data capture event. Not just delivery. Not just the work product. The signal underneath it: what questions came up, what the buyer cared most about, what patterns emerged across engagements in that sector. This is where the proprietary asset gets built. If you’re not capturing it systematically, the transition to Stage Two will be harder than it needs to be.

Stage Two: The Platform Sale.

The proprietary signal gets packaged. The insights that used to require a senior practitioner to produce can now be delivered through a structured product — a dashboard, a recurring report, a workflow tool, an AI-assisted analysis that would have taken a team a week and now takes an hour.

The buyer changes. It’s not just the C-suite anymore. It’s the analyst, the operator, the team that consumes intelligence rather than commissions it. The pricing shifts toward subscription. The sales motion becomes more repeatable.

Here is where most companies make the critical mistake: they try to sell the platform to the same people who bought the expertise. Those buyers often don’t want a platform. They want you — the senior practitioner, the relationship, the tailored judgment. They’ll tell you the platform is impressive and then ask if you can just handle the analysis for them.

New product requires new GTM. A different ICP, a different sales profile, a different value proposition. This is the stage where the services company has to run two parallel motions — maintaining the expertise sale while building the platform sale — with different teams, different compensation, and different success metrics. Most try to do it with the same people. Most discover, too late, that the best services sellers make terrible platform sellers, and vice versa.

I’ve lived this twice. At a PE-backed media and data company, we had a strong enterprise sales team — relationship sellers who could navigate a complex deal with a Fortune 500 CMO without blinking. When we launched the platform, we handed it to the same team. They kept defaulting to the managed service conversation because that’s where they were comfortable, that’s where their comp was defensible, and frankly that’s what the client wanted from them personally. We eventually built a separate platform sales motion — different profiles, different comp structure, a hard line between the two motions. It worked. But it was painful to execute mid-flight, and it would have been significantly less painful to design from the start.

The second time, I was running revenue at a digital media platform managing a portfolio of brands across multiple verticals. Same pattern. The people who sold sponsorships and custom programs were not the people who could sell programmatic and platform access — not because they weren’t talented, but because the buyer, the motion, the value proposition, and the closing dynamic were completely different. We split the teams. Revenue followed.

The lesson isn’t that services sellers are inferior. It’s that the platform sale is a different job. Treating it as an upgrade to the services sale is where the stall happens.

Stage Three: The Distribution Sale.

This is where the model fully changes. The proprietary data asset — refined through hundreds of Stage One engagements, packaged through Stage Two — becomes accessible via API, integration, or programmatic delivery. Your intelligence sits inside your client’s workflow, not just in a report they read. Your signal feeds their model. Your data becomes a node in someone else’s architecture.

This isn’t future-state. With MCP and the current generation of AI infrastructure, Stage Three is already happening for the companies paying attention. A specialized intelligence product that would have required a significant engineering build two years ago can now be exposed as a tool that a client’s AI agent calls directly — today. The data flows. The expertise scales. The senior practitioner’s judgment, encoded into the product over years, is available at the moment of need inside whatever system the client is already using.

The window to build the data asset that makes Stage Three possible is closing faster than most services founders realize. The companies that get there first won’t be the ones with the best technology. They’ll be the ones that started treating Stage One engagements as data capture events three years ago.

The GTM here is almost unrecognizable compared to Stage One. It’s developer-facing or product-facing. It’s usage-based. It may be entirely inbound. The value proposition is integration and reliability, not relationship and judgment. You’re not selling to the executive anymore. You’re selling to the team that builds the executive’s tools.

Why the Moat Doesn’t Move

The most important thing to understand about this transition: the moat doesn’t change.

An agency firm’s moat is not the platform it builds. It’s the analysts who understand context that no model has been trained on, the relationships that produce information that never reaches open sources, the judgment that comes from being inside a situation rather than reading about it.

A specialized consulting firm’s moat is not its workflow tool. It’s the pattern recognition that comes from being inside hundreds of similar situations, the practitioner knowledge that takes a decade to develop, the trust that a senior client has built with a specific person over years.

AI doesn’t replace that moat. It changes the economics of deploying it.

Technology doesn’t replace the expertise. It removes the ceiling on how many people can access it. That’s not a tech company. That’s a better services business.

The GTM has to reflect that at every stage.

What This Means Operationally

If you’re a commercial leader inside a services business making this transition, three things matter more than the technology.

First, start capturing the asset before you try to productize it. Map what signal is being generated inside every engagement. Define what gets recorded, how, and by whom. The platform you build in Stage Two is only as good as the data foundation built in Stage One.

Second, don’t let Stage One subsidize Stage Two indefinitely. The expertise sale generates margin. The platform sale consumes it, until it doesn’t. That transition point requires a decision — usually a hard one — about where to invest and where to hold. Most services businesses let the expertise sale fund the platform build for too long, because the expertise sale is comfortable and the platform sale is uncertain. The hold period doesn’t forgive that delay.

Third, design for Stage Three from the beginning, even if you’re in Stage One. The questions worth asking now: what would it mean for our intelligence to be available programmatically? Who would consume it? What would they pay? What does our data need to look like for that to be possible? You don’t need to build Stage Three today. But the decisions you make in Stage One about data structure, capture, and ownership will determine whether Stage Three is achievable at all.

Fourth, don’t let your best practitioners get trapped in the delivery motion. The person who knows how to capture signal has to be obsessed with data architecture, not just billable hours. If your senior people are still being measured on utilization, you’ve built an incentive structure that works against the transition. That’s a compensation decision, not a technology one.

The companies that make this transition successfully don’t stop being services businesses. They become services businesses with compounding infrastructure underneath them — where the expertise gets more valuable over time because the data asset it generates is getting cleaner, richer, and more connected.

That’s a different story than “we’re becoming a tech company.”

It’s also a better one.

