A/B Test: Confirming Choices Made During Checkout

Repeated clicks on the same option in session recordings, shoppers changing a selection before finishing, or support contacts about the wrong size or quantity hint that shoppers cannot tell whether a selection was saved. If the confirmation works, repeat clicks drop and shoppers pause less after a selection before completing the order. Before launch, verify that the message appears only once the selection is actually saved; if the problem is the selection control itself, fix that first.

Order completion rate

Compare completed orders to see whether written selection confirmation affects completion.

Also track

• Selection error rateMeasure errors only through predefined correction events or user reports.
• Repeat click rateTrack visitors who click the same option again after selecting it.
• Post-selection change rateCompare how often visitors change their selection before completing the flow.
• Order return rateTrack returns over the same follow-up period; report selection-related returns separately where identifiable.

• Written confirmationShow a short message after the selection is saved.
• Keep constantKeep the selection control, save behavior and existing color or icon states unchanged.
• Diagnostic measuresTrack selection errors and repeat clicks to help explain any change in completion.
• LayoutKeep the message from shifting content or causing accidental clicks.
• AccessibilityCheck mobile visibility and screen-reader announcement; test timing or icon changes separately.

1. 1Fewer repeat clicks or selection errors do not make a win unless the order completion rate improves too; those measures only explain a change in completion. Count selection errors only from predefined correction or user-report records.
2. 2A selection-related effect may surface later in returns, so track the order return rate in both versions over the same follow-up window, and report selection-related returns separately if you can identify them. The finding covers the written confirmation format in this checkout; check mobile apart from desktop, and keep timing or icon comparisons for later tests.
3. 3The effect of a confirmation may be small, so do not cut the test short. Set the number of visitors you need per version up front with the sample size calculator, and do not decide before you reach it. If you reach it and the gap is still unclear, do not treat that as evidence that the confirmation has no effect.