Checkout Progress Bar A/B Test Example

Drop-off concentrated at the address or payment step, sessions that move back and forth between steps, or survey answers calling the flow unclear point to shoppers who are unsure where they are in checkout. If the bar removes that doubt, shoppers who see the remaining steps leave the address and payment steps less often. If people leave over hard form fields, payment errors or late costs, fix those before adding a progress indicator. Confirm before launch that the bar reflects the actual flow, adds no steps, hides none of the remaining ones and uses step names shoppers understand.

Checkout completion rate

Compare orders among visitors who start checkout and see the progress bar.

Also track

• Step abandonment rateTrack exits from the address and payment steps separately.
• Checkout completion timeMeasure time from checkout start to order confirmation consistently in both arms.
• Backtracking rateTrack returns to earlier steps; this alone does not prove indecision or confusion.
• Page load timeCheck whether the progress bar affects page load time.

• Progress informationAdd a bar showing completed and remaining steps.
• Keep constantKeep checkout steps, their order, fields and CTAs unchanged.
• Step abandonmentTrack exits from the address and payment steps separately.
• MobileKeep the information readable without covering the main action.
• Separate testsTest checkmarks, explanatory copy or indicator placement separately.

1. 1Checkout completion rate is the deciding metric, but read abandonment at the address and payment steps separately. Drop-off that merely shifts from one step to another while the total stays flat is not enough to ship the bar. A higher backtracking rate alone does not prove confusion either; shoppers may simply be checking their details.
2. 2The bar's effect depends on the number and order of steps during the test; if the flow changes, measure it again. Plan a separate test for checkmarks, explanatory copy or indicator placement.
3. 3A small gap, or one with no clear direction, is no reason to conclude right away that the bar has no effect. First confirm that visitors who start checkout split between the arms as planned (sample ratio mismatch); if the split is off, fix assignment before you interpret the difference. Once assignment is sound, enter each arm's visitors and completions in the A/B test significance calculator to see whether chance could explain the gap.
4. 4Check whether the bar adds to page load time and keep it simple; a heavily animated or slow-loading bar can cancel out any benefit. On mobile, also confirm that it does not take up too much screen space or cover the main action.