we join based on the randomization unit you've assigned in the experiment report - so we pull a list of users (typically users, but could be anything) and then join by that id to other metrics like bookings or revenue. As long as those metrics can return a user id that can be joined back to the assignment id