Try Demo

Talk to Adoption Engineer

Whitepaper

​
1. What are Split Tests?

Split testing (A/B testing) in Corbado Connect allows us to experiment with different feature configurations and user experience variations in a controlled manner. Unlike Gradual Rollout which controls feature availability, and Passkey Intelligence which makes intelligent contextual decisions, split testing enables us to test the effectiveness of different approaches by randomly assigning users to different experimental groups. Split testing is implemented in Corbado Connect to make consistent and persistent assignment decisions across multiple user sessions, operating independently of Passkey Intelligence’s decision-making.

​
2. How do they work?

The split assignment decision is made during certain user interactions (e.g., login) using a rule-based system similar to Passkey Intelligence. The assignment combines:
  • User Context: User ID, project ID, client information
  • Random Number: Generated per user session to ensure randomness
  • Split Rules: Configured rulesets that determine group assignment logic
Once a user is assigned to a split group, this assignment is permanently stored. This ensures consistency - a user will always be in the same experimental group across all their interactions.

​
3. Example: Auto-Append Feature

We successfully used split testing to evaluate the “auto-append” feature, where we tested different strategies for automatically prompting users to create passkeys after login. This allowed us to measure the impact on user conversion rates and experience before rolling out the optimal approach.