Defaults
By default, CURE will run a univariate regression when enabled - the same as the standard implementation of CUPED. To add additional covariatesProject Defaults
In project settings under Experimentation, specify default covariates for your company:
Experiment Settings
Per-experiment, specify additional covariates or remove covariates specific to your analysis:
Metric:Metric
One tool our framework enables is using metric:metric covariates; for example, using units’ pre-experiment clicks as a covariate for units’ in-experiment revenue. We decided against using this for several reasons:- Consistency. It should not be the case that adding a new metric to your analysis significantly alters the results of other metrics
- It’s not necessary. If you have key metrics that function as covariates, we recommend explicitly providing these - for all metrics - as a covariate in an entity property source. This achieves the same result, without “black box” outputs coming from unknown metric covariances
Preventing Adjustments
You can turn off CUPED in your pulse results, and can create a project-level setting to enforce this. This will still run CURE, however, which entails some amount of compute cost. To avoid running CURE, you can turn it off on a given metric by un-checking the CUPED option in the metric’s setup page.