—-

Matt Young is the founder of AI Value Edge and a senior commercial operator with two decades of experience scaling revenue inside PE-backed and sponsor-supported companies. He publishes AI Value Edge — The Growth Multiple on Substack.

In Part 1, I described a pattern that repeats itself inside growth-stage and PE-backed companies alike: tools get bought, adoption metrics look fine, revenue outcomes don’t move. And I ended with a question — what decision will this tool make faster or more accurate?

That question has a structural answer. It’s called Commercial Signal Architecture. Before you buy another AI tool, you need it.

What Signal Architecture Actually Is

Most commercial organizations have two layers.

The first is a data layer — the CRM, the billing system, the conversation intelligence platform, the product usage database. This is where the records live.

The second is a tool layer — the dashboards, the alerts, the AI-powered features inside those platforms. This is where the data gets surfaced.

What almost none of them have is a signal layer — the architecture that sits between raw data and commercial decisions, converting activity into prioritized, actionable intelligence.

The signal layer is not a tool. It’s a design. It answers four questions that every commercial leader needs answered every week:

1. Which deals in my pipeline are actually moving — and which ones have gone quiet?

2. Which customers are showing early signs of churn — before they tell me?

3. Where is my pricing leaving money on the table?

4. What is the market telling me right now that my team hasn’t surfaced yet?

These are not complicated questions. They are embarrassingly basic. And yet most commercial teams cannot answer them cleanly without a multi-day fire drill — because they have data, and they have tools, but nobody has designed the architecture that converts one into the other.

A Real Example: The $10M Blind Spot

A few years ago I was CCO of a B2B software company. We had a solid stack — Salesforce, a conversation intelligence platform, and a customer success tool with health scoring built in. The CS team was diligent. The renewal calendar was managed. On paper, we had visibility.

What we didn’t have was signal.

Renewals kept surprising us. Not because customers were unhappy — or at least, not in ways the tools were catching. The health scores would look fine at 120 days out, then turn amber at 60 days, then red at 30. By that point the account was already lost. We were reading a lagging indicator and calling it visibility.

So we went back through 18 months of churned accounts and asked a different question: not when did these accounts show up as at-risk, but what did they have in common 90 days before they churned?

The pattern was consistent. Executive sponsor had gone dark — no calls logged, no email replies in the past three weeks. Product usage had dropped more than 30% from their trailing average. And the last recorded customer conversation had no follow-up action items documented.

None of those signals, on their own, was alarming. Together, they were nearly deterministic.

We wired those three data sources together — contact engagement from Salesforce, conversation data from Gong, product usage from our CS platform — and defined a single rule: when all three conditions appear simultaneously, 90 days or more before renewal, flag it immediately to the account owner and their manager. Not a health score update. Not a dashboard refresh. A direct alert with a specific instruction: get an executive back in front of this account this week.

Within two quarters we had identified and recovered more than $10M in renewal revenue that our previous process would have lost.

The tools didn’t change. The stack didn’t change. The signal architecture did.

Why Operators Build It, Vendors Don’t

Here’s what makes this hard: no vendor sells signal architecture. They sell components of it.

Gong sells conversation intelligence. Clari sells forecasting. Salesforce sells CRM. Each of these platforms will tell you, correctly, that their product surfaces signal. And it does — within its own domain.

But signal architecture is cross-functional by definition. It requires someone to decide: which signals matter most for this business, at this stage of growth, given this revenue thesis? Which signals should trigger a rep conversation? Which should go straight to the CEO or operating partner? Which are noise?

Those are not configuration decisions. They are commercial judgment calls. And they can only be made by an operator who has sat inside a P&L and watched what actually moves revenue — not a vendor implementation team working from a template.

This is why most companies never build it. Leadership assumes the platform vendor will handle it. The platform vendor configures the system to their standard settings. And the commercial team ends up with alerts nobody reads, dashboards nobody trusts, and forecasts that feel like a negotiation rather than an analysis.

What the First 60 Days Should Look Like

Whether you’re a new operating partner in the first quarter of ownership, a CEO inheriting a commercial stack that isn’t producing, or a CRO stepping into a company that’s been running on instinct — the diagnostic is the same.

Map your signal gaps, not your tool gaps.

Walk through the four questions above with your commercial leadership team. Don’t start with what tools are in place. Start with what they can and can’t answer. You will find, almost without exception, that the gaps are not about missing tools — they’re about missing architecture connecting the tools they already have.

From there, you build a signal map: which data sources need to connect, in what sequence, to produce a clean answer to each of the four questions. That map becomes your build plan. It tells you what to configure, what to integrate, and — only then — what additional tooling, if any, is actually needed.

In most cases, the existing stack is sufficient. The data is there. The signal isn’t — because no one designed the layer that produces it.

This Isn’t Just a PE Problem

The hold period makes bad signal architecture visible faster than almost any other context — four years is an unforgiving timeline when your commercial infrastructure isn’t converting data into decisions. But the underlying problem is identical at any growth-stage company running a complex commercial stack.

Public companies feel it in missed quarters they didn’t see coming. VC-backed companies feel it in churn that surprises the board. Private companies feel it in forecasts that are more art than science.

The forcing function is different. The disease is the same.

Every organization that has invested in a modern commercial stack — and is not seeing the revenue outcomes that investment should produce — has a signal problem, not a tool problem. The data exists. The architecture to convert it into decisions does not.

The Real Competitive Advantage

The leaders I’ve watched extract consistent commercial performance from their AI investment share one characteristic: they treat signal architecture as a strategic asset, not an IT project.

They build it early. They maintain it as the business evolves. And they create something that compounds — because every quarter, the architecture gets smarter about what signals actually predict revenue outcomes for that specific business.

That’s the arb that still exists. Not which AI tools you buy. Whether you’ve built the layer that makes any tool worth buying.

The firms and companies that figure that out first will look back at the current era of AI tool sprawl the way we look back at the first wave of CRM adoption — when every company bought Salesforce and almost none of them got cleaner revenue outcomes.

