At the start of a test procedure, we recommend you perform an A/A test on your site’s pages. In an A/A test, no change is made on the variant.
This type of test allows you to:
• Calibrate your test, and see if data is sent correctly.
• Make sure that your site’s AB Tasty tag works correctly.
• Make sure that outside elements don't degrade your report results.
• Gather data to give you test ideas using native objectives or plugins available in our library: for instance scroll time or time spent on the pages.
• Make sure that the link with your analytics tool works.
The procedure to follow is very simple:
You need to load one of your site’s pages into the editor and make no changes in the variant.
Implement plugins to gather new data
Scroll percent plugin
This plugin lets you see the number of visitors who scroll to the end of the page. To add it, simple select “engage” in the editor, and choose this plugin. You can configure the plugin so that it registers a conversion each time a visitor scrolls to: 25%, 50%, 75% or 100%.
Dwell time widget
In the plugin library (“Engage” in the editor), you can also install a widget that measures the amount of time your visitors spend on the pages. If, for example, you set an objective of 30 seconds, a conversion will be shown in your report each time a visitor spends more than 30 seconds on a page.
You can choose to target the pages you want for this A/A test. We recommend you perform an A/A test for each landing page.
For example, for an e-commerce site:
- Home page
- Product page
- Order summary
In an A/A test, you should directly allocate 100% of your traffic to the variant so as to send a maximum amount of data to your report.
You can allocate 50% of traffic to your variant, but we recommend that you stick with 100% to collect data more quickly.
Linking your A/A test with your analytics tool
We also recommend that you link your test with your analytics tool in the Advanced options step. This will allow you to verify that your AB Tasty data is collected correctly in your analytics tool.
Launch your test
Once you have completed these steps, you can launch your A/A test or schedule the launch data by clicking on the clock.
After conducting your test, you will have access to precious data that will allow you to come up with test ideas.
Example of a report on a scrolling objective
Note: an A/A test is conclusive if the conversion rates for your objectives are equal between the original version and the variant if you decided to split your traffic 50/50.