If week one is the contract, week twelve is the audit. Here is what is sitting in your shared folder the day we hand the keys back.
The closing posture
Most agency engagements end with a final invoice and a vague promise of "continued support." We end with a handover ledger, an exit-ready system, six published receipts, and every falsifier we wrote down at the start either survived or recorded as failed with a written reason.
If we cannot produce that closing set, the engagement is not finished — even if the calendar says it is. We will keep working at no extra cost, or we will say plainly which artifact we could not deliver and why (Class C — configuration written down where both sides can see it).
1. The handover ledger
A single document, versioned, that lists every decision made during the engagement with the date, the people in the room, and the evidence class it landed at. The ledger is the ground truth — not the slide deck, not the email thread, not the Slack channel. If a decision is not in the ledger, it did not happen.
The reason the ledger exists in this form is that evidence-disciplined practice treats the written record as the single source of truth (Class E — see the public methodology notes on /trust-receipts-standard). When the next team reads it six months from now, they should be able to reconstruct why each call was made without phoning anyone.
2. The exit-ready system
By week twelve, the work we did is running without us. That means:
- Every credential we touched is back in your name, with a written list of which ones changed and when.
- Every script, config file, and integration point is in a repository you own, with a one-page operating note for each piece.
- The exit-plan draft from week one is now the exit plan — updated, costed, and accurate to what actually got built.
- A named person on your side has executed at least one full operating cycle of the new system, with us watching but not driving.
If any of those four conditions are not met, we are not yet done, and the ledger says so out loud.
3. The six published receipts
Over twelve weeks we publish six short receipts — one every two weeks — that summarize what changed, what was measured, and what evidence class each result landed at. The receipts are public artifacts in your name, not ours, written so that your board, your team, or a future vendor can read them without needing us in the room.
The receipts are not marketing. They are the audit trail. If a receipt is missing or out of sequence, that is a finding and the ledger records why.
4. The falsifier verdicts
This is the one most engagements never produce. For each working hypothesis we wrote down in week one, the closing document records one of three verdicts:
- Survived — the falsifier condition did not occur, and the hypothesis is now closer to A-grade evidence.
- Failed — the falsifier condition did occur. The hypothesis was wrong, we wrote down why, and the work was rescoped.
- Inconclusive — the experiment did not run long enough or clean enough to decide. We say so, and we say what would resolve it.
A "failed" verdict is not a defect of the engagement. It is the engagement working as designed. A vendor who never records a failed falsifier is a vendor who is not actually testing anything.
5. The replacement note
The last artifact, and the most uncomfortable to write. A short memo addressed to whoever might replace us — naming the parts of the system that are most fragile, the decisions we are least confident about, and the places we would look first if something broke in month four.
We write the replacement note because the alternative is a vendor lock disguised as "institutional knowledge." If we have done the engagement honestly, the replacement note is short. If it is long, that itself is a finding (Class C — written into the closing ledger).
Why this matters
The whole point of trust receipts is that on the Monday after week twelve, with the handover ledger open and the exit-ready system running, you should be able to keep going without us — or replace us — without needing to ask anyone for permission, a slide, or a final status call.
If that is the way you want engagements to end, the /workshop is where we figure out whether we can actually deliver it for your specific situation. The /partnership-vs-vendor-contract is the short read first if you want to see the contract language that makes a clean handover the default rather than the exception.
The full standard the closing artifacts are written against lives on /trust-receipts-standard. The standard is public so that other people can hold us to it — and so that you can fork it for the next vendor after us.