# First win: a checked campaign performance brief

All records, claims and scenarios in this pack are fictional teaching material. They do not describe Joseph Carroll's experience or a real business. Files are version 2026-11-16.

## Use the pack in five steps
1. Define/check sources: read S1 campaign-performance.csv and S2 source-notes-and-definitions.md. Identify the outcome, grain, period, definitions and permitted actions. Reconcile the four source rows.
2. Brief/delegate: adapt working-brief.md. The lead writes unless a writer is explicitly assigned, and owns the final brief. Analyst and reviewer outputs are optional separate roles. State dependencies, permissions and the working arrangement.
3. Draft: use an approved assistant, write manually or use flawed-draft.md. No account or API is required. The supplied draft deliberately contains errors.
4. Inspect/correct: use the sources and the answer-free learner checklist, review-rubric.md. Record your own review before opening the facilitator answer keys, review-ledger.md and reviewed-reference.md. Recompute the numbers and inspect the claims attached to them.
5. Finish: write a concise executive brief, then compare with reviewed-reference.md. Record the evidence behind each of the four checks and retain unresolved questions. Different defensible wording is welcome.

The CSV uses the exact columns accepted by the AI Lab KPI workbench. Download campaign-performance.csv from this exercise and select it under “Load a campaign CSV”. The workbench’s default sample and its “Download sample CSV” action refer to different practice data. Confirm “Uploaded CSV” names campaign-performance.csv before using the totals. Download the calculated report before leaving; reloading returns to the default sample. Additional source context stays in the notes. The workbench checks numeric format and calculates sums, but it cannot verify the source assumptions.

For a group, follow facilitator-notes.md and its 60-minute agenda. For self-paced practice, plan around 30 minutes and take longer when needed. These are suggested allocations, not guaranteed completion times.

Open the guided exercise: https://carrollconsultingservices.com/ai-lab/first-win
Open the KPI workbench: https://carrollconsultingservices.com/ai-lab/tools/metrics
Run the team session: https://carrollconsultingservices.com/ai-lab/teach
Record a pilot comparison: https://carrollconsultingservices.com/ai-lab/my-plan#comparison

This pack supports drafting and review only. It grants no permission to access accounts, change budgets or publish. For a shared-screen session, the lead/operator keeps the group’s My AI Plan in one browser. For separate browsers, participants export Markdown to an approved lead-owned document; the lead integrates corrections manually. JSON restore replaces the receiving plan and never merges participant work. Export your work to keep or share an artifact; the kit does not create a shared team workspace.