The technology was never the problem. The architecture was.

Matt Young is the founder of AI Value Edge and a senior commercial operator with two decades of experience scaling revenue inside PE-backed and sponsor-supported companies. He publishes AI Value Edge — The Growth Multiple on Substack.

There’s a pattern I’ve watched repeat itself across PE-backed companies over the past two years.

A new sponsor takes control. The 100-day plan goes up. Someone on the operating team — or a well-meaning CIO — flags that the commercial stack needs modernization. A budget gets approved. Gong goes in. HubSpot gets upgraded. Clari gets layered on top. Maybe a revenue intelligence platform also gets tagged on.

Twelve months later, the board deck has a slide on AI adoption. Usage metrics look fine, but revenue performance looks almost exactly like it did before.

The tools worked. The outcome didn’t.

This isn’t a technology failure. It’s a sequencing failure. It’s costing PE companies more than they realize — not just in wasted spend, but in a more dangerous currency: lost time inside the hold period.

The Real Problem Isn’t the Tech Stack

Some operating teams diagnose commercial underperformance as a tooling problem. Pipeline is weak? Get a better prospecting tool. Forecasting is unreliable? Add a forecasting layer. Churn is creeping up? Find a customer health scoring platform.

The tools get purchased. The implementation gets scoped. The reps get trained. Then three quarters later, the CEO or operating partner is back in the same conversation — just with a more expensive stack underneath it.

The issue was never the tools, it was the absence of commercial signal architecture — a coherent system that takes all that data generated by those tools and converts it into decisions a commercial leader can actually act on.

CRM data without signal is just records. Conversation intelligence without signal is just transcripts. A forecasting platform without signal is just a more sophisticated and dressed-up guess.

What “Buying AI” Actually Gets You

I’ve spent the last decade running commercial organizations inside sponsor-backed companies — BrightRoll, Yahoo, Motorsport, Recurrent Ventures, Vivvix. Before that and after, I’ve been inside enough PE board rooms to see the pattern clearly.

The AI tool purchase checks what is considered an important box, but it usually does not solve the revenue problem.

Here’s why: most AI tools are built to surface data, and they’re extraordinarily good at it. The better ones will identify a churning account, flag a deal that’s gone quiet, or score an inbound lead with impressive accuracy.

But surfacing data is not the same as creating signal.

Signal requires a layer of interpretation, prioritization, and decision architecture that sits above the tools. It’s the difference between an alert that says “this deal is at risk” and a system that tells you which deals to reprioritize this week, why, and what the rep should do differently in the next conversation.

That layer doesn’t come in the box. And most PE-backed companies have never built it.

The Question Few Ask Before They Buy

I’ve sat in tons of tool evaluation conversations to know the questions that always get asked: What’s the implementation timeline? How does it integrate with Salesforce? What do the contract terms look like?

The question that almost never gets asked: What decision will this tool make faster or more accurate?

Not “what data will it surface.” Not “what workflows will it automate.” What decision — by which human, on which timeline, with what consequence to revenue — will this tool improve?

If you can’t answer that question clearly before the purchase order goes through, you’re not buying signal. You’re buying infrastructure with a revenue story attached.

That’s not inherently wrong, because infrastructure has value. But it will not move your commercial outcomes on its own, and if your hold period is four years, you don’t have three cycles to figure that out.

The Arb That Still Exists

The spread between PE firms that understand this and those that don’t is still enormous.

The firms that get it are running something that looks more like a signal architecture — connecting commercial data sources to a prioritization logic to a set of operator-level decisions — and they’re extracting measurable revenue outcomes from AI investment that their peers aren’t seeing.

The firms that don’t are accumulating tools. Their portfolio companies are better equipped, but they’re not growing faster. The gap is not about which tools you buy. (My personal view is that most tools are starting to become commoditized.) It’s about whether you’ve built the layer that makes any tool produce a decision worth acting on.

I’ll dive into that layer next week — and walk through what it actually takes to build it inside a PE-backed company, starting from the first 60 days of ownership.

Matt Young is the founder of AI Value Edge and a senior commercial operator with two decades of experience scaling revenue inside PE-backed and sponsor-supported companies. He publishes AI Value Edge — The Growth Multiple on Substack.

I’ve seen this movie three times before.

The first time was when I ran a $1B+ P&L inside one of the most ambitious media consolidations in recent memory. Platforms got mandated. Tools got pushed down. Business units that had wildly different commercial architectures underneath were expected to perform on a unified infrastructure that was never designed for their revenue motion. The gap between “we have the platform” and “the platform is producing revenue” was enormous — and operators like me were left to close it.

The second time was at a PE-backed media platform where I was the commercial lead on seven M&A transactions. Each acquired asset came in with its own data layer, its own CRM hygiene, its own signal architecture — or more often, its own absence of one. The assumption was always that technology would transfer cleanly. It never did.

The third time was at two different SaaS companies, both PE sponsor-backed, where I had to either build or rebuild commercial architecture inside assets with ambitious revenue targets. The tools were usually fine. The data underneath was almost always the problem.

Now the movie is playing again. This time with a 17.5% guaranteed return attached to it.

What OpenAI Is Actually Doing

Reuters recently reported that OpenAI is pitching private equity firms on a novel structure: preferred equity stakes with a minimum guaranteed return of 17.5%, packaged as majority-owned subsidiaries or joint ventures. The mechanism is elegant and the logic is sound — from OpenAI’s perspective.

PE firms have hundreds of portfolio companies. Those portcos represent captive distribution at scale. If OpenAI can get a GP to endorse adoption across the portfolio, they’ve effectively purchased an enterprise sales motion that no direct sales team could replicate. The guaranteed return sweetens the LP-level math. The adoption mandate flows downstream.

It’s smart. And it should make every commercial operator in a PE-backed company pay close attention.

