Send experimentation decision events to Databricks
Explore the following options for seamlessly sending experimentation decision events to Databricks to enable robust data integration and enhanced analytical capabilities:
- Experimentation Events Export (E3) – Optimizely recommends using Snowflake's secure data sharing feature to access this export, or you can use Fivetran as an alternative.
- (Web Experimentation only) Custom Analytics integration – Requires an event ingestion pipeline.
- (Feature Experimentation only) Notification listener – Requires an event ingestion pipeline.
Experiment Event Export (E3)
Use the Experimentation Events ExportExperimentation Events Export to securely access your Optimizely Experimentation event data, letting you flexibly analyze your experiment results.
Considerations
- Configure experiment event exports to enable your team to load these exports into the customer warehouse periodically.
- Prepare to manage long-term maintenance responsibilities.
Get data into experimentation
- Use the oevents CLI oevents CLI or Amazon Web Services SDK.
- Consider using FivetranFivetran as an alternate solution.
Custom integrations for Web Experimentation
Implementing custom analytics integrationscustom analytics integrations introduces extra code execution for an experiment. When you develop your own integration, consider how the target product expects to receive data and how Optimizely Web Experimentation and Optimizely Performance Edge transmit it.
- Modify frontend code to route new events appropriately.
- Develop custom instrumentation as necessary.
Historical data cannot be used, and customers are responsible for managing retries and batching.
Notification listeners for Feature Experimentation
Notification listenersNotification listeners let you programmatically monitor and respond to various events within the SDK in Optimizely Feature Experimentation. When you configure listeners, complete the following:
- Implement callback functionality within the SDK.
- Modify code to route decision eventsdecision events appropriately.
- Develop custom instrumentation as needed.
Historical data cannot be used, and customers are responsible for managing retries and batching.
Learn more about configuring Analytics platforms.Learn more about configuring Analytics platforms..