Reading paywall screenshots like a researcher

Paywall Copy Lab is easy to misunderstand. People arrive hoping we will pick a prettier headline. We do not. We print the screen, including the footnote that design usually crops out of the Figma frame, and we ask what the screen claims in a court-adjacent sense: price, term, auto-renew, trial length, and what happens if the partner fails.

Thai and English versions are both in scope. A mismatch between locales is not a localisation bug for later; it is two different offers. If analytics counts both as the same paywall_impression, your funnel is already fictional.

Toggles are claims

An annual/monthly toggle changes the legal sentence even when the headline stays still. We mark whether the selected state is reflected in the confirm button, the small print, and the receipt preview. Several Bangkok education apps in recent labs had an annual toggle whose footnote still described a monthly trial. Product called it a “copy miss”. Counsel called it a problem. Instrumentation should have shown two impression variants; it showed one.

What researchers do not do

We do not vote on brand voice. We do not A/B a joke. We do not promise that a shorter headline will lift first charge. The lab’s job is to make the claim on screen match the event you will later defend to finance. If that feels slow, it is slower than a regulator letter.

Bring screenshots with secrets removed. Bring someone who can change the build. Leave the mood board at the office.

Back to the journal