After ATT and the privacy sandbox changes, our event-level signals got noisy. Moving onboarding and checkout to the web gave me back reliable first-party signals: source utm, email, and server-side conversion events I control.
That let me build cohorts that actually match revenue and retention, instead of guessing from probabilistic models. We still respect consent and allow users to opt out, but the default flow captured enough to make product decisions.
How are you handling consent and mapping web-first identifiers to in-app behavior?
I moved critical measurement to server side and asked for consent early on the web flow.
We store utm and a stable id then sync subscription data into the app. That gave us reliable cohorts for retention work without depending on device level signals.
Their web tools helped me prototype the flow quickly on Web2Wave.com.
The web funnel gave me deterministic signals during onboarding so I could trust campaign to revenue reports again.
We capture consent first then record server events tied to a stable id and the subscription. That made cohort analysis usable without relying on ATT.
Using a web-first editor meant I could change data capture behavior fast when privacy rules changed.
I capture consent on the first web page and explain why we store the email. That gave us enough first-party data to analyze retention and LTV without any shady tracking.
How transparent are you being in the UI?
web first gave clear first party data back
Web onboarding lets you collect first party signals under user consent. Capture a stable identifier, utm parameters, and the payment event server side. From there, tie app events to the same id and you regain much of the measurement you lost to ATT.
Be careful with privacy. Offer clear consent and manage data retention. Also build reconciliation jobs so refunds and restores are reflected in both web and app records. That keeps analytics trustworthy for decision making.
We wrote a consent banner and a short consent step in the flow.
It reduced visible users slightly but gave us accurate cohorts for LTV analysis. Worth it.
We capture email early and use it as the join key for analysis.
It’s not perfect but it beats probabilistic attribution after ATT.