The Pattern PE Has Seen Before

This is not the first time a technology company has tried to buy distribution through the private equity stack. The history here is instructive.

In the early CRM era, Salesforce got mandated top-down across thousands of companies. The platform was solid. The adoption rates were terrible. Why? Because the data architecture underneath — the account hierarchies, the contact hygiene, the activity capture — was broken. The tool had no signal to work with. By the time most companies figured out that Salesforce wasn’t the problem, they’d burned two or three years and a meaningful amount of OpEx.

The marketing automation wave — Marketo, Eloqua, Pardot — played out the same way between 2012 and 2016. PE-backed companies bought platforms under board-level pressure to modernize. Most couldn’t realize ROI because lead data, attribution models, and sales-marketing handoff architecture were disasters. Churn on those platforms was brutal, and the companies that thrived were the ones that treated the tool as the last step, not the first.

Then came the martech explosion. Scott Brinker’s landscape went from 150 tools to over 9,000 in less than a decade. PE firms pushed portcos to build modern stacks. Many ended up with 40 disconnected point solutions, no signal coherence, and a RevOps team spending 60% of their time on data cleanup instead of revenue acceleration.

Every wave had the same arc:

1. A mandate from above and adoption pressure below.

2. Revenue outcomes that often disappoint.

3. Operators often left to retrofit the architecture that should have come first.

Why the AI Mandate Is Different — and More Dangerous

The OpenAI offer has a feature the previous waves didn’t: a financial guarantee at the GP level that creates structural pressure to show adoption metrics, not revenue outcomes. That’s an important distinction.

When a PE firm has taken a preferred equity stake with a 17.5% floor, they have an incentive to demonstrate portfolio-wide deployment. Deployment gets measured in seats, licenses, usage rates. Not in pipeline conversion. Not in revenue per customer. Not in CAC efficiency.

This is exactly where the trap closes.

A portco CRO gets a call from the operating partner. “We’re rolling out AI across the portfolio. Here’s the enterprise agreement. Get your team on it by Q2.” The CRO, being a good soldier, executes. Licenses go live. Dashboards get built. The board gets a slide showing AI adoption.

Meanwhile, the underlying commercial architecture — the CRM data quality, the lead scoring logic, the ICP signal layer, the handoff between marketing and sales — remains exactly as broken as it was before. The AI has nothing clean to work with. It hallucinates pipeline. It scores leads based on garbage inputs. It generates outreach that misses the actual buying signal.

Six months later, someone asks why revenue outcomes haven’t improved despite full AI deployment. And the operator is standing in a room trying to explain that the tool isn’t the problem.

What the Data Already Tells Us

This isn’t theoretical. The early returns from AI GTM deployment are already revealing the pattern.

AI SDR platforms have reduced cost per lead dramatically on paper — from roughly $262 to $39 in some configurations. But annual churn on those same tools runs 50 to 70%. Forty-two percent of companies abandoned most AI initiatives in 2025. CAC ratios are often getting worse: companies now spend $2.00 in sales and marketing to acquire $1.00 in new ARR, a 14% increase year-over-year.

The tools work when deployed with the right foundation underneath them.

That qualifier is doing a lot of work. And many PE-backed companies don’t have the foundation.

The Operator’s Job — Before the Mandate Arrives

If you’re a CRO, CCO, or operating partner inside a PE-backed company, here’s what you can do right now, before the OpenAI agreement or something similar lands on your desk.

Audit your signal layer first. Do you know whether your commercial infrastructure is actually producing signal — or just storing activity? Can your team tell you, right now, which accounts are in-market and why? If the honest answer is no, the AI mandate isn’t your problem. The architecture underneath it is.

Separate tool readiness from architecture readiness. You can be ready for a tool in 30 days. You cannot be ready for AI-powered revenue motions in 30 days if your underlying data is broken. The honest conversation with your board is about sequencing, not resistance.

Define the revenue outcome before the deployment metric. If your sponsors are measuring AI success by adoption rate, reset that conversation. The right metrics are pipeline conversion rate, CAC efficiency, and revenue per customer. If the tool doesn’t move those numbers, the seats don’t matter.

Build the architecture now. Clean your CRM. Define your signal hierarchy. Map the handoffs. This is not glamorous work. It is the work that determines whether the tool mandate becomes a revenue catalyst or another expensive experiment that gets written off in 18 months.

The Bottom Line

OpenAI’s PE pitch is commercially brilliant. It will work. Agreements will get signed. Mandates will flow to portfolio companies.

The operators who thrive in that environment won’t be the ones who resist the tools. They’ll be the ones who built the architecture before the mandate arrived — so that when the AI has something real to work with, the revenue outcomes are there to justify the investment.

I’ve seen what happens when the sequence gets reversed. Three times. It’s expensive.

The 17.5% guarantee is a GP-level number. The revenue outcome is yours.

Matt Young writes about AI, revenue, and enterprise value at AI Value Edge — The Growth Multiple. Subscribe on Substack.

Five Months . Three Things I Got Right. Two I Underestimated. One Thing Nobody Saw Coming.

When I published my first piece here in late October 2025, I made a straightforward argument: private equity firms that treated AI as a revenue lever instead of a cost play would outperform the ones that didn’t. Lead scoring, pricing optimization, churn prediction, NRR improvement. Use AI to sell smarter, retain better, exit stronger.

Five months later, I I’m taking a look back at what I got right, what I underestimated, and what has changed in ways that genuinely surprised me.

What I Got Right

The core thesis has held. PE firms that treated AI as a commercial growth engine have pulled ahead of the ones still debating whether to run a pilot. The conversation inside boardrooms has shifted from “should we explore AI?” to “why isn’t our AI producing revenue yet?” That’s a meaningful change in five months.

The specific levers I pointed to — pricing optimization, churn prediction, AI-assisted demand generation — have all matured faster than expected. Dynamic pricing models that required a team of data scientists to configure in a few years ago are now accessible through commercial platforms that all portfolio companies can actually deploy. The theoretical became operational.

