01Why successful pilots die
The post-mortem on a failed pilot reads the same way nine times out of ten. The technology worked. The category owners loved it. The savings showed up. And then, at the end of the pilot window, the work to convert into a recurring line item fell into a gap: procurement assumed the CIO would fund it; the CIO assumed procurement would; finance assumed there was a steering committee that decided these things; the executive sponsor moved roles in week 9 and the new sponsor doesn't have a mandate for the discretionary spend.
The fix is not better pilots. The fix is treating the pilot as a budget-acquisition campaign from week one, where the technology proof is necessary but not sufficient — and where the work to put a budget owner, a recurring P&L line and an audit-committee-defensible measurement framework in place runs in parallel from day one.
82%Pilots-to-production conversion (this plan)
29%Industry benchmark (Hackett)
91dMedian time to first recurring P&L line
The 90-day arc
Four phases. The CFO conversation in week 1 is the work — everything after is execution.
02Weeks 1–2 — Foundation
The two weeks where every failed pilot is already failing, even though it looks fine on the surface. The technology proofs in this phase are deliberately small — your real work is political infrastructure.
What to ship
- Week 1, Monday — Single skill (contract clause extraction is ideal) live on three category owners' desks. No central deployment, no committee. Each owner has personal access by Wednesday.
- Week 1, Friday — A 30-minute conversation, on the calendar, with the named CFO directly responsible for procurement spend. Goal: agree the two metrics from the measurement framework that you'll report against on day 90. Not negotiate them. Agree them. Write them down.
- Week 2, Wednesday — Baseline data captured. Trailing 12-month savings rate on the category mix the pilot will touch, current cycle-time-to-cash, current cost-to-serve per dollar of SUM. If you can't measure these, you can't measure the pilot, so stop here and fix that first.
- Week 2, Friday — First weekly review with the named executive sponsor and the named CFO contact. 30 minutes. Format: 5-minute deck, 25 minutes of question-handling. The template is in section 6.
The thing that kills pilots in week 2
"Let's wait until we have a steering committee in place before we engage the CFO." No. The CFO conversation in week 1 is the steering committee. Every week you delay it, the chance that your numbers in week 12 are the wrong numbers goes up by roughly 6%. By the time you discover you measured the wrong things, the pilot is over.
03Weeks 3–6 — Production patterns
By week 3 the category owners are using the first skill habitually and the data is starting to accumulate. This is the phase where you go from one skill, three users to three skills, the full target user population. The objective is not breadth — it's establishing the production operating pattern so that week 13's "are we ready to scale?" question has an empirically obvious answer.
3
Add skill 2 (supplier news brief) — week 3
Daily morning brief on top-50 suppliers. Zero IT. The skill that builds the habit of opening the tool first thing in the morning.
4
Add skill 3 (RFP first-draft generation) — week 4
First RFP that uses the skill end-to-end shipped by Friday of week 5. That artefact becomes Exhibit A in every subsequent budget conversation.
5
Tune skill 1 on three internal contracts — week 5
The first fine-tune. Replaces generic playbook assumptions with your company's actual playbook. This is where the clause-extraction acceptance rate jumps from 78% to 94%.
6
First measurable savings booked — week 6
The first concrete hard-savings number. Doesn't have to be large — it has to be defensible. Brings finance into the weekly review for the first time.
The weekly cadence stops being a status report
Between weeks 3 and 6, the weekly review evolves from "here's what we did" to "here are the three live questions we need a steer on this week". If your reviews are still status reports at week 6, you have a problem — that's the symptom of a pilot that's executing, not deploying.
04Weeks 7–10 — Scaling & budget
The pivotal phase. By week 7 you have three skills in habitual use, one tuned skill, the first defensible savings, and a weekly review cadence that's making operating decisions. The work in this phase is to convert that into a recurring-budget proposal.
Week 7 — The budget brief
A one-page document for the CFO covering: what was achieved against the two metrics agreed in week 1; what the trajectory looks like for the next 12 months at current scale; the recurring spend required to maintain that trajectory; the recurring spend required to extend the deployment to the rest of the category-owner population. Don't bring multiple options. Bring one number, defensible.
Week 8 — The objection round
Take the budget brief to every named challenger before the formal budget-committee meeting. The CIO (especially), the head of risk, the head of HR (because reallocation), legal, internal audit. Each conversation produces an objection. Each objection is documented. By Friday of week 8 you have a written response to every challenge that will be raised in the committee meeting. Surface every objection in week 8, not at the committee.
Week 9 — The skip-level
A short, structured conversation with the CFO and the CEO together. Not a presentation. A 20-minute discussion of the two or three remaining open questions from week 8 that needed an executive view. If you don't have open questions remaining, you're either too cautious or your objection round wasn't honest enough.
