Skip to content

Reading your analytics

Overview is the screen you land on. It groups its numbers so you can read them in a sensible order rather than scanning a wall of tiles.

Group Answers
Traffic How many people came, how many were new
Engagement How much of your storefront they actually looked at
Session Quality Whether visits were meaningful or glancing
Conversion How many moved toward buying
Commerce KPIs What that was worth
Data Quality Whether to trust the rest of it

Start at Data Quality if a number looks surprising. It is quicker than investigating a figure that was never reliable.

Most are familiar from any analytics tool. A few are specific to a swipe-first storefront, and those are the ones that tell you something new. The names below are exactly as they appear on screen.

Details Opened Rate. How often someone went past a Feed card into the cube behind it. The closest thing to “did the card do its job” — a card that looks good but is rarely opened is not working.

Detail Retention Score. How far into a cube people typically get. It averages how many sessions are still present at each successive card, across the depth most sessions reach. Low retention means your detail loses people; the session view shows where.

Avg Swipes (Mainfeed / Details). How many detail cards a session goes through on average once it opens something.

Swipe Back Rate. How often people go back. Occasional is normal browsing. A lot of it can mean people are overshooting something they wanted.

Product Interaction Rate and Gatekeeper Interaction Rate. How often people engage with product controls, and with gating steps.

Bounce Rate and single-cube sessions. Visits that did essentially nothing. Lower is better for both.

Tracking users without events. A data-quality measure — visitors recorded with no activity attached. If it climbs, treat the other numbers with suspicion and check your setup.

Every metric can be compared against the previous period, and the change is coloured by whether it is good news — which is not always “up”. Bounce rate, single-cube sessions and tracking-without-events all count as improvements when they fall.

Pick a range long enough to mean something. On a storefront with modest traffic, a single day tells you very little, and two days rarely differ for reasons worth acting on.