[GA4] Integrating with a third-party experiment tool

By integrating with a third-party experiment tool, you can run experiments through the third-party tool and interpret the results through Google Analytics. This article provides an overview of the integration between Google Analytics and third-party experiment tools.

For detailed integration steps, refer to the documentation from the third-party tool you integrate with.

Note: For partners who want to integrate their experiment tool with Google Analytics, see Create an experiment integration with Google Analytics for step-by-step instructions on how to set up the integration.


With the simplified integration between Analytics and third-party experiment tools, you can use the reports and Intelligence capabilities of Analytics while benefiting from the cutting-edge technology of third-party tools.

The integration standardizes around audiences rather than dimensions and metrics. Google uses an audience to group together users for each experiment variant. Then, each audience is deleted when the experiment ends. This means you can use the audiences you need for an experiment without exceeding audience limits when the experiment ends.

When an experiment ends, you can export your data to BigQuery, which enables you to pass the data back to a third-party tool. The third-party tool can then compute statistical inferences on the events to determine, for example, which variant is a winner. You can also export your data to BigQuery and use your own systems to run custom-made inference formulas to determine a winner.

How it works

When you create an experiment in a third-party tool, you choose the Google Analytics 4 property you want to associate with the experiment and consent to the integration.

When the experiment begins, Analytics creates an audience for each experiment variant. Then, the Experience-variant ID dimension surfaces the data associated with each user in an experience variant.

Before an experiment ends, you can analyze the results of the experiment using the Experience-variant ID dimension. When the experiment ends, Analytics deletes each audience so you don't exceed the audience limit. But, you can still access the experiment variant data in Explore and BigQuery.


Third-party tool


Google Analytics


Create the experiment and select Google Analytics 4 property as measurement option

Consent to the third-party tool creating Google Analytics objects


Start the experiment

An audience is created per variant


Your experiment runs on your website

Google Analytics associates users with an experiment using an Experience-Variant-ID


View experiment data in a report in the third-party tool

Audiences are available in reports and explorations


End the experiment

Audiences are deleted, but Experience-variant IDs can be used in Explore and BigQuery Export

Audience targeting

Audience targeting enables you to focus an experiment on a group of users who exhibit specific behaviors on your website. For example, you can use audience targeting to target high-value customers with special offers or incentives. Through audience targeting, only users in the audience can be a part of an experiment.

The audiences used for audience targeting are different from the audiences used for each experiment variant.

For each experiment variant, Google Analytics creates an audience that contains the users who are allocated to the variant. Each variant audience contains some of the users from the targeted audience. For example, if you create an A/B experiment, an audience is created for variant A and another one is created for variant B. These variant audiences are created only for measurement purposes.

When an experiment ends, the variant audiences are deleted, but data about the variant audiences are still available in your reports through the Experience-variant ID dimension for the data range of the experiment. Additionally, the audience used for targeting is unaffected by an experiment ending.

Was this helpful?

How can we improve it?
Clear search
Close search
Google apps
Main menu