Pilots that convert

LAST REVIEWED 2026-09 · SOURCED FROM 1 SESSION, SEP 2026

The pilot is not the deal

The session was for founders who already have real customer interest and a pilot on the table. Design partners and co-builds were explicitly out of scope: a customer pilot is someone buying the product as prescribed, proving a couple of things inside their organization before they roll it out. They are not co-creating the product with you. “Enterprise” was used loosely — SME through large — because the motion is the same: interest, pilot, proof, purchase order.

The reason this motion got harder: the bar moved. Investors want more revenue. Buyers, after 18 months of rushed AI spend that did not deliver, want a pilot as a way to de-risk the deal — the same instinct that makes a VC ask for more traction. Indirect savings are not enough. The two things that actually propel a deal are reducing cost or risk and increasing revenue, in numbers the buyer can take into a business case. A third failure mode the room kept hitting: the software cannot survive security, procurement, and rollout past the first team.

“A pilot should not end with the customer liked it. It should end with a purchase order.”

— Venture partner and former enterprise seller (Workday, Microsoft, SAP) who now sits on startup boards · session, Sep 2026

That is why collapsing a six-figure enterprise pilot into a cheap monthly subscription can feel easier and still be a trap. One industrial-AI founder in the room had done exactly that — annual contract billed monthly, a two-to-three-month validation window, terminate if it fails — and the mid-market deals continued. The challenge from the front: you just anchored yourself at the lower price. The pilot is supposed to be the foot in the door for land-and-expand, not a substitute for the commercial deal.

Qualify the close before you start

Work backwards from the number you actually need. If the objective is a $100K annual license, the first qualification is not “will they try it?” It is: if we prove we can solve this problem, will they buy, and at what price? A successful pilot against the wrong price is a failed deal.

Always charge. Free pilots get less value, less engagement, and — in the speaker’s experience — do not convert. The working pattern from her own selling years: a paid diagnostic (on the order of $50K, not a token) to map the enterprise problem and how the product addresses it. The customer can shop that diagnostic around. Because the success criteria, stakeholders, and decision path were in the agreement, the claimed conversion rate was 99% — the 1% that said no bought something else, watched it fail, and came back.

Two related qualification questions:

Budget threshold. One portfolio company closed a named chip manufacturer in four days because the economic buyer could put $20K on a credit card. Above that number it would have gone through AP for months. Price the pilot under the threshold that lets them start, then go to the six-figure phase two.
Compelling event. Extract why they have to start on this date and finish on that date. An acquisition cutover, a system that disappears on April 1, a safety mandate that can be purchased outside the normal budget process — those close. “We should probably do a pilot” does not.

Scope it like an MVP

Hone in on one problem, one use case, often one department. Write three to five success criteria into the agreement before anyone starts. They have to be measurable, and you have to be able to show the evidence — ideally instrumented in the product.

Weak	Strong
“The team likes it”	80% of the team uses it weekly, unprompted
“It should save time”	Cycle time drops from four hours to under one
“We’ll improve conversion”	Prospect-to-close conversion improves by X%
“It handles the records”	500 records completed at X accuracy

Direct savings beat indirect ones. An airline-hardware founder in the room had completed four field trials that qualitatively won ops, safety, unions, and ramp agents — fatigue down, bags less damaged, turns faster — and still lost finance, because the savings were soft. The next trial was scoped to three numbers finance could underwrite: load time equal or better than the manual standard, unload time faster, and greater than 90% of bags completing the automated handoff. Same product; different proof.

Turn those criteria into an ROI model at the end — some portfolio companies now stand one up in a couple of hours with AI tools, including a calculator the customer can play with. Plug in labor cost, volume, and what happens to the hours you free. “Six months in, this has already paid for itself” is the slide that moves a commercial decision. Put a version of the calculator on the website so prospects can run it before they ever talk to you.

Keep the clock tight. One to two months. The longer a pilot runs, the more it drifts into a science project that never becomes a deal. Write the start date, the reason it has to start then, the weekly milestones, the end date, and the reason it has to end then. Treat in-scope and out-of-scope the way you would an MVP; uncontrolled additions are how pilots become consulting.

“Ideally your pilots are one to two months. The longer a pilot is, the more likely it is to drift into not ending up in an enterprise deal.”

— Same session, Sep 2026

(The first-customer contract playbook still holds for design partners: three to six months, discount traded for proof points. That is a different motion. Do not mix them.)

