The Messaging and Mobile A/B testing features are not available for purchase and will be deprecated from the product on January 1st, 2022. Read more information on the Mixpanel blog.
In most reports, the quickest way to analyze Events and compare impacts based on which variant the user received is to use the Experiments Property. This Property is readily available in your Funnels, Retention, and Formulas report.
Users remain in the same variant within an experiment. When you create an experiment, the library passes the distinct_id to the endpoint to query which experiments are available for a user. Since the variant for each using is calculated based on the distinct_id, the variant will remain the same as long as the distinct_id is the same. The variant will change if you change the distribution of an experiment. For example, if the original experiment distribution was 50/50 and you change the distribution to 40/60, users that were included in the 40% variant could now be included in the 60% variant, although no users from variant 1 would be included in variant 2.
For example, here is the Property as it appears in the Formulas report:
The Experiments Property shown above is also available in Insights. However, rather than a human-readable name, the Property reflects the numeric Experiment ID and the Variant IDs.
To find the Experiment ID and Variant IDs for your A/B test for use in Insights:
- Click on the A/B Testing tab in your project.
- Click the Analytics Icon for the experiment you'd like to analyze.
- The last number in the URL is the Experiment ID. For example: mixpanel.com/report/your_project_id/ab_test/#analytics/experiment_id mixpanel.com/report/123/ab_test/#analytics/678
- To get the Variant IDs, from the same page, open your developer console (Mac: option + command + j on Google Chrome).
- Refresh the page and, once loaded, search for “segmentation” under Network, and the Variant IDs will show under Preview:
Once you know the Experiment ID and Variant IDs, you can segment by those values in your reports to understand which users were in the original or variant of your test and do further analysis on your experiments.