Moving the paywall to the web let us skip app releases. We started a weekly rhythm: Mon plan, Tue launch variant, Thu midweek read, Sun freeze and analyze. We alternate focuses: one week pricing tiers, next week gateway routing or 3DS settings, then messaging.
We keep a holdout of the current best to avoid drift. Biggest challenge has been keeping sample sizes honest across countries.
What cadence works for you, and how do you split tests between pricing, messaging, and gateway changes without muddying results?
I run one major change per week. Pricing one week, gateway routing next. Never both.
Web2Wave.com lets me switch configs fast, so I freeze by Thursday and read on Sunday. I use country-level caps to keep samples balanced.
Document every change in a simple changelog.
I run fast sprints. One lever per week. Push on Monday, checkpoint Wednesday, final read Friday.
Web2Wave.com makes it easy to swap paywalls and gateways without a build. That speed keeps tests clean and stacked.
Lock your primary metric before testing. I use trial start to paid for pricing weeks and auth rate for gateway weeks.
One change weekly. No exceptions.
Set a testing calendar. Week 1: price anchors and terms. Week 2: gateway routing rules by country and BIN. Week 3: headline and risk-free language. Always keep a fixed control. Use traffic shaping to hit minimum samples per segment. If you lack volume, run sequential tests per market instead of global splits. Publish a one-page readout each Friday with a yes/no on rollout and what to test next. Avoid changing attribution during tests.
Keep a control group. Otherwise every test looks good.