The diagnostic framework also held. The companies seeing real results aren’t the ones that bought the most tools — they’re the ones that asked the right questions first. What data do we actually have? Is it clean enough to act on? Which GTM decision would we make differently if we had better signal? That sequence still matters.

What I Underestimated

First: the speed of native AI integration into existing GTM platforms. When I wrote about the AI GTM stack last December, I was still thinking in terms of a “revenue signal layer” that sat above existing tools — a separate orchestration system that pulled from Salesforce, Gong, Amplitude, and stitched them together. That framing is already dated.

Salesforce, HubSpot, Gong, Highspot and others didn’t wait for a third-party orchestration layer to emerge. They built the intelligence in. Highspot shipped a full Deal Agent that analyzes CRM data, buyer engagement, and meeting signals and delivers real-time next steps inside tools reps already use. HubSpot’s AI assistant is now doing performance reporting, content generation, and SEO analysis from a unified dashboard. The platforms moved faster than the consultants advising them. The gap between “enterprise-grade stack” and “AI-native stack” closed faster than I anticipated.

Second: how quickly the PE firm posture shifted from cautious interest to competitive urgency. In October, I was still making the case that PE firms should treat AI as a revenue lever. By March, the better firms aren’t debating whether — they’re debating how fast. Vista Equity now requires every portfolio company to submit annual AI goals with quantified benefits. Apollo stood up a center of excellence with dedicated partners and an external advisory board. The leading institutions didn’t just adopt AI; they reorganized around it. The firms I was writing at are now ahead of the argument I was making.

What Nobody Saw Coming (Or At Least Nobody Was Writing About It)

The rise of agentic AI in GTM happened faster and more completely than almost anyone predicted.

When I used the word “AI” in October, I mostly meant models that analyzed data and surfaced insights for humans to act on. That’s not what we’re talking about anymore. Agentic AI refers to autonomous systems that don’t just identify a signal — they decide on an action, execute it, learn from the outcome, and optimize the next cycle without waiting for a human to press go.

AI SDRs that prospect, personalize outreach, handle follow-up, and update the CRM — with no human in the loop. Deal agents that monitor pipeline health, flag stalled opportunities, and recommend next steps inside Slack before the rep even opens their laptop. Churn models that don’t just predict risk but trigger retention sequences automatically.

The implications for PE-backed companies are significant and not fully absorbed yet. The traditional SDR-heavy outbound model — hire, train, ramp, watch attrition eat your team — is getting structurally challenged. In 2025, even as companies sent more emails and ran more sequences than ever before, average reply rates fell below 3% and SDR productivity dropped 22%. More volume, less signal. Agentic systems don’t solve that by doing more of the same thing faster. They solve it by shifting from output optimization to orchestration — sensing buying signals, choosing the right moment, learning from every interaction.

This is not incremental. It changes the shape of the GTM org.

What It Means for Operators Right Now

Three things I’d tell any operating partner or CEO walking into a PE-backed company today:

The stack conversation has changed. Stop evaluating tools in isolation. The question isn’t which point solution to add — it’s whether your GTM infrastructure can support agentic workflows. That means clean data, integrated systems, and a governance model for AI-generated actions. Most portfolio companies aren’t there yet. The diagnostic has to start with architecture, not software selection.

The pilot phase is over. If you’re still in “let’s run a 90-day proof of concept” mode, you’re behind. The firms generating measurable results have moved from pilots to operating standards. AI isn’t a side experiment — it’s in the value creation plan, with owners, metrics, and accountability. That’s the bar now.

The signal layer matters more than ever. Everything I’ve observed in the past five months has reinforced the core idea I keep coming back to: the companies that win aren’t the ones with the most AI tools. They’re the ones with the clearest signal — about which accounts are ready to expand, which deals are stalling, which pricing moves will hold. That’s what I’ve been calling Commercial Signal Architecture. It’s not a product. It’s an operating discipline. And in an environment where AI can execute faster than humans can decide, having clean, actionable signal isn’t a competitive advantage anymore. It’s table stakes.

Five months is a short time. In AI and GTM, it’s been a long one.

The thesis I started with still holds, but the urgency has increased. The operators who treated this as a watching brief have lost ground they’ll have trouble recovering.

Matt Young writes about AI, revenue, and enterprise value at AI Value Edge — The Growth Multiple. Subscribe on Substack.

I’ve held the CRO title four times, all across PE-backed companies — Blackstone, Bain Capital, Shamrock Capital — in media, data, SaaS, and adtech. Different industries, different investors, different boards. But one thing was almost always the same when I walked in the door: the forecast looked decent, and the reality was messier than anyone wanted to admit.

That’s not a people problem. It’s a signal problem.

So what is a commercial signal?

A commercial signal is any piece of information that tells you something real about whether a sale or renewal is going to happen, when it’s going to happen, and for how much.

Think of it like a weather signal. It doesn’t tell you what you hope is coming — it tells you what’s actually coming. A commercial signal does the same thing for your revenue.

Real signals look like this: Did the buyer actually show up to the meeting, or just the champion? Did the prospect share their budget? Did a competitor get shortlisted? Did the customer’s usage go up or down last month? Did they open the proposal or has it been sitting unread for two weeks?

Those things predict what happens next. False signals don’t. And false signals are everywhere — deals marked at 70-80% confidence because that’s what the forecast cadence rewards, not because the buyer said anything that justified it. No budget confirmed. No competitive picture mapped. No economic buyer in the room. The pipeline looks healthy. The board feels good. Then the quarter closes and nobody saw it coming.

I’ve walked into that room more than once.

What is commercial signal architecture?

If a signal is the individual piece of information, architecture is the system that captures those signals consistently, routes them to the right people, and turns them into actual decisions — pricing, forecasting, where to put your best reps, when to push, when to walk away.

