Results & Analytics

The Results page is where you see whether your experiment worked. It shows you how each variation performed against your metrics, how confident the statistical model is in those results, and whether there are any health issues you should be aware of. This section explains every number on the Results page and how to use it to make good decisions.

In this section

Reading Your ResultsOverview of the Results page — summary cards, variation table, and what each number means.

Statistical MethodologyHow A vs B uses Bayesian statistics to measure confidence — no p-values required.

Winning ProbabilityWhat winning probability means, when to call a winner, and the 95% threshold.

Health GuardrailsSample Ratio Mismatch, statistical confidence, and data quality indicators.

Filtering by SegmentView results for a specific device type, country, browser, or custom segment.

Segment LiftSee per-segment lift in one view, with multiple-testing correction and interaction tests.

Time SeriesConversion rate trends over time — detect novelty effects and seasonal patterns.

Secondary MetricsHow secondary metrics provide supporting context alongside your primary metric.

Exporting ResultsDownload your experiment data as CSV or save a visual PDF snapshot.

When to check your results

One of the advantages of A vs B's Bayesian approach is that you can check your results at any time without inflating false-positive rates. That said, it is still best to let experiments run until they reach the target sample size or a convincing winning probability. Checking and stopping too early — even with a Bayesian model — increases the risk of acting on noise.

A good rule of thumb: check in once or twice a week to make sure the experiment is healthy, but do not make a final decision until the experiment has been running for at least one full business cycle (usually one or two weeks) and has collected enough visitors.