Summary
The weekly Ta-Da! site stats appear to undercount Android usage. In practice, Android users are active, but the weekly stats message is not reflecting that activity in its sessions figure.
Observed
The weekly stats report sent on 2026-06-29 said:
Traffic stayed very quiet this week: 8 pageviews from 3 visitors across 4 sessions.
That looks too low given known Android usage, especially from installed/home-screen PWA use.
Expected
Weekly stats should include Android user sessions the same way they include desktop/browser sessions.
Suspected scope
This looks like an analytics collection gap rather than a rendering problem in the report itself.
Likely places to inspect:
- Umami tracking in the client app, especially installed Android / standalone PWA launches
- whether Android home-screen launches create/record sessions correctly
- whether route changes or app resumes in the PWA are being tracked consistently
- whether the weekly stats job is querying Umami in a way that excludes or collapses Android traffic
Pointers
- Analytics client hook:
app/plugins/umami.client.ts
- Env/docs worth double-checking while here:
docs/ENVIRONMENTS.md (the documented NUXT_PUBLIC_UMAMI_HOST example includes /script.js, while the plugin currently appends /script.js itself)
Acceptance
- Android usage appears in weekly stats without obvious undercounting
- Installed Android PWA sessions are represented in Umami and in the weekly summary
- If the issue is config rather than code, document the correct production env shape clearly
Summary
The weekly Ta-Da! site stats appear to undercount Android usage. In practice, Android users are active, but the weekly stats message is not reflecting that activity in its
sessionsfigure.Observed
The weekly stats report sent on 2026-06-29 said:
That looks too low given known Android usage, especially from installed/home-screen PWA use.
Expected
Weekly stats should include Android user sessions the same way they include desktop/browser sessions.
Suspected scope
This looks like an analytics collection gap rather than a rendering problem in the report itself.
Likely places to inspect:
Pointers
app/plugins/umami.client.tsdocs/ENVIRONMENTS.md(the documentedNUXT_PUBLIC_UMAMI_HOSTexample includes/script.js, while the plugin currently appends/script.jsitself)Acceptance