Users & retention

Both reports count only the platform or app selected in the header. They do not pull website visits in. Definitions are in App metric definitions.

ConsoleApp analyticsUsersUsers

Retention is ConsoleApp analyticsUsersRetention. Mini program analytics uses the same labels and the same pages. In the mini program view, the channel split becomes scene values.

Users

The header switches among new users, active users, starts, one-time starters, and cumulative users.

• Active users: a start or a screen view in range. Mini program starts include $mp_launch and $mp_show. This tab also draws daily, weekly, and monthly active users. Weekly is the seven days ending on the last day of the range, deduplicated. Monthly is the thirty days ending that day.
• The main daily-active card is the daily average, because a range total adds each day's people without deduplicating across days.
• Other metrics show the range total and can be split by version and channel. Reinstalling does not count as a new user.

Retention

Switch between New-user retention and Active retention, at day, week, or month grain.

• New-user retention starts from the first day someone appears and asks whether they start again in later periods.
• Active retention starts from people active in a period and asks whether they are still active later.
• Cells that are not due yet are marked as not due, not zero.

Beside the table are User freshness (the share of active users first seen within 1, 7, or 30 days) and Slipping away (cumulative users inactive for 7, 14, 30, 60, or 90 days as of the range end). To choose your own start and return events, click Custom retention, which opens Retention. Week-to-week new, returning, and churned users are in Composition.

Engagement

Headcount is in Users & retention. This page shows only distributions, with no date comparison.

ConsoleApp analyticsUsersEngagement

The mini program item is ConsoleMini program analyticsUsersEngagement.

Five distributions

Distribution	Each bar is	Meaning
Duration	One session	How long a single use lasted
Frequency	One user	How many starts in the selected range
Interval	Two neighboring starts	How long between opens
Screens	One session	How many screens in that use. Mini programs count pages
Starts	One user-day	How many opens that day

Duration and interval labels are buckets, not raw seconds. The starts chart counts each user-day, so two days from the same person are two contributions, not one summed bar.

Time of day

The Times of Day grid at the bottom is weekday by local hour in the project time zone. Switch it between starts and active users. It describes habit (weekday lunch or weekend evening), not a single day's log.

Session duration is defined in App metric definitions. Screen names and flows are in Screens. To configure intervals or distributions on your own events, use the interval and distribution models in behavior analytics.

Composition

It is weekly and does not replace the daily retention table in Users & retention. For lifecycle states on your own events, use Lifecycle.

ConsoleApp analyticsUsersUser composition

Mini program analytics uses the same label and the same page.

What the report shows

Each week's active users split three ways upward: new (first seen this week), continuing (also active last week), and returning (active earlier, absent last week, back this week). The downward bar is churned: active last week, absent this week. The change ratio is this week's active users divided by last week's. The flow chart connects last week's groups to this week's, and only those two weeks.

Two more blocks sit outside that split:

• Revisiting users: active this period, with the previous active day within 7 days, 8–30, 31–90, or 91–180 days.
• Silent users: a start in this period, but no screen change, click, or custom event.

The weekly table can be exported.

How to read the result

The change ratio says whether this week grew or shrank against last week. It does not judge the quality of new users. Silent users still count as active, because they started the app, so active users can look healthy while part of the mix goes no further.

Returning and continuing require the same person. If you count by device, a new device looks like a new user. See Visitors, users & identity. For people inactive for 30 or 60 days, use Slipping away on the retention page instead of adding up churn bars.