Docs

App Store Connect's New Cohort Filters: What to Track

App Store Connect now builds cohorts by download date, source, and offer start date with up to seven filters — free retention data most teams paid for.

Analytics

Apple’s overhaul of Analytics in App Store Connect added cohort analysis by download date, download source, and offer start date, with up to seven filters stackable on any metric view. That’s retention-by-acquisition-attribute analysis — “how does the region-B cohort’s day-30 retention compare to region-A’s” — that used to require exporting raw events into Amplitude or Mixpanel to answer. For teams already paying for a third-party product analytics tool, the free version living inside App Store Connect changes what’s worth building versus what’s worth just reading off Apple’s own dashboard.

The feature is genuinely useful, but it has a boundary worth knowing before a team restructures its reporting around it: Apple aggregates cohort data to protect user privacy, which means small cohorts get suppressed or bucketed, and there’s no row-level export to join against other data sources. It answers “what happened to this cohort” cleanly, but it can’t tell a team why, and it can’t be joined against a CRM record, a support ticket, or a custom event your own SDK captured. Treat it as a fast first read, not a replacement for an analytics pipeline that needs to answer causal questions.

Data Points to Track

  • Cohort definition fields used (download date, download source, offer start date), logged per report pulled, so retention comparisons across weeks are actually comparing like-for-like cohort definitions rather than silently shifting scope
  • Filter combination applied (up to seven at once), recorded alongside any screenshot or export, since a filtered view that isn’t documented is unreproducible the next time someone asks how a number was generated
  • Minimum cohort size before Apple suppresses or aggregates the row, tracked empirically by testing small filtered segments, so the team knows where the tool’s privacy floor sits before relying on a thin cohort’s numbers
  • Retention curve shape by acquisition source, not just the headline day-1/day-7/day-30 figures, since two sources can share the same day-30 number with very different drop-off shapes in between
  • Delta between App Store Connect’s cohort retention and your own first-party retention tracking, checked periodically, because the two won’t always agree — Apple’s counts what re-opened the app, your own SDK may count a more specific engagement event

Setup Steps

  1. Baseline your own first-party retention numbers first, before treating App Store Connect’s cohort view as a source of truth — you need something to reconcile it against.
  2. Pick a small standard set of cohort filter combinations your team will check weekly (e.g., download source × region, offer start date × platform), rather than ad hoc one-off filters that don’t build a comparable history.
  3. Screenshot or export each cohort view with its filter state visible, since App Store Connect’s UI doesn’t version or save filtered views for later comparison.
  4. Test the suppression floor directly by filtering to a deliberately small segment and noting where Apple stops returning a broken-out number, so the team knows the tool’s limits before a real analysis runs into them.
  5. Route anything requiring row-level joins — support tickets, CRM data, custom events — to your existing analytics pipeline, keeping App Store Connect’s cohort view as the quick acquisition-attribute check it’s built for.

Actionable Insights

A cohort acquired through a specific download source retaining meaningfully worse than others by day 7, visible directly in the App Store Connect filter, is a fast signal to pause spend on that source before waiting for a full LTV model to confirm it. An offer-start-date cohort with strong day-1 retention but a steep day-7 drop suggests the offer itself is attracting users who churn once the introductory period ends, worth checking against your own subscription-renewal tracking. And if App Store Connect’s cohort retention consistently reads higher than your first-party numbers for the same population, the gap is worth investigating rather than picking whichever number is more flattering — it usually means the two tools are counting different definitions of “returned,” not that one of them is wrong.

Expert help

Need help tracking this in your app?

Our team sets up analytics pipelines for mobile and web teams every day. Talk to us and get your first events flowing in under an hour.

Talk to an expert