Amplitude is the leading product intelligence platform that helps companies use their customer data to build great product experiences for systematic business growth.

Split’s Amplitude integration allows product and engineering teams to deliver impactful features and personalized customer experiences at scale.

Build cohorts in Amplitude based on past or predicted user behavior to easily identify relevant sample populations to run beta tests and experiments on. Send cohorts to Split as segments to target specific customers when rolling out a new feature. You can also send Amplitude event data to Split through your Customer Data Platform (Segment or mParticle) to power measurement and learning.

Send Split impression data as a feed to Amplitude. Combine feature flag data with user data to understand how customers are interacting with each treatment and determine which features are producing the most favorable outcomes.

What can I do with the Split Amplitude Integration?

• Target customers by behavioral attributes to serve relevant flag treatments to the right audiences. Build behavioral cohorts in Amplitude based on user events (such as cart abandonment), time range (new users in the last 30 days), and/or specific property (location, plan type, etc).
• Save time and avoid errors by manually creating and managing target lists in several places. Once integrated, Amplitude cohorts can be imported into Split as a static segment or as a synced segment that updates at a specific time interval (hourly, daily).
• Personalize customer experiences to drive desired outcomes by targeting customers with in-app offers and services they are most likely to engage with based on past or predicted behavior.
• Create a behavioral cohort for heavy users of a specific feature and test a new version of that feature on the cohort.
• Create a predictive cohort for users most likely to achieve a desired outcome and test a new feature used to produce that outcome on the cohort.
• Run deeper analysis on impression data in Amplitude to compare user behavior across flag treatments and determine why certain metrics changed as a result.