Overview
The shipped mini program SDK covers three platforms only: WeChat, Alipay, and Douyin. See Mini program platform differences.
Console menu
After a mini program source is registered, switch to mini program analytics.
ConsoleMini program analyticsOverviewOverview| Group | Items | Contents |
|---|---|---|
| Overview | Overview, Realtime | Core counts, and who is active right now |
| Users | Users, Retention, User composition, Engagement | The same pages as app analytics |
| Scenes & sharing | Scenes, Sharing | Entry scenes and forwards, specific to mini programs |
| Pages & events | Pages, Custom events | Pages by path, and your own events |
| Devices & quality | Devices, Stability | Hardware, crashes, and stalls |
| Configuration | SDK code, Mini-program settings, Settings, Business settings | Install the SDK and drop test devices |
The header can filter by platform or a single mini program. Overview, realtime, users, retention, composition, engagement, devices, and stability use the same rules as app analytics, locked to mini program platforms. Versions and attribution from the app shell are not in this sidebar; sharing and scenes are the two extra pages here. Boards and configuration follow the app shell: no boards group without an assigned board, and privacy center only for owners and admins.
Where to start
Register the mini program id under ConsoleMini program analyticsConfigurationSDK code. After install, confirm $mp_launch on Realtime, then read scenes in Scenes. To line the mini program up with a website, see Multi-platform overview.
Scenes
The code is $scene on the launch. It is not the install channel $channel. Share returns are in Pages & sharing.
The app channels page can also switch to scenes when both kinds of source exist and no platform is selected.
What the report shows
The math matches channel analysis in Versions & channels: average daily new users, next-day retention, and share of active users, with a matrix of new users against next-day retention. Missing scenes are labeled (no scene).
Known common codes display as the number plus a name, such as the discover entry, a one-to-one chat card, or a scan. The list is a frequent subset for labels, not the full official catalog of every platform.
An extra table, Tagged links / QR codes, groups sessions and users by utm_source and utm_campaign. Those are parameters you put on the path, unrelated to $scene, so you can see both "opened from a scan" and "the link carried this campaign." If the header has selected one mini program, the table is only that app id.
Limits
Codes missing from the label list are shown as the raw number. Alipay and Douyin codes that do not match the list also stay numeric. That does not change the count. Next-day retention has no point until a day has elapsed. Realtime in the mini program view also ranks scenes over the last 30 minutes. That short window is not this daily table.
Pages & sharing
ConsoleMini program analyticsPages & eventsPagesConsoleMini program analyticsScenes & sharingSharingPages
In the mini program view the column is the page path, from $mp_view_screen. Time on page comes from $mp_page_leave. A path is the mini program's own path string, not a website host plus path. Neighboring paths in the same session become a flow; paths do not connect across sessions. The same page in the app shell groups by screen name. See Screens.
The top row has the most views, not the longest stay. Read average time for that.
Sharing
The metrics are share count, reopens that carry $share_distinct_id, distinct sharers, return rate, and $share_depth. The chain table is by path.
Return rate divides returns by share count, not by sharers; one person forwarding many times adds to the denominator each time. Average depth is weighted by hop. It is empty when the depth field is missing. Without a mini program source, the page says sharing applies only to mini programs.
Scene entry (a scan, a chat card) is Scenes. Do not mix that code with share depth.
Limits
Both pages follow the header date range, and exports keep the current filter. Neither page offers a previous-period comparison.
Users, retention, devices, stability
All platforms in this header still means mini programs only. Android stays in app analytics. The two shells keep their filters on their own paths, so neither overwrites the other.
Console menu
ConsoleMini program analyticsUsersUsers| Item | Group | Difference from app analytics |
|---|---|---|
| Users, Retention | Users | Splits use scene values, not install channels |
| User composition, Engagement | Users | Same rules. Engagement counts pages instead of screens |
| Devices | Devices & quality | Still model, OS, network, and place |
| Stability | Devices & quality | Still crashes, not-responding, freezes, and error groups |
There is no Versions item here. If $app_version is present, Users can still split by version. Attribution is not in this sidebar.
Reading the numbers
Starts add $mp_launch, $mp_show, and $app_start together, so returning from the background after the session timeout counts again. New users are still first seen in range. A reinstall is not new. Scene labels change the display name only, not the headcount. See Scenes.
The formulas are in App metric definitions. Composition and engagement read the same way as in app analytics. See Composition and Engagement. Whether each SDK reports crashes by itself is in SDK overview. Before the start rollup exists, the stability page asks you to wait for the daily rollup or recompute.