Live Event
Interleaving: a faster way to compare ranking algorithms
Use interleaving when you have two good candidates and not enough traffic to tell them apart.
At Berlin Experimentation Meetup, GrowthBook's Hampus Poppius breaks down how interleaving works and when you would use it. The method mixes the output of two competing algorithms into one experience and reads which one users actually pick, reaching a verdict far faster than a standard A/B test on the same traffic.
GrowthBook is an open-source platform for feature flags, experimentation, and product analytics that runs directly in your data warehouse.
Why teams run GrowthBook
- Open source, self-hosted or cloud. Read the code, run it in your own infrastructure, and stop carrying an internal tool as a side project.
- Metrics in SQL, results in your warehouse. CUPED, sequential testing, Bayesian and frequentist engines, and multiple-testing corrections, with the query visible behind every analysis.
- Everyone can test. PMs, growth, and marketing launch their own tests in the AI Visual Editor or with agents, inside guardrails the platform team sets.
- Your agents ship too. The agent that writes the feature can wrap it in a flag, launch the experiment, and report back, under the same approvals and auto rollback as everyone else.
Register for the event here, or book time with the team while we're in Berlin.
More events
Ready to ship faster?
No credit card required. Start with feature flags, experimentation, and product analytics — free.