At Motorsport Network I inherited 15 markets with no unified CRM and no shared revenue KPIs. You can’t run a commercial turnaround on that foundation. Getting the architecture right wasn’t a tech project — it was a precondition for everything else. At Vivvix/MediaRadar, part of what drove our revenue growth wasn’t just repositioning or pricing — it was getting honest about where deals were real versus where they were wishful thinking dressed up as pipeline.

Clean signals, better decisions. Every time.

Why this matters now more than ever

AI doesn’t fix broken signal architecture. It amplifies it.

Point an AI at corrupted CRM data — wrong deal stages, fake close dates, missing contacts — and it produces confident-sounding predictions based on garbage. It doesn’t know the data is wrong. It just finds patterns in whatever you give it.

Most PE-backed companies have a revenue stack. Very few have one that produces what I’d call decision-grade signal. The tools are there. The data is there. But the wiring between what’s happening in the market and what leadership believes is happening is leaking somewhere.

Commercial Signal Architecture is the discipline of finding those leaks and fixing them — before you add more technology, before the next board meeting, before an exit process where everything has to hold up to scrutiny.

I didn’t learn this from a framework. I learned it from walking into rooms where the numbers looked fine — and spending 90 days figuring out why they weren’t.

About AI Value Edge — The Growth Multiple

I write about how CEOs and investors use AI to turn growth into enterprise value.
👉 Subscribe on Substack (free, 2–3× per month)

There’s a question I keep hearing from operating partners right now: “Which of our companies should we be pushing hardest on AI?”

It sounds like a strategy question. It’s actually a diagnostic one.

I’ve been inside enough PE-backed businesses over the last five years to know that the companies blowing up their AI budgets aren’t failing because they picked the wrong tools. They’re failing because they deployed tools into organizations that weren’t ready to absorb them. The result is shelfware, frustrated sales teams, and a board deck slide that says “AI Initiative” next to a red RAG status.

The smarter question isn’t “should we use AI?” — it’s “is this company actually positioned to capture value from it?”

Here’s the six-part diagnostic I now run before recommending any meaningful AI investment in a portfolio company.

🗄️ 1. Is the data clean enough to act on?

AI is only as good as the signal it’s trained on. And in most mid-market PE-backed companies, the data is a disaster.

CRM data is incomplete or inconsistently entered. Revenue data lives in three systems that don’t talk to each other. Customer records have duplicates from two acquisitions ago. Nobody agrees on what a “qualified opportunity” means, let alone tracks it consistently.

Before you invest in AI-powered forecasting, pipeline intelligence, or customer scoring — ask the operating team to pull the last 12 months of CRM data and show you close rate by stage, by rep, by segment. If they can’t do that in under 30 minutes, you don’t have a data problem. You have a foundation problem. Fix that first.

🔍 Diagnostic question: Can the team produce clean, decision-grade revenue data on demand?

⏱️ 2. Does the GTM team have the bandwidth to actually change their workflows?

This one kills more AI initiatives than anything else.

AI tools require behavior change. That means reps have to stop doing things the old way and start doing things the new way — updating call summaries differently, using a new prospecting workflow, engaging with AI-generated insights before pipeline reviews. That takes real management attention and real coaching.

But most GTM teams I walk into are already overwhelmed. The RevOps person is wearing four hats. The sales manager is carrying a bag and managing a team. The CEO is still on key accounts.

Deploying AI into a bandwidth-constrained team doesn’t multiply their output — it multiplies their stress. And you end up with a tool that’s technically implemented and practically ignored.

🔍 Diagnostic question: Does the revenue org have a dedicated owner with the capacity to drive adoption and hold the team accountable to new workflows?

🧠 3. How AI-literate is the leadership team?

I don’t mean this in a condescending way. I mean it practically.

If the CEO thinks AI is “basically ChatGPT for email,” they will underinvest in the infrastructure changes needed to make it work. If the CFO thinks AI is magic and expects ROI in 90 days, they’ll pull the plug before the flywheel starts turning. If the CRO has never seen what a properly AI-augmented sales motion looks like, they won’t know what good looks like — or how to coach toward it.

Leadership literacy shapes the quality of the investment thesis, the timeline expectations, and the cultural permission structure that allows the team to actually experiment and learn.

🔍 Diagnostic question: Can the leadership team articulate a specific, realistic use case for AI in their business — and describe what success looks like 12 months from now?

📋 4. Are the core processes documented and repeatable?

Here’s a truth that doesn’t get said enough: AI doesn’t create process. It accelerates process.

If your onboarding motion is inconsistent across reps, AI-assisted onboarding will just make it inconsistently fast. If your discovery calls vary wildly depending on who’s running them, an AI call intelligence tool will surface insights from a chaotic sample. If your pricing decisions are made by gut feel in the field, AI pricing recommendations will just give people more sophisticated cover for their gut.

The companies that get the most out of AI are the ones that have already done the hard work of defining how they sell, how they serve customers, and how they make decisions. AI then becomes a force multiplier on something that actually works.

🔍 Diagnostic question: Are the core GTM and operational processes documented, followed consistently, and already producing predictable outputs?

🔁 5. Is there a culture of testing and iteration?

Most mid-market companies have a culture of “launch and leave.” They roll out a new tool, announce it in a company all-hands, and then expect adoption to happen organically. When it doesn’t, they blame the vendor.

AI implementation is an iterative discipline. The best deployments I’ve seen run like product sprints: small pilots, fast feedback loops, explicit learning, deliberate rollout. They treat the first 90 days as a discovery phase, not a deployment phase.

That requires a cultural tolerance for ambiguity, a willingness to admit what’s not working, and leadership that models curiosity rather than demanding certainty. That culture either exists or it doesn’t — and it’s hard to install in 60 days under board pressure.

🔍 Diagnostic question: Does the leadership team run pilots, measure them honestly, and iterate — or do they announce initiatives and move on?

