set-up-ab-testing
Installation
SKILL.md
Set up A/B testing
Experimentation velocity is the single most-cited differentiator between top-grossing subscription apps and the rest. Cal AI ran 123 experiments across 46 trigger points (~5 real experiments/month); Nami ML reports top performers run "hundreds of experiments across all funnel pages." (Superwall, Lindsay Giachetti on LinkedIn)
A team that runs dozens of experiments per month compounds small wins into outsized revenue. A team that relies on opinion and ship-and-pray does not.
What to test (by funnel stage)
| Stage | Experiments that move the needle |
|---|---|
| Onboarding | Number of steps; quiz question order/wording; "building your plan" animation; personalization depth; pain-amplification copy |
| Paywall | Placement (onboarding_end vs feature_gate vs winback); price point; trial length (3 vs 7 vs none); anchor framing; social proof; trial timeline graphic vs text; hard vs soft vs hybrid |
| Pricing | Weekly vs annual default; decoy tier presence; lifetime option; regional price points |
| Retention | Streak mechanics; notification copy/send-time; re-engagement cadence |
| Review | Gate trigger; satisfaction-question copy; timing |
| Store listing | Icon, screenshots, short description (Google Play Store Listing Experiments; Apple Product Page Optimization) |