Highlighting One Option With Background Color A/B Test

Plan this test when visitors seem to hesitate on a selection screen, such as a long gap between the start and selection events, session recordings of the cursor moving back and forth between options, or support tickets asking which option fits. A helpful highlight means visitors choose in less time, selection completion rate does not fall and later changes do not climb. If the hesitation comes from unclear option descriptions, fix the wording first. Pick the option to highlight because it fits what visitors actually need, and do not add a "most popular" label without data to support it.

Selection completion time

Use the same start and selection events in both arms, and interpret completion time with the share who do not complete.

Also track

• Post-selection change rateTrack later changes among users who selected an option over the same follow-up period; not every change means regret.
• Visibility of other optionsUse the same event definition to check whether non-highlighted options remain visible.
• Selection completion rateCompare completed selections among all eligible assigned visitors.
• Post-selection satisfactionIf surveyed, use the same question, timing and sampling rule in both arms; omit this measure if there is no survey.

• ChangeChange only the background color of one predefined option.
• Selection mixRead shifts in selection alongside overall completion.
• Later changesTrack corrections and cancellations over the same period.
• AlternativesKeep other options' text, order, size and access unchanged.
• ExperienceReview first-time and experienced users separately.

1. 1Always read selection completion time alongside the share who never select; faster completers can hide more people giving up. If choices shift toward the highlighted option, weigh that against overall completion and the post-selection change rate, without treating every change as regret.
2. 2The highlight color must keep text contrast at an accessible level. If visibility of other options drops, or they become hard to read or click, do not treat a time gain as a reason to ship.
3. 3What you learn is specific to the highlighted option and the selection screen you tested; highlighting a different option is a separate test. Review first-time and experienced users separately, and track corrections and cancellations over the same follow-up window.
4. 4Time measures can be pulled by a few very long sessions, so look at the spread instead of a single average. The A/B test significance calculator shows whether the gap in selection completion rate is beyond chance; it cannot test the time gap itself, so read that with the same caution.