Experiments

The Experiments tab compares the variants of an A/B experiment so you can tell what to ship.

Choosing an experiment

Use the experiment picker in the top-right. It defaults to a running experiment if one exists. The header summarizes the experiment's goal (its primary metric — completion rate or save rate), total sessions, and status, along with a significance badge:

• Significant — the difference is statistically significant (with the p-value shown).

• Trending — a difference is emerging but isn't significant yet.

• Not enough data — too few sessions or conversions to call.

Comparing variants

Two cards show Variant A (Control) and Variant B side by side: the variant's rate on the primary metric, its sample size, and the underlying counts (completed/abandoned for completion-rate experiments, or saved/cancelled for save-rate experiments). When the result is at least trending, the leading variant is marked as the winner.

The Verdict explains the result in plain language, including the relative lift of Variant B versus A and whether the experiment is significant yet or should keep running.

If no experiments exist, the tab invites you to create one on a journey.