Evaluate experiments in Varify.io
Table of contents
In short
You can now link your Google Analytics 4 properties directly to Varify. This allows you to evaluate your experiments directly in Varify.io with one click and determine a winning variant.
Step-by-Step Tutorial
Set up GA4 evaluation in the tool
Link Google Analytics Property
- Navigate to "Go to Tracking Setup" in the dashboard and then click on "GA4 Report Integration" on the left-hand side.
- Here you can connect your Google Analytics account with Varify.io. To do this, click on "Sign in with Google". A Google login window will open. Sign in with the account that is also linked to the Google Analytics property and confirm access to your Google account through Varify.io. IMPORTANT: The Google account must have at least the "Marketer" permission level.
- When linking, make sure that both requested authorizations are activated if they are not selected by default.
- You can then use the two drop-down menus to select the corresponding Google Analytics account and the appropriate property.
Calling up the experiment evaluation
The reporting feature is only available for newly created experiments after the GA4 connection has been established.
As soon as you start a new experiment, the "Results" section will appear at the top right of the experiment. Click on it to go to the evaluation report.
Creating the report
You can add various Google Analytics goals on the report page.
- To do this, click on the "Add new goal" button.
- Select the metric you want to evaluate from the list
- Confirm the selection with the green tick.
You can assign several goals to a report, which are then displayed separately in tables.
The goal has been added to the report. In the report itself, you can see the list of all variants of the experiment and the corresponding metrics.
The reporting period can be set using the date input fields and confirmed with "Update Results".
Evaluation of the test results
A conversion rate and an uplift are calculated based on the number of visitors in the variant and the values of the selected metric.
To check whether your experiment is statistically significant, you can use the confidence level for the evaluation. Variants with a confidence level of 95 % or higher are highlighted with a green "Yes".
