#experiment

8 notes tagged “experiment”.

• External payment links vs in-app purchase (US, 2025+)RevenueCat's Dipsea test after the 2025 US ruling — web-only checkout earned $1.96/user vs $2.09 for IAP at a 30% fee (≈ a wash), fewer trials but higher trial conversion and far lower auto-renew-off (3.5% vs 19%); at the 15% fee IAP wins; Skylight saw +100% conversion moving a QR flow from web to Apple.
• Free-trial toggle — let users choose trial vs pay now (and price them differently)Put a "skip free trial" toggle on the paywall so users pick trial or immediate purchase — Riz charges $9.99/week with a 7-day trial or $6.99/week without; Cal AI keeps one price; on transaction-abandonment offers give a deeper discount (90% vs 80%) for buying without a trial; ~10% pick no trial.
• Paywall before sign-up (account creation after purchase)Moving a forced sign-up screen after the paywall raised paywall views from 40% to 94%, trial starts +190% and sales +52% in a fitness app.
• Raising subscription prices (Skylight $39 → $79)How Skylight doubled its subscription from $39 to $79/year with minimal loss of attach or retention — price tests (incl. hardware discount), choosing below the ARPU-max $99, grandfathering existing subscribers, honouring the old price on request.
• Removing the free trial (pay-up-front plans)When dropping the free trial wins — Natal's no-trial monthly grew monthly subs 2,000% (quarterly +46%, annual +21%), Claim went paid-only after 34% trial starts and 50–55% trial-to-paid — and the prerequisites, plus counter-cases (Ladder, Skylight).
• Send paid traffic to a web quiz or straight to the App Store?Evidence from Built With Science (90% of paid traffic to a long web quiz after it beat direct-to-store head to head) and Simply (web flows made Facebook work for the first time, ~20% of Simply Piano revenue) on when a web quiz outperforms an app-install campaign — lean-back social users, older non-app-native audiences, bundles.
• Showing two paywalls in onboarding (and the 8-paywall-views rule)A no-trial paywall early plus a trial paywall later in onboarding lifted start rate from ~3% to 7.7% and monthly sales ~3x; paywall views keep paying off until roughly 8 views.
• Web pricing page — show annual first with a trial timeline (60/40 → 75–80% annual)Built With Science's biggest 2026 win — a web pricing page showing only the annual plan with a Blinkist-style day 1 / day 7 / day 12-reminder trial timeline moved plan mix from ~60/40 to 75–80% annual with no price change; adding more context hurt; a "pay now, 20% off" no-trial toggle lost until both options were framed positively.