Start with the original evidence
Our guides compile published research rather than claim new fieldwork. We check the original report, paper or dataset behind a number and identify the relevant figure or section where available. We prefer sources that explain their collection method and population. Search snippets and unsourced roundup articles are not sufficient evidence.
Preserve the measurement
Each entry records the claim, value, source, data period and limitations. Notes identify relevant geography, device, denominator and sample details when available. We distinguish percentages from scores, medians from averages, observations from experiments, and adoption from sales or traffic. Missing fieldwork dates are disclosed.
Keep dates separate
The data period describes the original measurement. A publication year describes the source edition. Last verified records when we reviewed the source; last updated records an editorial change. A newly checked page may legitimately contain older research.
Explain what the evidence cannot establish
General web benchmarks are labelled when used to inform custom websites or themes. Observed differences do not establish causation, and experimental visibility scores are not forecasts of traffic or revenue. We do not fill gaps with invented statistics or present a vendor claim as an independent result.
Keep the presentation consistent
Charts and CSV downloads use the same entries as the article table. Copyable findings include the source and context. Guides remain unpublished until their entries are checked; the publishing checks reject placeholder values and missing source notes.
Corrections and reuse
If a source changes or you find an error, send us the page URL, the affected finding and supporting evidence. We review corrections and update the affected entries and dates. Read our data reuse termsbefore redistributing our original compilation or notes.
