Qmr

QMR: the quarterly marketing review

You assemble the quarter's evidence and narrative skeleton; the team brings the judgment. The QMR is trustworthy because every number traces to a snapshot in data/ and means what data/ontology/ says it means.

Needs: nothing wired; data/ontology/metrics.md and funnel.md filled. It does more with crm, web-analytics, ads, marketing-automation and seo-data wired (which vendor fills each is the Wired table in integrations/README.md; the checklist maps every item to its category), and for each category that is not, it names the export and the data/<domain>/snapshots/ path to drop it at. A missing snapshot is a gap in the report, never an estimate.

A person runs this, at quarter end, in a session: it needs the team's answers and the exports it asks for. The snapshot pulls it depends on may run on a schedule instead (docs/operating-model.md).

Procedure

1. Open the quarter's folder. reports/qmr/<year>-q<n>/ (ask which quarter if ambiguous). Copy in the pack from reports/_templates/qmr/: report.md and data-checklist.md. If the folder exists, resume; the checklist tracks what's still missing.

2. Work the data checklist. data-checklist.md lists every snapshot the QMR needs, its expected path, and how to get it. For each item:

   • Already in data/*/snapshots/ and covering the quarter → check it off with the path.
   • Category wired (the Wired table in integrations/README.md) → pull it through snapshot-pull, which saves it under the expected snapshot name; check it off with the path.
   • Category not wired → give the human the exact export instruction from the checklist, the category's manual route ("In the CRM: … export as CSV, drop it at data/crm/snapshots/<name>.csv"), and leave the item open. Batch all the export asks into one list; nobody wants seven interruptions.

3. Compute. With the snapshots in hand: quarter's headline metrics vs targets, funnel conversion by stage, deltas vs the previous quarter's QMR (read reports/qmr/<prev>/report.md; its Data-used section names the comparable snapshots). All terms per data/ontology/metrics.md and funnel.md. Content shipped: grep content/ frontmatter for status: published with a published: date inside the quarter. Projects: read projects/*/status.md and _archive/ for what closed.

4. Fill report.md. Numbers and evidence fully; narrative sections (wins, misses, lessons) as drafted claims the data supports, each flagged for the team to confirm or rewrite. Decisions-needed section: collect open questions from project statuses and anything the quarter's data obviously raises.

5. Dashboard. Build dashboard.html next to the report via the make-dashboard skill: headline metrics with vs-target and vs-last-Q deltas, the funnel, trend lines where ≥2 quarters of snapshots exist.

6. Report status. End with: checklist items still open (and who owes which export), numbers that couldn't be computed and why, and the flagged narrative claims awaiting confirmation.

Rules

• A gap is a gap. An uncomputable metric appears in the report as "missing: needs ", never an estimate silently presented as fact.
• First QMR ever? No previous quarter to diff. Say so, and establish this one as the baseline.
• If data/ontology/metrics.md is unfilled for a headline metric, that's a blocking question for the team, not a definition you pick.