A/B Test: Payment Screen With Named Load Steps

Analytics showing visitors leaving before the transaction finishes, session recordings of shoppers refreshing the loading screen, or support contacts asking whether a payment went through are signs that shoppers find the wait during payment unclear. If the step display works, fewer shoppers leave while waiting, with actual processing time unchanged. If shoppers leave because of slow infrastructure or frequent failed transactions, look at those first, and make sure the loading screen does not bypass or hide security checks.

Payment-step abandonment rate

Among visitors who see the step-based progress screen during payment, compare those who leave before the process finishes.

Also track

• Perceived wait time (survey)If surveyed, measure perceived wait time; abandonment alone does not measure trust.
• Actual processing timeMake sure the step screen does not increase actual processing time.
• Support after an errorTrack support requests after failed transactions.
• Payment success rateMake sure overall payment completion does not fall.

• Progress displayShow steps tied to real transaction events instead of a spinning icon.
• TimingKeep processing, verification and error handling intact; do not add artificial waiting.
• Perceived waitMeasure perceived waiting separately from actual processing time.
• ErrorsShow existing error states promptly when a transaction fails.
• ConnectionReview mobile and slower connections without changing the progress display during the test.

1. 1The payment-step abandonment rate is the primary measure, but lower abandonment does not justify shipping if the payment success rate or actual processing time gets worse. Lower abandonment alone does not show more trust; rely on the survey and direct feedback before saying anything about trust or perceived wait time.
2. 2This test changes only how the loading screen looks. Hold the number of payment steps fixed in both versions, or the screen cannot be credited with the difference. Look at mobile and slower connections on their own, check that support contacts after a failed transaction do not rise, and validate the result again before reusing it on other waiting screens.
3. 3Abandonment is counted only while payment is processing, so it may be rare; the sample size calculator shows how much traffic each version needs before a small change in such a rare event becomes visible. Confirm too that the number of visitors who saw the loading screen in each version matches the planned split (sample ratio mismatch); a mismatch can mean the screen-view event is under-recorded in one version. Should the gap remain unclear after these checks, do not conclude that the step display leaves abandonment unchanged.