Skip to main content
Consent records and statistics

Cookie banner A/B test ideas and examples

Four hypothetical tests you can adapt, with what to change, what to keep and how to decide.

These examples are hypothetical. A result applies to the browsers that responded to your banner in that configuration and period, so another website’s result is not evidence for yours. Give each test a meaningful name and make one main change, so that the result is easier to explain. You can preview both versions, and a test always splits one eligible opt-in configuration 50/50.

1. Test a shorter introduction

Question: does a shorter introduction change full acceptance? A keeps the current wording. B shortens the introduction and keeps every required explanation, purpose, vendor disclosure and policy link; everything else stays the same. Write the text from what your website actually does, and leave standardized TCF wording unchanged. A suitable name is Clearer introductory text.

A result shows how the wording performed among responding browsers. It does not show that the text is legally sufficient.

2. Test clearer typography or spacing

Question: does text that is easier to read change the outcome? B changes the font size, line spacing or spacing within the same banner type, while content, choices and button actions stay identical. Before you start, check in both versions that text wraps readably, that the banner works when zoomed, that touch targets are large enough and that keyboard focus is visible.

A result shows how the new typography performed; it does not prove that one version is more accessible.

3. Test two accessible color palettes

Question: does another palette change full acceptance? B uses different colors. Both palettes must meet the required contrast, and the accept, reject and settings choices must be equally easy to see and understand in both. The editor does not stop you from choosing colors that fail contrast, so check them yourself.

A higher rate does not make a design compliant, and no color works best on every website.

4. Test a position or width

Question: does where the banner appears, or how wide it is, change the outcome? B changes a permitted position or width within the same banner type. Modal behavior, consent actions and required content stay the same. Check small screens and zoom before you start.

A different template family, consent framework or geographic audience is not a valid version B.

5. Decide what to keep

Compare the Full acceptance rate of each version, the difference in percentage points and the status. A higher rate for B without a clear difference is inconclusive. Then decide whether to keep A or apply B. Accessibility, accurate information and a meaningful choice remain requirements whatever the result.

Do not hide or weaken the reject choice, preselect optional categories, change purposes, make visitors choose again or shorten how long consent is remembered to collect more observations. No test guarantees an uplift, a revenue gain, a sample size or a required duration.

To set up, read and end a test, see Create your first cookie banner A/B test, Understand your banner A/B test results and End or troubleshoot a banner A/B test. For design and accessibility, see Customize your cookie banner and Check your banner on mobile and with a keyboard.