A/B testing
A/B testing splits your traffic between your current banner and a variant, then measures which one earns more consent. You change layout, colors, and button radius in the variant. Visitor consent choices stay valid in either case.
Create a test
- Open A/B Testing in the dashboard at my.optsens.com.
- Click New test and give it a name.
- Set the Traffic split between variant A and variant B.
- Variant A is the Current banner control. It runs your live banner with no changes.
- In variant B, set the overrides you want to test. The table below lists what you can change.
- Set Scheduling if you want a fixed start and end, and a minimum number of impressions per variant.
- Click Create A/B test, then start it from the test list.
One test runs per domain at a time. Starting a second test is refused until the running one is stopped or completed.
What you can change in a variant
| Override | Options |
|---|---|
| Banner position | Popup, bottom or top bar, side panel (left or right), corner (bottom-left or bottom-right) |
| Banner theme | Keep current, or set custom background and text |
| Button radius | 0 to 50 px |
| Accept & Reject buttons | Background, text, border |
| Manage & Confirm buttons | Background, text, border |
Button colors apply to the pair. A color you set for Accept also applies to Reject, and a color for Manage also applies to Confirm. Paired buttons keep equal prominence, and rejecting stays as easy as accepting.
Contrast warnings appear under any color you set that may be hard to read. A dark-mode warning appears if dark mode is on for the domain, since dark colors override your custom variant colors for visitors with a dark device.
Run and read results
Start a draft test, stop it from the list, or let scheduling stop it. Expand a running or completed test to see per-variant results:
- A consent-rate donut split into Accept, Custom, Reject, and Dismiss.
- Counts for accepts, rejects, customizes, and dismisses.
- The rate at which the preference center was opened, and average time to a decision.
- A statistical-significance reading with a confidence level, and whether the result is significant yet.
- Sample-size progress against your minimum impressions per variant.
When you are confident, click Declare winner on a variant to complete the test with that winner. When a scheduled end stops the test, the variant with the higher accept rate is recorded as the winner automatically, if it collected enough impressions. Without enough impressions the test completes with no winner.
Plan gating
A/B testing is gated by plan. On a plan without it, the test list stays empty and creating a test returns a message that A/B testing is not available on your plan. See plans and billing for what each plan includes.