Data sources
The Collection Agent reads four logical source groups. In the demo these are simulated by a documented generator so you can evaluate the output format immediately. In production you point the same four groups at your own systems — a CSV or JSON export is enough to start.
User growth
Signups, activation, funnel conversion by stage.
Metrics. new_signups, activation rate, funnel stages
Cadence. Daily / T+1
Revenue
Total revenue split by subscription, advertising and value-added services.
Metrics. revenue, subscription revenue, ARPU, paid conversion
Cadence. T+1 (settlement)
Activity
Active users and engagement depth across web and app.
Metrics. DAU, WAU, MAU, avg session, opens per user
Cadence. Near-real-time (5 min)
Retention
Cohort retention and satisfaction signals.
Metrics. D1, D7, D30 retention, NPS
Cadence. T+1
Bringing your own data
Supply a CSV whose first column is the week label and whose remaining columns are metric keys, e.g. week,new_signups,dau,revenue. Values that are missing or non-numeric are flagged by the Cleaning Agent rather than dropped silently.