In the context of our A/B tests, 'outliers' are users that make many more searches than the average user.
The searches associated with those users are removed automatically in order to make results more accurate - we don't want the abnormal behavior of those users to affect the performance metrics such as CTR and CVR.
It is possible to override the default removal outliers from A/B tests using the exclude parameter if configuring A/B tests via the API however this is not recommended.
A very large number of outliers can be indicative of bots crawling your web pages and performing thousands of searches as bots are counted as users. You can read our documentation on avoiding outliers due to bots here.
To better understand outliers, and the reason for their removal from A/B tests, we recommend reading through the following article: https://www.algolia.com/blog/engineering/removing-outliers-for-a-b-search-tests/