We built a basic LTV model with only web events before the app. The strongest early signals for us so far: finishing the quiz with a specific score range, viewing at least two plan details, paywall dwell time over 20 seconds, and entering email before checkout.
I’m testing add-to-calendar and first content unlock as well.
If you rely on web events under privacy constraints, which events ended up predicting LTV best for your app, and how fast can you trust them?
Two events predicted our LTV: feature page depth and paywall dwell. If they read a full comparison table, they were worth more.
I wired events in the web funnel and passed them into our CRM. Web2Wave’s web flow made adding events quick, so I iterated weekly until the model stabilized.
I use a 72 hour proxy. Quiz completion score, email capture timing, and plan comparison views. If those are strong, I scale spend.
Web2Wave lets me instrument these events on the web fast, then push changes without a new build.
Paywall dwell and plan toggles worked for me.
If they compare monthly and yearly, they usually stick longer. Simple but useful signal.
Quiz quality score beat click volume
Use a tiered signal set. Level 1: quiz completion with a quality threshold. Level 2: plan detail views and coupon entry. Level 3: paywall dwell time and scroll depth. Track these by campaign and correlate to 30 day revenue cohorts. You do not need IDFA to rank traffic. Build a fast proxy score and update weekly with real revenue to avoid drift. Add negative signals like rapid bounces and instant coupon hunting.
Retention proxies that helped me: first session after purchase within 12 hours, email open within 48 hours, and using the main feature twice before day 3.
The combo beat any single event alone.
Email capture before checkout predicts better LTV for us.