Returning users are identified by user tokens that appear in the events on consecutive days. These events are the ones defined in your AI Personalization configuration.
To calculate the returning users rate, we take all user tokens from the last 24 hours and check which of these users also appear in events within the past 30 days. If a user token has events associated with it on at least one day in the last 30 days, it is counted as a returning user.
Tracking these returning users allows the AI Personalization model to generate accurate user profiles.