# Acceptance checklist: executive report (v1.0)

Run every line before a report leaves your hands — AI-drafted or not.
Each item is binary: it passes, or the report goes back. This is the
quality gate at step 7 of the
[Research to Executive Report workflow](/en/workflows/research-to-executive-report/).

## Bottom line

- [ ] Section 1 states a recommendation, a cost, and a deadline in ≤ 120 words.
- [ ] Section 1 survives being read alone (no reference to later sections).
- [ ] The title states the decision, not the topic.

## Scope

- [ ] The decision request is a yes/no or a budget figure, not a theme.
- [ ] Out-of-scope adjacencies are listed, each with a one-clause reason.

## Evidence

- [ ] Every factual claim in sections 1, 5, and 7 maps to ≥ 1 row in the
      evidence table.
- [ ] Every evidence row carries a working link or a named document.
- [ ] Every confidence entry says why (measured / single source / vendor
      claim / inference) — bare "high" fails.
- [ ] Section 4 lists "what we did NOT verify" — or, if no gap was found,
      records what was checked and how (a bare "all verified" fails).

## Provenance

- [ ] Input materials are linked in the appendix.
- [ ] Method + version noted (who/what produced the draft, which revision).
- [ ] AI-drafted passages were human-read in full (not skimmed headings).

## Risks and options

- [ ] Each risk has an observable trigger, not just a severity word.
- [ ] Options table includes reversibility AND portability (what happens to
      the artifacts if we stop or switch tools).
- [ ] Contradictions in the inputs are surfaced as risks, not averaged away.

## Delivery

- [ ] Dates and numbers spot-checked against the input notes.
- [ ] The exported file opens cleanly outside the authoring tool (the
      portability smoke test).
- [ ] Next action has an owner role and a date.

---

Scoring: 18 items. Anything unchecked → revise and re-run. Keep the checked
list with the report — it is your audit trail when someone asks "how do we
know this?" six months later.