Holdouts

Holdouts measure the aggregate impact of multiple features. A holdout keeps a group of users back from a set of features for measurement. While each A/B test or experiment compares control and test groups for that feature, a holdout compares the holdout group (Control) against a balanced group of users who were not held out and continued through the normal rollout or experiment behavior for the included features.

​

How to use Holdouts

1. To create a new holdout, navigate to the Holdouts section on the Statsig console (it is a specialized kind of experiment).
2. Click the Create New button and enter the name, description and unit type of the holdout that you want to create.
3. You can choose to either create a global or a selected holdout. A global holdout is automatically added to any new feature with the same unit type, and is meant to capture the aggregate impact of all features developed after the holdout began (individual features may be opted out as needed). A selected holdout captures the aggregate impact of a specific selection of features that you want to hold off.
4. By default Holdouts apply to a % of all users (Population = Everyone). You can optionally target the Holdout at a subset of users by applying a Targeting Gate (Population = Targeting Gate). e.g. If you wanted an iOS users only Holdout, you could apply a Targeting Gate that only passes iOS users.
5. You must set the percentage of users to be held-out between 1% to 10%. Statsig recommends a small holdout percentage to limit the number of customers who don’t see new features.

​

How to read Holdouts

Holdouts on Statsig use the same “equal variant” methodology as Feature Gate rollouts, whereby metric lifts are computed by equal sized groups to calculate holdout lift. You can read more about the advantages of this methodology in “A/B Testing Intuition Busters: Common Misunderstandings in Online Controlled Experiments” by Ron Kohavi, Alex Deng, & Lukas Vermeer. Accordingly, the Cumulative Exposures panel for a given Holdout shows total exposures of the Holdout, broken down into three groups:

1. In holdout (Control): Units that were included in the Holdout and used for analysis.
2. Not in holdout (Test); used for analysis: Units that were not included in the Holdout and were selected for comparison against the holdout group.
3. Not in holdout (Test); not used for analysis: Units that were not included in the Holdout and were not used in the lift calculation.

For units not included in the Holdout, Statsig generates the two “Not in holdout” groups using random sampling. The group used for analysis is sized to balance the comparison against the holdout group. Holdout metric lifts represent the cumulative impact of launched and active experiments on the Holdout group vs. the same percentage of the rest of the population continuing through the normal behavior for the included rollouts and experiments. The “Not in holdout (Test); used for analysis” group is not necessarily made up of users who saw every treatment. Those users follow the normal non-holdout behavior for each included gate or experiment. In the example below, the 1% Holdout compares the metric values of users in Holdout vs. 1% of users not in Holdout. It does not compare the 1% Holdout against the full remaining 99% of users. The launched features are having an overall negative effect on the “Add to Cart” metric.

​

Best Practices

1. Size - Statsig recommends a low single-digit holdout percentage, say 1% – 2%, to limit the number of customers who don’t see new features.
2. Duration - Statsig recommends operating holdouts for a period of three to six months, and then releasing the holdout. Prolonging the holdout period may increase the complexity of your software as you’d have to maintain a functioning product with no new features for a longer period.
3. Back testing - Occasionally you may want to turn off a set of features that you have already released to measure the effectiveness of those features. Statsig doesn’t recommend this as it turns off features that users are already using and relying on. However, when a “back measurement” is critical, you can use Holdouts to turn off a set of features and automatically compute the impact of this set of features.

​

Unit ID Types

By default, holdouts are based on User ID. To use a different ID type, select it from the drop down menu during the holdout creation.

​

Holdout effects on Gates & Experiments SDK methods

​

Feature Flags/Gates

• For users in holdout, gates will always return False.

​

Experiments

• For users in holdout, if the experiment is not in a Layer, calls to get experiment parameters will always return the “default value” passed in code.
• For users in holdout, if the experiment is in a Layer, calls to get experiment parameters will return the values defined in the Layer defaults in the Statsig console.

​

Ending a Holdout

To end a holdout and allow users in the holdout group to see all held-out features, you can disable the holdout. Disabling it stops tracking the effects of those features, but the results will still be retained for future reference. Alternatively, you can delete the holdout if it was created by mistake or if you no longer need to keep the results.