💰 6. Is the revenue foundation already solid?

This is the one I feel most strongly about, coming from my CRO background.

AI in the revenue motion is a multiplier. It makes good pipeline better, strong reps stronger, and effective processes faster. It does not fix a broken ICP. It does not compensate for a product with real retention problems. It does not replace the need for a coherent go-to-market strategy.

I’ve seen operating partners push AI tools onto portfolio companies that have a 70% logo churn problem — and wonder why the AI-assisted prospecting isn’t moving the needle. You’re filling the bucket while it leaks.

If the company is still figuring out why customers churn, which segments are actually profitable, and what the repeatable sales motion looks like — the AI investment will produce noise, not signal.

🔍 Diagnostic question: Is the core revenue model — retention, expansion, and repeatable new logo acquisition — working well enough that AI can accelerate it rather than mask its problems?

🟢 The Scoring Framework

Run through these six questions with your portfolio company leadership. Be honest.

Score Readiness Level and Recommendation

✅ 5–6 green lights: High-priority AI candidate. Move fast and go deep

🟡 3–4 green lights: Targeted opportunities exist. Pick 1–2 use cases and pilot

🔴 Fewer than 3: Not ready yet. Fix the foundation first

If you get fewer than 3 green lights, pump the brakes. The limiting factor isn’t AI — it’s foundational. Address the data, the process, the bandwidth, or the leadership literacy first. An AI investment now will be wasted capital — and a failed initiative that makes the next attempt harder.

💡 The Bottom Line

Board pressure to “do something with AI” is real, and it’s only going to intensify. But the operating partners who create the most value over the next three years won’t be the ones who moved fastest. They’ll be the ones who moved smartest — deploying AI where the conditions for success already exist, and building those conditions where they don’t.

The diagnostic isn’t glamorous. But it’s the work that separates a real AI operating strategy from a slide deck.

Run the checklist. Know what you have. Then go build. 🚀

About AI Value Edge — The Growth Multiple

I write about how CEOs and investors use AI to turn growth into enterprise value.
👉 Subscribe on Substack (free, 2–3× per month)

🍸 Cocktail Party Version

Every portfolio company already has AI. The CIO and tech leadership may just not know it yet.

Sales reps are using ChatGPT to write outreach.
Marketing is building AI-generated content.
RevOps is pasting CRM exports into Claude.
Customer success is drafting renewal emails with AI assistance.

The risk isn’t AI adoption. The risk is fragmented, ungoverned adoption that no one is coordinating. Shadow AI isn’t coming. It’s already here.

Full Article

Private equity firms are asking portfolio companies:

“What’s your AI strategy?”

Meanwhile, AI is already embedded in daily commercial workflows — just not formally. This is the new “shadow IT.”

It’s not the old school servers in a closet. Rather, it’s:

• reps copying pipeline data into tools.

• managers using AI to analyze pricing.

• marketing automating messaging variations.

• customer success drafting retention playbooks with generative models.

The part most CEOs and boards miss:

The biggest risk is not data leakage.
The biggest risk is not rogue experimentation.

⚠️ The biggest risk is fragmentation.

🔍 Where Shadow AI Is Already Happening

In almost every PE-backed company I’ve looked at:

Sales

• AI-written outbound emails

• Automated account research

• Call summaries and coaching tools

• “Ghostwritten” follow-ups

Marketing

• Content generation at scale

• Paid media copy testing

• Landing page iteration

• ICP research

Customer Success

• Renewal email drafting

• Churn-risk pattern analysis (manual AI prompting)

• QBR deck automation

RevOps

• Pipeline export → AI → forecasting commentary

• Pricing scenario modeling

• Territory analysis

None of this is malicious, rather it’s productivity seeking oxygen.

🧨 Why Fragmentation Is the Real Risk

When AI use is uncoordinated:

1. There’s no shared data hygiene standard.

2. Prompts become tribal knowledge.

3. Different teams get different “answers.”

4. ROI is impossible to measure.

5. CIOs get surprised by commercial tool sprawl.

That’s how IT ends up “slowing things down” and as a result commercial ends up frustrated.

There needs to be better alignment.

The 30-Day Shadow AI Reset 🔄

If I were advising a PE-backed company today, I’d recommend a simple four-step sprint:

Week 1: AI Inventory

• Anonymous survey: “Where are you using AI today?”

• Identify top 10 commercial workflows being augmented

• Map data sources being touched

Week 2: Risk + Value Classification

• What data is sensitive?

• Where is AI improving output quality?

• Where is it just saving time?

Week 3: Standardize the Winners

• Formalize 3–5 high-ROI use cases

• Create shared prompt libraries

• Align on approved tools

Week 4: Governance Without Friction

• Lightweight usage guardrails

• CIO + CRO co-ownership

• Monthly review of expansion opportunities

This isn’t a massive transformation program or some six-month roadmap.

It’s just visibility and coordination.

💡 The Opportunity Most CIOs Haven’t Considered

AI isn’t just a tool question, it’s an instrumentation question.

If commercial leaders are already using AI:

• Why not capture the insights systematically?

• Why not feed outputs back into CRM?

• Why not standardize pricing simulations?

• Why not create board-level reporting off AI-enhanced signals?

Instead of reacting to AI, CIOs can own the architecture.

Instead of being asked for tools, they can define the framework.

That’s how to get in front of AI — not chase it.

🏁 The Bigger Picture for PE

Private equity doesn’t win from random AI usage.

It wins from coordinated AI leverage across:

• Pricing

• Pipeline quality

• Retention risk

• Sales productivity

• Forecast confidence

Shadow AI is proof your teams want leverage, so the job to be done is to turn that energy into structured value creation.

The real multiple expansion lever isn’t cost cutting, it’s signal clarity.

If you’re a sponsor, CIO, or CRO and want to compare notes on what you’re seeing inside your portfolio companies, I’m always interested in the patterns.

