Adding covariates to the post analysis can reduce variance. One instance of this is CUPED by there are lots of covariates which are easier to add (eg request ty

Adding covariates to the post analysis can reduce variance. One instance of this is CUPED by there are lots of covariates which are easier to add (eg request type, response latency, day of week, user info, etc).