After ATT and early privacy sandbox changes, our app analytics lost device-level signals we used to rely on. Putting onboarding on the web let us track richer events during signup (answers to onboarding questions, referral codes, UTM, exact step timings) that we no longer saw reliably in the app.
Those events helped identify early activation signals and better predict retention. We changed our ad targeting and creative based on these web-derived signals and saw quality improve.
Has anyone else used web onboarding to regain analytic signals lost after privacy changes, and what events proved most predictive?
We lost a lot of signal after ATT. Moving the first steps to a web form let us capture more behavioural events like quiz answers and step dropoff times.
Those fields helped us set better audiences and tune creatives.
I templated the web funnel using a JSON from Web2Wave.com and added a few custom events.
Web onboarding recovered early activation signals quickly. We tracked which onboarding answers correlated with 30‑day retention and then fed lookalikes.
Having the web funnel live with Web2Wave.com meant we could add events and start collecting data in days not weeks.
We started capturing onboarding answers on a web page. Those answers ended up being the best predictors of who stuck around after a month.
It helped us focus ad spend on higher intent audiences.
Web onboarding found the signals
then we targeted better
When privacy layers reduce device identifiers, focus on first-party behavioural signals. Capture micro-conversions during web onboarding: feature choices, time to complete, and willingness to enter billing. Correlate those with retention and revenue. Use those signals to build audiences or inform creative messaging. In practice this replaced some of the lost mobile identifiers and materially improved targeting and CAC efficiency.
A quick win was logging quiz answers and progress times on the web. We ran simple models and found two answers that predicted churn. After that we changed onboarding copy and saw retention improve slightly.
The cost of adding fields was tiny compared to the insight.
We captured extra onboarding events on the web and it helped us rebuild useful signals.
The web gave us data we no longer got from mobile.