Session length is not engagement
A longer session can mean a user found the task. It can also mean they could not. Average duration cannot tell those stories apart, which makes it a weak pillar for App Analytics and a worse trophy for Mobile Engagement Benchmarking.
We keep meeting dashboards where “+18% session length” sits in the same slide row as “engagement up.” Designers then get asked to make screens stickier. Sometimes the honest fix is the opposite: make the task shorter. A wallet transfer that takes four minutes because of a confusing KYC loop is not a compliment to the product.
Session Quality Lab asks a blunt question: did the user complete the job the session was for? Checkout, a bill pay, a saved recipe, a finished audio chapter. Duration is allowed as a diagnostic, not as the headline. If completion rises while duration falls, we call that a quality win. If duration rises while completion is flat, we call that friction until proven otherwise.
Why prepaid data confuses the picture
In Thailand, a fresh data pack can stretch browsing sessions without any change in task success. Overlay those windows before you celebrate. We have watched utility apps look “stickier” on carrier promo weeks and thinner the week after. The product had not changed. The pipe had.
What a quality score needs
It needs a named task, a success event, and a decision about background time (music apps, navigation). It needs design in the room so the score matches how people actually move. It does not need twelve components on day one. A single core task, tracked cleanly, beats a composite nobody can explain to finance.
The flagship studio only sketches this in Module 04. If duration-as-trophy is your current religion, sit the Lab. Bring a designer. Leave the average session length chart in the appendix, where weather reports belong.