The champion is not the buying committee

Single-threading is how a well-run pilot dies when someone changes jobs. Map, in writing, who approves, who funds, and who uses. Score each person positive, negative, or neutral.

Role	What they actually do
Champion	Wants you to win; usually your first contact; knows the problem
Economic buyer	Owns the budget and writes the check; sometimes is also the champion — that is the ideal
User / operator	Lives in the workflow; can validate value, or quietly kill the deal because they fear the change
Procurement	The path the PO has to walk; you need it mapped before the commercial decision, not after
Security / IT	Can block install or data access; start them in parallel with the pilot when you can

Negative users convert when they feel they are contributing. Give them one-on-one time, let them shape the pilot, and the fear often turns. Neutral procurement and security do not convert by charm; they convert by a documented path you started in week one. Healthcare in particular will ask for security review and AI-governance sign-off before the pilot. Ask what they actually need — HIPAA, SOC 2 Type I vs Type II — and pressure-test whether a gap is a deal-breaker or something you can parallel-run. One honest line from the room: “We have Type I now; Type II is a couple of months out. Is that a deal-breaker, or can we start?”

On trust: if they are asking for case studies before they will even pilot, they are not early adopters. Keep nurturing them, and go find the buyers who want to be first and will do the press release, the webinar, the event. Hardware founders get a later exception: once the product is proven in market, you do significantly fewer pilots, the way airlines do not test-fly a Boeing before they buy.

Run the clock

Before kickoff, the agreement already has problem, scope, criteria, stakeholders, and the path to a PO. Kickoff is not a celebration — it is a re-alignment, because people show up who were not in the signing meeting.

Then keep momentum. On a one-month pilot:

Week 1–2 — a visible quick win. Signing-to-kickoff energy dies in a lull. A first success in the first two weeks is what keeps the committee engaged.
Midpoint — a formal review. How are we tracking against the criteria? What blockers appeared? Do we adjust, or are we on track?
End — tick the boxes, then ask. These were the success criteria. We hit them. Do you agree to move forward on the commercial deal?

Always be closing, through the whole thing. By the time you reach the commercial conversation you already know what security review and procurement look like. Finding out then is too late.

Domain still beats a generic model. When buyers say “we already have Claude,” the wins in the room were companies that could talk the customer’s workflow in their language — and, in one case, a $35K paid side-by-side of the product versus doing the same job in a general-purpose model. Verticalized pain is the answer; “we also use AI” is not.

When it stalls, restart it

Stall	What to do
No real economic buyer	You never engaged the check-writer. Go get them.
Success is too vague	Rewrite the criteria so they are quantifiable, or you have nothing to close against
It became a science project	Pull activity back to the buying decision; new requests go out of scope
Champion goes quiet	They left, got acquired, or changed roles. This is why you multi-threaded
Procurement takes over	Loop the champion or economic buyer and have them apply pressure. Tie it to the compelling event: if we miss this date, you lose the people who would install it
They love it, no budget	Re-qualify the cost of no decision with the ROI model. Then reduce scope — one team, one site — rather than discount. Discounting trains the next deal

“I’m not a big believer in discounting. It’s more a reducing scope.”

— Same session, Sep 2026

And if they needed case studies you do not have yet: say so. “Do you need more proof before you can buy? We have customers in the pipe; once those land we can connect you.” Sometimes they wait. Sometimes they decide they want in now. Either way you stopped pretending.

Numbers from the room

Figure	Value
Pilot duration	1–2 months; longer drifts away from a deal
Success criteria	3–5, measurable, in the agreement before kickoff
First visible win	Inside week 1–2
Paid diagnostic (one pattern)	On the order of $50K
Credit-card / AP threshold (one deal)	$20K on a card, four days pitch to close, then a six-figure phase two
Claimed paid-pilot conversion	99% when the agreement carried criteria, stakeholders, and a decision path
Side-by-side vs. a general-purpose model	One company charged $35K to run it
First-customer / design-partner duration (different motion)	3–6 months; see First customer contracts

Sources

One working session (Sep 2026) with a venture partner and former enterprise seller — Workday, Microsoft, SAP; selling since the late 1990s; now on the boards of the startups her firm backs — on converting a paid customer pilot into a purchase order. Founders in the room were running live pilots in industrial AI, airline ground operations, and healthcare.