Journal · 12 Mar 2026
Retention after a listing refresh
A listing refresh that lifts conversion and drops D1 is not a mixed blessing. It is a mismatch between what the store promised and what the first session delivered. We treat it as a single story in App Store Performance Analytics, not two dashboards that happen to share a week.
Consoles make the split easy to miss. Conversion lives on the store listing card. Early retention lives in a product analytics tool, often with a different timezone and a different definition of “new user.” If your measurement spec does not name both owners, the screenshot team will celebrate and the onboarding team will file a separate ticket three weeks later.
What we ask before calling it noise
First: did the creative change the promise? Frame one that suddenly leads with bill-pay, cashback, or a festival skin will recruit people who wanted that object. If onboarding still opens on a generic wallet tour, D1 should wobble. That is not seasonality. That is copy doing its job too well.
Second: did the audience mix change? A conversion lift driven by a new query cluster can look healthy while the cohort is simply less qualified. We look at impression source notes where the console allows it, and we do not average Thai and English locales into one conversion cell during the window.
Third: did the product ship something else the same week? Listing tests that coincide with an auth rewrite are unreadable. We mark those cards “contaminated” in class and refuse to let them into a board pack as evidence.
How it appears in a board pack
We write three short paragraphs: what changed on the listing, what happened to conversion, what happened to D1. We include the pre-registered gate from the experiment card. If conversion was the only gate, we say so — and we say that was a design error, not a mystery in the data.
The wallet case in our voices page rolled back frame one after fourteen days because the dual gate failed. They kept a quieter bill-pay frame in position three. Conversion gave back some of the lift; D1 recovered. Finance received the episode in plain English instead of a buried footnote.
What this is not
It is not a lecture on cohort retention math. It is not a demand that every listing test wait for D30. D1 is a listing signal when the promise changed. D7 is useful when onboarding was also touched. Past that, you are doing product analytics, which is a different room.
If you want the desk version of this, it sits in module 04 of the ASO Measurement Studio and in the shorter Retention Signal Reading seminar.