The paywall that kept changing, but never on purpose
On a travel app I worked with, the team had already tried tweaking the paywall a few times — new copy here, a different price display there. Nothing stuck, because nothing was actually being tested. Each change was a fresh guess, not a validation of the last one.
The fix wasn't a better paywall design. It was turning vague dissatisfaction into four specific, testable hypotheses.
Hypothesis, not redesign
Instead of asking "what would make this paywall convert better," the useful question was "what specific belief is stopping someone from converting?" That reframing produced four distinct hypotheses:
Unclear trial mechanics — users weren't sure when they'd be charged or how to cancel, so we split the paywall into a clear timeline before plan selection.
Pain-point mismatch — the copy listed features instead of naming the actual frustration (hours spent planning without knowing if it's any good), so we rewrote it to lead with that pain directly.
Missing social proof — as an early-stage product with no brand recognition, the paywall had nothing to lean on for trust, so we tested adding traveler reviews and a "join the club" framing.
Investment framing — the biggest hesitation was spending money before knowing the value, so we tested showing a direct cost comparison: planning alone vs. with the product.
Why this matters more than the design itself
Each hypothesis targeted a different psychological reason for hesitation. That's the part that's easy to skip — testing four visual variants of the same paywall tells you which one people prefer, but not why. Testing four hypotheses tells you which belief was actually blocking conversion, which is far more useful information for every future paywall decision, not just this one.
The winning variant lifted paywall conversion by 8% — a modest-sounding number until you remember it came from isolating one specific, provable cause rather than a general facelift.
The takeaway
If your paywall (or any conversion screen) isn't performing, resist the instinct to just redesign it. Write down the specific reasons someone might be hesitating, turn each into something testable, and let the data tell you which belief was actually true.
