On this page
To analyze A/B test results in CUX, you need a clear way to distinguish between the variants. Without this distinction, the platform cannot separate the behavioral data, and analysis will not be possible. CUX supports multiple methods of variant identification:
- Two separate URLs Each version has its own address, making segmentation straightforward. This option is ideal if you are testing different layouts of an entire page that serves one main purpose (e.g., a registration or contact form page), but also will help to differentiate the versions if you are testing a change in a single element. Example:
- Variant A - example.com/landing-A
- Variant B - example.com/landing-B
- Same URL with parameters Both variants share the same base URL but are differentiated by query parameters. This method applies to the same use cases as above, where you compare larger page changes. Example:
- Variant A - example.com/landing?version=A
- Variant B - example.com/landing?version=B
- Two versions of the same element with unique IDs This option is useful when testing a variant of an existing clickable element, like CTA buttons. Ensure the IDs are unique and descriptive; avoid overly short values (like just A or B) because this can complicate configuring Goals or Waterfalls. Example:
- Variant A - id=”cta-var-a”
- Variant B - id=”cta-var-b” Please note that this method only captures visits where users clicked a variant. To include visits where a variant was viewed but not clicked, use URL or Data Layer methods.
- Data layer values The most flexible option. Regardless of what you are testing, sending a variable to the data layer ensures CUX captures which version each user has seen or interacted with. Example:
- Variant A - test: var-A
- Variant B - test:var-B
Successful analysis in CUX depends on correct test setup. If variants are not tracked properly, the results may be incomplete or misleading. Depending on the method of variant identification you’ve selected, make sure that:
- CUX snippet is installed on both variant URLs.
- Goals configured with Click events using element IDs do not accidentally capture clicks on other elements.
- DataLayer capture integration is enabled BEFORE starting the experiment.
Important! To confirm dataLayer Capture availability and enablement for your project, contact support in the app.
Contact support in the app
- Sign in to the app and select the project related to your question, if applicable.
- Open Help and support in the top bar and select Contact Support.
- Describe what you tried, what happened and what you expected, then select Send request.
To check progress or add details, open Help and support > My support requests, select the request and reply in the same conversation. Continue an existing request when it concerns the same issue.
Do not include passwords, project tokens or unredacted customer data.