Skip to content
Independent statistics publisher. Not affiliated with any government, UN body or statistics office.

Methodology

Updated

1. Primary sources only

We link the organisation that produced a number: a national statistics office, a regulator, a company filing, a research institute, an industry body or the company that ran a survey. We do not cite a blog or an aggregator that repeated it.

2. One claim per sentence

Each statistic is one sentence that names its subject, number, unit, place, year and source. We keep the source's own scope: "US adults" stays "US adults".

3. Checked by an editor

Before a statistic goes live, an editor opens the source and confirms the figure, the scope, the year and the sample. Anything that does not match exactly is dropped.

4. Labelled by source type

  • Official: government and regulator statistics.
  • Filing: company reports filed with regulators, such as SEC 10-K and 10-Q filings.
  • Research: academic and independent research organisations.
  • Industry body: trade associations and industry groups.
  • Vendor report and vendor survey: data published by a company about its own platform or customers. We show the sample and who paid for it.
  • Our analysis: figures we calculate ourselves. The method and inputs are shown under each one.

5. Our own calculations

We add value by calculating figures that do not exist elsewhere: year-on-year changes, rankings, per-person rates and long series stitched from yearly releases. We only combine inputs with the same scope and the same year basis, and we never mix survey and census figures in one calculation.

6. Dated and archived

Every statistic shows its data year and the date we verified it. Every source gets an archived copy so the proof survives if the original page moves.

7. Reviewed on a schedule

Each page shows its next review date. Fast-moving topics are reviewed quarterly, slower ones yearly, and any page is updated when a key source publishes new data.

8. Corrected in public

If we get something wrong, we fix it and log the change on the page and in our corrections log.

Untraceable statistics

Some widely repeated figures cannot be traced to any study. When we cannot find the original source, we say so on the page instead of repeating the number.