Critique · 5 December 2025
When n-day retention misleads the product meeting
An n-day number is a compression. Compression is useful until it hides three different arguments: when the clock started, what “came back” meant, and whether the calendar was comparable. Product meetings that treat the number as a single fact will rewrite roadmaps on a misunderstanding.
In critique we unpick the number in a fixed order. Students are not allowed to jump to feature hypotheses until the window itself has been interrogated.
First: which n, and why
Day-7, Day-14, and Day-28 are house windows at Proxy Cloudhub because they are common enough to compare across teams, not because they are sacred. If your product’s meaningful return cannot occur inside that n, the number is decorative. Say so before you discuss onboarding copy.
Second: which event returned
Opened the app, played a track, passed the paywall, completed an order — these are different returns. A subscription audio team in the atelier mixed trial-start with first-open inside one Day-14. Paid conversion looked fine; returning listeners after trial looked weak. Splitting the clocks ended a two-month argument about paywall wording that was never the main issue.
Third: which days were comparable
A holiday week, a one-off campaign, or a timezone cut can move n-day retention without any product change. Thailand seasonality is the example we use most, but the principle travels: if the calendars do not match, the delta is not yet a product story.
What we ask leadership to do
Ask for the paragraph, not only the percentage. If the analyst cannot read clock-start, return event, and calendar notes aloud, the meeting should stop. That sounds severe. It saves more roadmap time than another hour of speculation.
This unpicking is the spine of our method and of the flagship atelier. If you already have a number that is driving a decision this month, send it to the studio and say which n you are quoting.