Statsig has implemented measures to track bot traffic, including search engine crawlers, which can significantly inflate the number of stable_ids, making your user base appear larger. However, these bots typically do not trigger any events and therefore do not contribute to metrics. As a result, this behavior should not impact the computation of lift or the estimate of the confidence interval in most cases.
In some instances, bots may cause SRM issues. If you observe this, it could potentially be bot-related. To mitigate this, you can manually add filters to exclude bots using a targeting gate. For instance, a filter that excludes traffic where the browser contains "bot" can be quite effective.
While Statsig is actively working on a more comprehensive approach to handle bot traffic, it is currently recommended to use any deterministic detection on your side to filter out bot traffic. This can help align the exposure data more closely with other analytics tools and ensure that bot traffic does not affect the Stable ID assignments or the experiment results.
Please note that the impact of bot traffic should be evenly distributed between control and test, so any impact should cancel out. However, if you have further questions or need more specific advice, please reach out for assistance.