A/B testing

Start a test

1. Open a checkout page in the builder.
2. Switch on the A/B test panel.
3. Design your variants and set their traffic weights.
4. Publish.
5. Read the results under A/B testing.

A/B testing unlocks when your plan is active. A test can cover a whole page or a single section. You can limit it to first-time visitors, certain countries or a device. Visitors outside those filters see your live layout and are not counted.

How shoppers are counted

• Each shopper gets a variant by their checkout session, so a refresh does not change it.
• Variant A is the control, which is your current design.
• A conversion is counted once per paid order, against the variant the shopper saw.

Read the results

The table shows impressions, conversions, conversion rate, revenue and the change against A. The verdict reads Not enough data yet, Trending (not conclusive) or Significant (95%). The figure is directional and not a guarantee. Let each variant reach about 100 impressions before you trust it.

Finish a test

• Promote copies a winner into your live checkout and stops the test.
• Stop & archive ends the test and keeps it under Previous tests with its final stats.
• Reset stats wipes all counters.

A/B tests on the post-purchase card are set up in Promotions and show here too.