Sessions · 10 min
Reading session depth without vanity metrics
“Average scroll: 72%” is a costume. On a phone it often means the shopper’s thumb was bored, not that they considered a size.
Desktop heritage taught us to love time-on-page. Mobile catalogues punish that love. A long session can be a person waiting for a Line reply with the PDP open in the background. A short session can be someone who knew the SKU, opened the size guide, and paid.
In class we replace scroll percentage with a small set of intent taps: size guide, fabric note, delivery promise, and variant change. Those four are not universal. A grocery app may care about substitution preference instead of fabric. The principle holds: count the tap that would change a warehouse or a buyer’s mind.
How to argue with a dashboard vendor
Ask where session timeout is set. Many tools still use thirty minutes, which turns a backgrounded app into fake depth. Ask whether a screen view fires on resume. Ask whether video autoplay on a live replay counts as engagement. If the vendor cannot answer without a CSM, you already know the number is decorative.
We then build a one-page depth note: for a single SKU family, show intent taps versus add-to-cart, split by new versus returning, split again by Bangkok evening versus daytime provincial. The chart is ugly. The merchandiser can use it.
Where this fails
If your app is a thin wrapper on a desktop site, the tap targets may not exist as events. Instrument first; do not invent depth from heatmaps. Heatmaps are a conversation starter, not a metric.