Feature Rollout
The progressive rollout of a new feature to a subset of users before a full launch.
Progressive rollout (gradual release): A method consisting of enabling a new feature for a limited subset of users, before a global deployment. Within a CRO and A/B testing approach, this method allows you to measure the real impact of a change (on conversion rate, revenue, retention...) while reducing technical and business risks. Progressive rollout is often managed via feature flags and orchestrated in stages (e.g., 1%, 5%, 20%, 50%, 100%).
However, it is essential to pay attention to the size of the exposed populations: samples that are too small can lead to a high standard error (SREM), making statistical analyses unreliable. It is therefore recommended to combine this type of rollout with solid metrics (RPV, success rate per session, etc.) and a rigorous calculation of statistical confidence before drawing conclusions.
Platforms such as ABtasty Feature Experimentation, Split.io, Optimizely, or Firebase Remote Config allow you to manage these rollouts with built-in analytics tracking.
Autres termes : A/B testing
View the full glossary- A/B TestingA/B(n) testing is a method that compares two or more versions of a web page, an application, an algorithm, advertising banners, etc., to determine which performs best.A/B testing
- AA TestA type of test where two groups receive the same experience to ensure that the testing system is reliable and that observed differences are not due to chance.A/B testing
- Client-side testing (A/B test)Client-side testing modifies the content visible to the user via scripts executed in the browser.A/B testing
- DatalayerA data structure used to transmit information between a website and a tag management tool.A/B testing
- Feature FlaggingA technique for enabling or disabling features remotely without deploying new code.A/B testing
- Test hypothesisA test hypothesis is a statement formulated to predict the outcome of a specific change on a website or application as part of an optimization process.A/B testing