Engagement
Volume and quality did not move together. Device mix explains part of the Paid Social trough for me: mobile is the majority, and Android WebView is more than a third of users.
49.1%
Engagement rate in this slice
30s
Session-weighted avg time
50.3%
Best month in slice
Jul 2026
45.1%
Weakest month in slice
Sep 2026 MTD
Rate and time
View series as a table
| Month | Engagement rate | Avg time (s) |
|---|---|---|
| Jul 26 | 50.3% | 42 |
| Aug 26 | 49.5% | 25 |
| Sep 26* (MTD) | 45.1% | 21 |
What I noticed. February 2026 is 64.6% when present. April is 37.5%. July 2026 time is 42 seconds when paid recedes.
How I am reading it. A session increase that coincides with Paid Social usually looks like a quality decrease to me. Time recovering when paid disappears looks like a healthier mix, though one month is not enough for me to call it a new baseline.
What I would look at next. I would keep engagement beside sessions in every paid review.
Device (full window)
| Mobile | 54.4% | 39.1% | Majority device, weaker engagement |
| Desktop | 43.8% | 65.2% | Higher quality |
| Android WebView | 36.1% | 35.4% | 9s average time |
| Chrome | n/a | 63.9% | 27s average time |
Where I would go next
| What I noticed | The numbers | How I am reading it | What I would do next |
|---|---|---|---|
| Mobile is the conversion constraint | 54.4% of users, 39.1% engagement vs desktop 65.2%. | The majority device is getting the weaker session. | I would fix mobile landings before buying more in-app traffic. |