Metrics dashboard / product analytics view

Decision Define activation as completing 3 of these 5 events within 7 days of signup: 1. workspace_created 2. first_project_created 3. first_teammate_invited 4. first_dispatch_completed 5. first_invoice_sent Why these five Cohort regression on the last 6 monthly cohorts shows these five events predict day-30 retention at r = 0.71. Adding a sixth event drops r to 0.69 (overfitting). Removing any one drops r below 0.55. Instrumentation Segment track() calls, all server-side from the API layer (no client gaps). Properties on every event: account_id, plan, signup_source, signup_at. Backfill: 90 days from event logs. Owner: Data team, ETA 4 days. Dashboard Single Mode dashboard: "Activation cohort tracker". Tiles: - 7-day activation rate by signup week (current vs trailing 8-week avg). - Event-completion funnel (which of the 5 events drops off most). - Activation by signup source. - Pre/post comparison (set the baseline cutoff at the deploy date). Cadence Weekly Thursday review with PM + CSM lead. 15 min, decision-first format: - What changed this week. - One thing we're testing next week. - One thing we'll kill if no signal by [date]. Risks - Server-side gaps for legacy mobile clients. Mitigation: client-side fallback for 2 deprecated versions only. - Backfill assumes event log retention; spot-check before kicking off cohort regression rerun. Next step Segment plan ships day 1. Dashboard live day 3. First cohort report at week 2.