Daily active users
Intent to walk funnel
Distinct users hitting each step in the range. Steps are counted independently, so a later step can exceed an earlier one if someone arrived by a different path.
Which time slots do people want?
Saved walk intents per slot, split by activity.
Slot demand by day of week
Darker means more intents. Separates "7am is popular" from "7am on weekdays is popular".
Does demand convert?
A slot with lots of intent but a low match rate is a liquidity problem, not a preference signal.
Where do people want to walk?
Demand by neighbourhood pin.
Hotspots
Meeting points chosen by the matcher, and — if route sampling is enabled — where sessions are actually walked. Both are aggregated to a ~110 m grid.
Experiments
Flags and A/B tests. Toggle enabled in the Supabase SQL
editor; the client picks it up on next launch with no release.
Read out a result
Conversions count distinct users who fired the event after they were enrolled. Check the arm sizes are roughly balanced before reading anything into the rates.