January 27, 2025

Experiments

This feature allows teams to test ideas and validate them using data. By running controlled experiments, you can better understand what works for your users, leading to more confident decisions and improved outcomes.

Experiments leverage your metrics data to calculate results by comparing user groups segmented through a feature flag. Let’s have a look at the outcome:

Above is an engagement rate analysis for an experiment. We tested two versions of our website controlled by a feature flag — "on" and "off." The results clearly show that the "on" variant outperforms the "off" variant. When running experiments, it's important to validate whether the results are statistically significant. FeaturesFlow makes this easy by providing the "probability to be best" metric for each outcome, helping you make data-driven decisions confidently. This value is calculated using our advanced statistical engine powered by Bayesian statistics, which are increasingly replacing the traditional frequentist approach in experimentation. Our engine ensures that your experiment outcomes are analyzed efficiently and accurately using this modern methodology. You can add and remove success and guardrail metrics under their tabs within the experiment:

You can easily control the traffic split and treatments directly under the Feature Flag tab within an experiment:

Experimentation helps you make better decisions, backed by real data. With Experiments in FeaturesFlow you can test ideas, learn what works, and improve with confidence.

Give it a try and let us know what you think!