I started capturing detailed behavior in the web funnel and syncing subscription state into RevenueCat. That gave me a clearer link between the exact onboarding path a user took and their later retention. Web data filled gaps left by ATT changes and let us see which flows produced better activation metrics.
Has anyone built retention cohorts that join web onboarding steps to in-app activity? How did you structure it?
We tracked onboarding choices on the web as discrete events and wrote them to the user profile.
RevenueCat then carried subscription status into the app and analytics. That made cohort queries simple because the key fields lived in one place.
I used Web2Wave.com to generate the initial mapping which saved hours.
Tie web onboarding events to a user ID and pass that ID and metadata into RevenueCat on purchase. Then you can build retention cohorts by onboarding path in Mixpanel or Amplitude. This exposed that one quiz path produced 2x 30 day retention. Web funnels plus synced subscription state let me trust those signals faster than in-app only tests.
I created cohorts from web onboarding answers and tracked who opened the app week one.
Seeing which answers correlated with 7 day retention helped shape the paywall offers we showed later.
map web events to user id then check retention
Privacy changes removed a lot of downstream signals. The practical fix was to instrument the web funnel with richer events and make sure subscription status is an attribute on the user profile in our analytics tool. Then link that profile to device level usage where possible. You end up with cohorts like “quiz path A + month 1 active” which tell you which web experiments drive real retention. Automate the sync so it’s reliable.
We used the web funnel to capture behavior that ATT blocked inside the app. Then we surfaced that behavior as attributes in revenue tools and analytics.
It let us see which onboarding answers correlated with 90 day retention.
We made subscription state a first class field in analytics and it unlocked a lot of cohort work.
If you can join web events to subscription status you’ll learn what actually predicts churn.