Within-month completion/start ratios fell from 20% (200/1,000) to 15% (180/1,200): down 5 percentage points, or 25% relative [S1]. These are provisional ratios, not established cohort conversion rates; completion and start populations may differ. Starts rose 20%, while completions fell 10%. Setup confusion is a candidate explanation, not established cause [S2]. Priority 1: use part of the two-day capacity to check event definitions and compare cohort completion windows; owner proposed: available engineer; completion evidence: reconciled counts with matching observation windows. Priority 2: review a small redacted sample of setup failures and support cases to locate a repeated blocker; owner unassigned. If a consistent blocker is found, scope one reversible fix for the remaining capacity. Success measure: completion rate for comparable trial cohorts over an agreed window; no improvement target is approved yet. Reassess before building if the metric change is a tracking/cohort artifact. Plan is provisional; no experiment or fix was executed.