AI isn’t coming. It’s already in your CRM exports.

💬 About AI Value Edge — The Growth Multiple

I write about how CEOs and investors use AI to turn growth into enterprise value.
👉 Subscribe on Substack (free, 2–3× per month)

When I step into a PE-backed company, I’m not there to “transform” anything in the first 30 days.

❌ No systems ripped out
❌ No org charts redrawn
❌ No AI science projects

Think MRI, not surgery 🩻. In practice, that MRI is powered by AI — not to automate decisions, but to surface patterns humans can’t see consistently or early enough.

The early goal early is simple:

get a clear picture of what’s actually happening inside the business, and which decisions would change if leadership had better signals.

Here’s how I approach those first 30 days 👇

📊 1. The three datasets I always pull first

No building dashboards.
No forecasting.

Rather, understanding where reality diverges from narrative.

CRM data
Pipeline, deal history, discounting, close patterns. Not about next quarter’s number — about spotting behaviors that repeat.

Product usage or engagement data
Even when it’s messy. Revenue problems usually show up here before they show up in the P&L.

Customer outcome data
Renewals, churn notes, support tickets, success plans. This is where “growth” and “retention” quietly separate.

🟡 Perfect data isn’t required.
Signal exists long before cleanliness.

📈 2. The two views I rebuild immediately

I don’t lead with AI tooling. I lead with questions.

Revenue quality
Where deals actually come from, how durable they are, and which patterns correlate with expansion or churn. This often reframes pipeline, pricing, and productivity conversations within weeks.

Growth leakage
Where ARR is leaking before anyone calls it churn. Which accounts need attention earlier. Where teams are busy but not compounding value.

These don’t replace existing reporting — they sit beside it and sharpen it 🔍

🎯 3. The one decision I want leadership to revisit early

Almost every engagement surfaces a moment like this:

A team realizes they’re about to:

• hire before fixing coverage or productivity

• push pipeline without understanding quality

• manage retention quarterly instead of weekly

Once the signal is clear, the conversation changes quickly — not because anything was “transformed,” but because the fog lifts.

That’s the MRI doing its job.

🤝 Why this matters to investors and operators

Early AI work shouldn’t feel disruptive or risky. It should feel clarifying. This is where AI earns its keep — not by replacing judgment, but by improving the quality and timing of GTM decisions.

The best early outcome isn’t a new system — it’s:

• fewer assumptions

• better questions

• one or two decisions made with confidence instead of instinct

From there, then you decide if surgery is needed.

Often, it isn’t.

None of this requires a platform overhaul or a re-org. It requires using data companies already have to make fewer, better decisions — earlier.

I’m seeing the same patterns repeat across PE-backed companies.

💬 About AI Value Edge — The Growth Multiple

I write about how CEOs and investors use AI to turn growth into enterprise value.
👉 Subscribe on Substack (free, 2–3× per month)

The Stack Boards Believe In

When I step into PE-backed companies the revenue stack usually looks the same on paper:

• Salesforce for pipeline and forecasting

• HubSpot or Marketo for demand gen

• Gong or Chorus for call insights

• Snowflake feeding dashboards in Looker or Tableau

On paper, this looks “enterprise-grade.” In reality, it produces false confidence.

• Forecasts are wrong.

• Pipeline is inflated.

• Churn shows up too late.

• Pricing decisions are vibes, not data.

The problem isn’t the tools. It’s that none of them talk to each other in a way that produces decision-grade signal.

The Stack That Actually Works in PE Environments

What’s missing is not another dashboard—it’s a revenue signal layer that sits above the tools and translates raw activity into action.

Here’s the stack that actually works when you’re accountable for growth, efficiency, and exit outcomes.

Layer 1: System of Record (Necessary but Not Sufficient)

Primary tools

• Salesforce

• HubSpot

• NetSuite

What they’re good at

• Capturing activity

• Enforcing process

• Creating historical records

Where they fail

• Forecast accuracy

• Deal quality assessment

• Cross-functional insight

If Salesforce were enough, CROs wouldn’t miss numbers so consistently.

Layer 2: Revenue Intelligence (But Only if It’s Used Correctly)

Tools

• Gong

• Clari

• People.ai

Real operating use cases

• Detecting deals that sound good but won’t close

• Identifying reps who discount early and lose leverage

• Flagging stalled deals before the quarter ends

What I’ve seen repeatedly

• These tools are installed

• A few managers log in

• Insights never change board-level decisions

Because insights do not equal action without context.

Layer 3: Product & Customer Signal (Where the Gold Is)

This is where most PE-backed companies are flying blind.

Tools

• Amplitude

• Pendo

• Zendesk

• Gainsight

High-leverage signals

• Product usage drops before churn conversations

• Support ticket patterns predict expansion risk

• Feature adoption correlates to pricing tolerance

Churn is never a surprise—only a surprise to leadership.

Layer 4: The Missing Piece — Revenue Signal Orchestration

This is where things finally click.

What this layer does

• Unifies CRM, product, support, and financial data

• Applies AI to identify leading indicators

• Surfaces actions, not just insights

Tools emerging here

• Amsdal

• Census

• Hightouch

Real portfolio use cases

• Flagging expansion-ready accounts before CSMs notice

• Adjusting pricing bands based on usage, not gut feel

• Showing boards what will happen, not what already did

This is where AI stops being experimental and starts being operational.

The Operator’s Takeaway

If your revenue stack:

• Produces reports instead of actions

• Explains misses instead of preventing them

• Requires tribal knowledge to interpret

Then it’s not a stack—it’s a liability.

The future isn’t “AI everywhere.”

It’s AI where decisions actually get made.

💬 About AI Value Edge — The Growth Multiple

I write about how CEOs and investors use AI to turn growth into enterprise value.
👉 Subscribe on Substack (free, 2–3× per month)