Personalising the paywall with onboarding signals

TL;DR

• Most paywalls are one tested winner shown to everyone; few apps personalise. The onboarding knows why the user came, the paywall “never finds out”.¹²
• Fastic (weight loss): 20–30+ playful onboarding steps with varied inputs (sliders etc.), ~80% completion; the paywall then shows the user’s target weight (“you want to get to 50 kg?”).³⁴
• Even the user’s name on the paywall is a start; implicit signals (screens or features touched in the first session) work too.⁴
• Source signals: the Apple Ads search term or ad group the user came from is intent you can pull into the app to choose the paywall.²
• Give a personalised artifact before the ask: Gleam’s onboarding ends with a persona scorecard, what to improve and suggested first courses — proof the questions weren’t just to get a checkout.⁵
• Validate every personalisation with an A/B test — “your mileage may vary”.⁶

Details

Why it works. Users now expect long onboardings and notice when answers change nothing; personalising “wows” them and builds trust (Kelvin Zhang).⁵ Host Freddy Fang says Adapty’s study of top-converting onboardings found personalisation was the number-one principle — “this app was tailor-made for me”.⁷

Signal types to use (Andy Carvell’s framework, Integrated growth — moving user signals between store page, ads, paywall and CRM): intent (goal, target weight, the keyword searched), engagement (features tried), value (precursors of purchase).⁸

One paywall per entry point (2022). Dan Burcaw: with several custom product pages, it makes sense to run several paywalls aligned with each page’s message or campaign — and let marketers change them without an app release.⁹

Implementation sketch.

1. Store onboarding answers and attribution (keyword/ad group/creative) on the user.
2. Template paywall headline and visuals from the top answer (“Reach 65 kg by March”).
3. Map acquisition intent → paywall variant (e.g. “edit PDF” keyword → editing-focused paywall).
4. A/B test each against the generic winner on net revenue (Judge paywall tests on net revenue after refunds (matured cohorts)).

Disagreements / caveats

• Long onboarding isn’t free: see Long or short onboarding before the paywall? (avoid the middle) — Fastic’s 80% completion depends on making each step engaging.
• Explicit “just for you” labelling can backfire (TripAdvisor example in that note); personalise quietly.
• No paywall-lift numbers were given for Fastic or Gleam’s personalisation.

See also

• Paywall content — sell the outcome and add context (long paywalls)
• Hard paywall only for paid traffic (gating by "How did you hear about us?")

Footnotes

1. Why your paywall is failing (and how to fix it) | Andy Carvell, Phiture @ 00:00 — Andy Carvell ↩

2. Why your paywall is failing (and how to fix it) | Andy Carvell, Phiture @ 15:46 — Andy Carvell ↩ ↩²

3. Why your paywall is failing (and how to fix it) | Andy Carvell, Phiture @ 17:01 — Andy Carvell ↩

4. Why your paywall is failing (and how to fix it) | Andy Carvell, Phiture @ 18:13 — Andy Carvell ↩ ↩²

5. How two ex-engineers built a $140K/month social skills app | Gleam AI @ 42:51 — Kelvin Zhang ↩ ↩²

6. Why your paywall is failing (and how to fix it) | Andy Carvell, Phiture @ 19:03 — Andy Carvell ↩

7. How two ex-engineers built a $140K/month social skills app | Gleam AI @ 43:37 — Freddy Fang ↩

8. Why your paywall is failing (and how to fix it) | Andy Carvell, Phiture @ 23:56 — Andy Carvell ↩

9. Why and How to Develop Subscription-Based Apps with Dan Burcaw of Nami ML @ 10:12 — Dan Burcaw ↩