What it contains
Conversion by segment — device, traffic source, new versus returning, landing page template. Segmented, because a blended rate conceals almost everything useful. Funnel step completion — where visitors leave between product page, cart, and purchase. Revenue per visitor — the metric that matters more than conversion rate alone. Average order value — so a conversion gain that cost you AOV is visible. Test results — variants run, the metric judged, sample size, and outcome. Page performance — Core Web Vitals on key templates, from field data.Reading it
Segment or learn nothing. A flat sitewide conversion rate can hide mobile falling while desktop rises. Always read device and source separately. Revenue per visitor over conversion rate. A variant that converts more at a lower order value can reduce revenue. If the two disagree, revenue per visitor is the one to trust. Compare against your own baseline, not an industry average. External benchmarks vary so much by category, price point, and traffic source that they cannot tell you whether your number is good. See benchmarks.Reading test results
Each test reports the metric it was judged on and the sample it reached. Both matter.- A result without adequate sample is not a result. Small samples produce large apparent effects that do not persist.
- No difference is a real finding. Most tests show no meaningful effect — that tells you to stop working on that element.
- At 95% confidence, one in twenty findings is false. Across many tests, some winners are noise.