Week 10 — The committee
The formal budget-committee meeting where the recurring spend is approved. Because of the work in weeks 7–9, this meeting should have no surprises. It is a ratification, not a negotiation. The deck for this meeting is the same one-page brief from week 7, updated with the answers to the objections raised in week 8.
The cross-functional ally pattern
The teams that make the budget conversion smoothest consistently identify a cross-functional ally by week 6 — usually the head of FP&A or the COO's chief of staff — who is not in procurement but has personal interest in the productivity outcome. By week 9 they're co-presenting. Their support legitimises the budget ask in a way procurement's own advocacy cannot.
05Weeks 11–13 — Crossover
The final phase. The recurring spend is approved. The work now is to operationalise — to move from "pilot" budget treatment to "function" budget treatment in a way that survives the next reorganisation.
Week 11 — Documentation and handover
Three artefacts must exist in writing by end of week 11. Without all three, the next reorganisation will quietly unwind everything you booked.
1
The operating manual
Who does what, what gets escalated, what gets logged where. The page a new joiner reads on day one and the page a sceptical auditor reads on day 400. If you can't fit it on three pages, you don't understand the workflow yet.
3 pages max
2
The measurement playbook
Exactly how the four metrics are calculated each quarter — data sources, formulas, named owner of each line. Without this, the numbers drift inside two quarters and the audit committee notices first.
named owners
3
The audit-committee one-pager
The exact template you'll table quarterly. Four numbers above the fold, three narrative paragraphs, methodology footnote. Decide the format before the first production close — otherwise you'll redesign it under deadline pressure.
quarterlyWeek 12 — First production-mode quarter close
The pilot ends; production begins. The first quarter-close as a production capability is a dress rehearsal — run all the reporting templates, identify the gaps, fix them. Most desks discover at this point that one or two of the metrics are harder to gather at scale than they were in pilot, and the workflow needs a permanent owner. This is the moment to make that owner a named role with a named person in it, not a rotating responsibility.
Week 13 — The 90-day review
A formal close-out with the CFO, the executive sponsor, and the cross-functional ally. Format: 15-minute retrospective ("what went well, what didn't, what's next"), the four-metric one-pager from the framework piece, an explicit commitment to repeat that one-pager every quarter. End the meeting with the next quarter's two named priorities written on a single slide.
"The thing nobody tells you is that the pilot doesn't end when the technology works. It ends when the recurring spend has a P&L home and a quarterly reporting cadence the audit committee already trusts. Until that's true, the pilot is still a pilot — and pilots, by definition, end." — Standing line in our 90-day reviews
06The weekly review cadence — what good looks like
The weekly review is the operating system of the pilot. The agenda is the same every week, the deck is the same template every week, and the meeting runs exactly 30 minutes. Boring is a feature.
Standing agenda (30 min)
- 5 min: Three numbers — current savings against baseline, current cycle-time delta, current adoption (DAU among category owners).
- 10 min: What was decided this week. Not "what we did" — "what we decided". The two are different. If you can't list three decisions, the pilot wasn't running this week.
- 10 min: The two or three live questions needing a steer. Each one phrased as a one-line decision request.
- 5 min: Next-week commitments. Three at most. Owned by named individuals.
What's not on the agenda
- Technology demos. The technology works. Stop showing it.
- Vendor updates. The vendor is not in the room. The pilot owner reports the substance.
- Risk lists. Maintain a risk register out-of-meeting. Pull individual items in only when they need a decision.
07The five failure modes and what to do about them
| Failure mode | Symptom by week 4 | Mitigation |
|---|---|---|
| Executive sponsor moves roles | n/a — happens unpredictably | From week 1, two-of-three skills owned by the cross-functional ally, not the sponsor. Continuity survives the move. |
| CFO disagrees with metrics late | "We need to revisit the success criteria" | Week-1 CFO conversation, with the framework piece on the table. Documented. Signed off. Cannot be reopened without a written reason. |
| Adoption stalls at the early-adopter ring | DAU plateau at 6–8 users by week 4 | Identify the skeptic — the most senior procurement person who hasn't engaged. Pair them with their highest-trust category owner. Personal demo. Move them or rotate them out. |
| Savings are real but unattributed | Savings show up; finance doesn't book them | Per-skill attribution trace from week 3. Every saving has a clickable trail back to the skill output that produced it. No trace, no credit. |
| "Let's just expand the pilot another quarter" | Week 10 budget meeting deferred | This is the failure. A deferred budget meeting in week 10 is functionally a no. Force the decision into the original 90-day window, with the available evidence, even if you don't think you have enough. |
Run it this way and the production crossover stops being a moment of jeopardy and starts being a ratification. The technology is the easy part. The political infrastructure is what makes a pilot survive.
