How to calculate bounce rate with A/B testing?

Technical Questions CMS
1

In the Data provided in the A/B testing insights, I see no way to track the bounce rate - which is exactly what we want to measure in our tests. How do I add this custom metric?

Will the different test pages have a custom URL I can create a GA4 dashboard with?

What does “Clicks where is unique is true” represent?

2

Hello @elizabeth,

To add a custom metric like the bounce rate to your A/B testing in Builder.io, you can follow these steps:

  1. Create a Custom Event: In your application code or within the Builder UI by adding a custom script to track when a user bounces from the page:
    import { builder } from '@builder.io/sdk';
builder.track('bounceRate', {
  value: 1,
  label: 'Bounce'
});
  2. Customize Metrics in the Insights Dashboard:
    • Navigate to the Insights tab in Builder’s Visual Editor.
    • Click on the three dots in the upper right corner and select Customize Metrics.
    • Choose Manage Custom Events and add the custom event ‘bounceRate’ to appear in your dashboard.

Builder.io does not automatically create unique URLs for different test variations in an A/B test. Instead, all variations of the test share the same URL, and the user is allocated to a variant using cookies and managed with server-side logic to display the correct version of the content. This means that each test variant does not have its unique URL, but rather a common URL with different content served based on the A/B test configuration.

For reference, you may read about A/B testing and its implementation in Builder.io here.

The term “clicks where is unique is true” in the Builder.io A/B testing insights refers to the number of unique sessions in which a visitor has clicked on the content. This means that for each session, a click is counted only once, ensuring that multiple clicks by the same visitor in a single session are only counted as one unique click. This metric is useful for measuring distinct user engagement, rather than the total number of clicks, providing an accurate representation of how many individual users interacted with the test content at least once during their visit.

If you have any more questions or need further assistance, feel free to ask!

Thanks,