Loading analytics…
Turn metadata only — transcripts are never logged. Verify-fails count model outputs the server rejected (invented IDs, bad JSON).
Distinct real users who engaged with each feature. Bar = % of real users (all-time); count shows month-to-date / all-time.
Month-to-date activity per user across schedule events, tasks, meal plans, groceries and the smart-add parser. "Last active" is derived from real product activity, not just sign-in.
| User | Role | Events +/✎ | Tasks +/✎ | Meals | Groceries | Reviews | Parser | Profile | Last Active | Action |
|---|
Send a gentle, SparkDay-branded nudge — with fresh AI insights on schedules + meal prep and tips for getting the most out of the app — to real users inactive for 3+ consecutive days. Demo and unsubscribed accounts are always excluded. Preview and "Find inactive" never send.
| Name | Role | Subscription | Plan | Source | Renews/Ends | Original Tx | Last Purchase Event | Signed Up | Action |
|---|
| Reason | Source | Unsubscribed |
|---|