Configure analytics platforms

6 min

These are third-party integrations and not Optimizely subprocessors. See Optimizely's Third-Party Add-Ons & Platform Integration Terms.

When you create flags, you want to know when users are accessing a feature flag and what experience they receive in production.

Although Optimizely Feature Experimentation does not offer analytics on feature flags by default, you can still track analytics on feature flag usage by adding a notification listener to send events to the analytics provider of your choice.

Notification listenersNotification listeners trigger a callback function that you define when specific actions are activated in the SDK.

The most common use case is sending a stream of all feature flag decisions to an analytics provider or an internal data warehouse to join it with other data you have about your users.

To track flag usage, complete the following steps:

Configure an analytics provider of your choice.

Configure a notification listener.

Follow your analytics provider's documentation and send events from within the listener callback.

The experiment ID and variation IDs generated by Optimizely are 16-digit numbers. This can potentially cause issues with custom integrations that might misinterpret these IDs as credit card numbers, resulting in data being flagged or removed.

To prevent this, consider incorporating a letter prefix or using a combination of name and ID (for example expID(expName) or varID(varName)) in your custom integration configuration. This cannot be done in the Optimizely UI, instead this would be done in your custom integration. Adjustments may be necessary for custom integrations to ensure data is captured correctly.

The following sections are suggested implementations for some common analytics platforms. Use these suggestions as presented or adapt them to meet your specific needs.

Analytics integrations

Analytics integrations let you filter your third-party data by Optimizely experiments and variations. This enables you to measure experiment performance based on events that the third party tracks. Analytics integrations also let you create custom third-party reports that include only users who saw a particular experiment.

Analytics integrations are conceptually simple, but there are some important considerations in configuring your integration. When a user activates an experiment, you need to notify your third-party analytics service about which experiment and variation the user saw.

Generally, each third-party analytics platform has its own event format for this notification. For example, Segment's Experiment ViewedSegment's Experiment Viewed event has expected properties that contain your experiment and variation. With Google AnalyticsGoogle Analytics, you can use a non-interaction event with a custom dimension that contains your experiment and variation. Check your third-party analytics platform's documentation to confirm how they expect you to send this information.

You need to notify your third-party analytics platform when a user sees an experiment, which is when you use the Decide method. You can make use of the notification listeners to bind this functionality to the client’s Decide method.

If you already implemented an integration but are questioning the data or want to know what third-party data Optimizely will support, read about data discrepancies in third-party datadata discrepancies in third-party data.

Audiences

Audience integrations help you target your experiments to segments of users that are identified in a third-party data management platform (DMP).

You can target your experiment to users by values for the attributes you have configured in your Optimizely project. Add the audience attributes you want to target to your project before passing them to Optimizely in your code. For information, see Define attributesDefine attributes.

After you create the attributes within Optimizely, you can pass your DMP's attributes when you build the OptimizelyUserContext object.