A/B tests

Run experiments to optimize your messaging. An A/B test compares users’ responses to multiple versions of the same campaign, while a multivariate test extends this to two or more variables. In Braze, the terms are used interchangeably because the setup process is the same. Use A/B testing with Intelligent Selection to automatically optimize your results.

Section articles

• 

  Concepts
• 

  Create tests
• 

  Optimizations
• 

  Analytics
• 

  FAQ

When to use A/B tests

• Trying a new messaging type: Experiment and learn what resonates with your users.
• Onboarding campaigns or recurring sends: Ensure high-traffic campaigns are as effective as possible.
• Multiple message ideas: Run a test and make a data-driven decision.
• Challenging assumptions: Test whether conventional marketing tactics actually work for your specific audience.

Tips for running effective tests

• Use large samples to ensure results reflect your average user and aren’t swayed by outliers.
• Randomize test groups so that differing response rates reflect message differences, not sample differences.
• Know what you’re testing. Isolating a single change identifies which element had the greatest impact; testing multiple differences lets you compare broader approaches.
• Set a test duration upfront and don’t end the test early, even if early results look promising.
• Add tests before launch. Adding a test to a running campaign produces inaccurate results. Clone the campaign, stop the original, and add the test to the clone.
• Include a control group to measure impact versus sending no message at all.