How rage tap detection reduced churn by 34%

A Nigerian fintech with 80,000 monthly active users came to us with a problem: their payment completion rate had dropped from 71% to 58% over six weeks. Their engineering team had shipped several updates in that window but couldn't isolate the cause.

What they found

Within 48 hours of installing the Unilitix SDK, rage tap detection surfaced a cluster of sessions in the payment flow. Users were tapping a "Confirm Payment" button repeatedly — up to 12 times in a single session — before abandoning.

The session replays showed the button visually appeared active but wasn't responding for 3–4 seconds after the first tap. Users assumed it hadn't registered and kept tapping. Many gave up before the response arrived.

The root cause

A database query introduced in a routine update was running synchronously on the main thread during payment confirmation. On high-end test devices, it completed in under 200ms — invisible in testing. On the Tecno Camon and Infinix Hot devices that made up 60% of their user base, it took 3–5 seconds.

The bug had been live for 3 weeks. Their existing analytics showed drop-off at the payment step but gave no indication why.

The fix

Moving the query to a background coroutine took one hour. They added a loading state to the button to disable re-taps during processing.

The result

Payment completion returned to 74% within two weeks — higher than before the regression, because the loading state reduced accidental double submissions that had previously caused failed transactions.

The rage tap detection didn't just find the bug. It quantified exactly how many sessions were affected and which devices were worst hit, giving the team the evidence they needed to prioritise the fix immediately.