Skip to content

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.

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.

New Stuff!