Skip to content
BREAKING◆Markets Rally as Green Energy Infrastructure Bill Passes Committee
science

Fact-Check Lab

How we verify claims, check numbers and label what we cannot confirm.

The Fact-Check Lab is where we explain our verification habits in the open. It is less a department than a method: a set of questions that every claim must pass before it reaches you.

How a claim gets checked

Before publication, an editor asks four things of each significant claim. Who is saying this? How do they know? What would change the picture? And can we confirm it independently, or only report that it was said? Claims that cannot clear that bar are attributed, hedged or cut. We do not repeat a claim simply because it is interesting.

The three-tier source ladder

We rank sources by how directly they can be checked:

  1. Tier one: primary material. Original documents, datasets, filings, recordings and direct observation. These carry the most weight.
  2. Tier two: credible secondary sources. Reputable published reporting and peer-reviewed research. We cite them and try to trace them back to tier one.
  3. Tier three: self-reported. Claims from the organization or person concerned. They are useful, but never treated as confirmed on their own.

A story that rests mainly on tier three will say so.

Checking numbers and units

Numbers are where good stories most often go wrong. We check each figure against its source, confirm the unit and the time period, recalculate percentages and totals, and watch for common traps such as mixing absolute and relative changes, or capacity and output. Where a figure cannot be verified, it is hedged with words like 'roughly' or 'about', or it is left out.

How we label 'reported by' figures

Self-reported figures appear with their origin attached, for example 'the operator reports' or 'according to the programme'. That phrasing is deliberate. It tells you the number is real as a claim and unconfirmed as a fact, and it lets you decide how much weight to give it.

Illustrative scenarios

During this launch edition, our stories are illustrative scenarios built around generic actors rather than named real organizations. We flag that clearly, and we do not attach invented statistics, quotes or claims to real people or institutions. When we move to live reporting, scenario material will be labeled as such wherever it appears.

Ask us to check something

If you have seen a claim that deserves scrutiny, whether in our work or elsewhere, send it to [email protected] with a link or description and what concerns you. We cannot promise to check everything, and we will not provide legal, medical or investment judgments, but we read every request.

Update stamps

Each story carries an update stamp. If we revise a story after publication, the stamp changes and a note explains what was altered. Errors follow our Corrections Policy, and the standards behind all of this are set out in our Editorial Standards.

Last updated 2026-10-04