Why Your Org Can't Cross (Run Your Own Pilot Numbers)

Part 3 of the field-guide companion to Builder-Leader. The first two chapters were personal: where you stand, what you hear. This one is organizational, and it ends with a number you can say out loud to your CFO.

Compute your own ratio

The headline figure from the research is blunt: roughly 88% of enterprise AI proofs-of-concept never reach production.²2The widely-cited enterprise-AI-adoption figure behind this chapter: most large surveys of corporate AI pilots put production conversion in the 10-15% range, with the 4-of-33 (about 12%) account cited here as one specific, checkable instance of the pattern. In one widely-cited account, 33 pilots yielded 4 productions. That's not a sector average you can reason around. It's the shape of what governance-heavy deployment produces when no senior person in the loop has personally crossed the build gap.

Stop using the industry number as a shield and compute your own.

On one sheet, list every AI pilot your organization has run or funded in the last 18 months. Two columns: name, and status as exactly one of in pilot, killed, in production, production claimed but dormant. Count the real productions, count the total, divide.

If you're worse than 4-of-33 (about 12%), you're in the worst quartile. If you're better, ask the harder question: does "in production" mean daily use, or a demo that earned a green slide? In most orgs the real answer moves a row or two from production to dormant. The number at the bottom of that sheet is yours, and it's more useful than any survey.

Check which myth has your budget

Institutions fail to cross the gap because their default moves are the right answers to a different question. Rate each of these on how your org actually behaves, not what the chief AI strategist says on stage:

1. Literacy myth: "Train everyone on AI fundamentals and adoption follows."
2. Procurement myth: "The right vendor plus the right contract plus the right SOC 2 review produces the capability."
3. Platform myth: "Centralize on one platform and teams cross the gap because the friction's gone."

Each is a correct answer to capability deployment. None answers personal capability acquisition by senior people, which is what crossing the gap actually requires. The myth you half-believe is the one quietly absorbing budget and producing pilots that stall. Circle it. Write one sentence on what your org does in practice that reveals the belief.side quest

The substitution that names the gap

Here's the exercise that makes the thesis concrete. Pick one live decision in your AI program: a vendor proposal review, a pilot sign-off, a go/no-go on a production push. Write one paragraph on how it would run differently if the person making the call had personally built, shipped, and operated an AI-assisted workflow on their own laptop for the last six months.

The difference usually lands in three places. They ask a different first question. They catch a specific hand-wave nobody else flags. And they change the shape of the pilot itself, not just its approval status.

If that paragraph is easy to write, your org has already crossed and you can skip ahead. If it's uncomfortable to write because the person in that seat is you, the exercise is working. The chapter's thesis is one line: you cannot lead what you have never worn.side quest

Why this is the institutional chapter

The first two parts of this series were about the individual. This one is the bridge: it shows why personal crossing isn't a nice-to-have for a senior leader but the missing input that institutional programs keep designing around. Literacy, procurement, and platform are all attempts to buy a capability that only gets built by a person doing the reps. The pilot ratio is what it looks like when you skip that step at scale.

The full Chapter 3 field guide, with the research behind the 88% figure, is in the public repo: field-guide/03-why-your-org-cant-cross.md. The book is at builder-leader.com.

Next in the series: Part 4 opens the exoskeleton and inventories the six components of a harness.