I got tired of waiting on app releases just to test a price change. We moved onboarding and the paywall to the web and kept the native app mostly for the product experience.
The flow now:
- Ads and in-app promos deep link to a web onboarding sequence with a paywall and checkout.
- We flip offers on the web (prices, trials, copy, bundles) without a new build.
- After purchase, our backend creates/updates the subscriber in Adapty/RevenueCat, and the app unlocks on the next launch. We also keep a simple in-app paywall as a safety net.
What worked tactically:
- We run 1–2 tests per week and avoid overlapping major events. We target ~1k–2k qualified sessions per variant or two business cycles before reading it.
- Variables that mattered: annual vs monthly framing, trial length vs intro price, guarantee language, and how we display savings. Trust badges and FAQs helped more than I expected.
- We tag campaigns with UTMs on the web and pass a token into the app so we can tie installs back to the original session.
Curious how others run this: what do you consider a clean sample size, how do you guard against seasonality, and which offer tweaks gave you the biggest lift without spiking refunds?
I shipped the same setup.
Web onboarding with a web paywall. RevenueCat handles entitlements. The app just checks status.
I used Web2Wave to generate the funnel JSON and their SDK reads it. Changes go live without builds.
Keep variants simple. One lever at a time.
I treat web funnels like a live lab. Spin a variant, ship it in minutes, measure, then replace losers fast.
Web2Wave lets me tweak copy, price, and layout on the web, and the app reflects it instantly. I aim for weekly cycles with clear stop rules.
Keep the test list short.
Pick one price lever and one trust element.
Run it for a full week so weekday and weekend behavior evens out.
Then switch. Smaller steps made it easier to see what worked.
Test fewer things faster on web
Define the decision before launching the test. If you need 10%+ lift to switch, put that in writing.
Split traffic server-side and log assignment. Don’t change creatives mid-test. Use UTMs and a user token to link web to app.
Run at least one full pay cycle if you test trials. Check refund and chargeback rates alongside conversion. Winners should raise paid retention, not just front-end conversion.
Two quick plays that worked:
- Intro price beat free trials for us. Lower immediate churn and fewer support tickets.
- Short copy wins on mobile. Keep bullets above the fold and push FAQs below.
We only test one money lever at a time.
For seasonality, we use paired weeks. Variant A runs week 1 and 3, Variant B runs week 2 and 4, same budgets and geos. It reduced false wins. Also, always cap discount tests so they don’t leak into organic traffic.
Try testing trial length and headline first. Easy wins usually.