We don't skim your page. We test every claim on it.

A 2020¹ Stanford study found remote workers were 13% more productive². As one researcher put it, "the commute is the killer"³. Because they skip the commute, remote employees work 1.4 extra days a month⁴. We think hybrid is the future.

Follow one claim from your page to the public record

Five kinds of claim get tested. Opinion is left alone.

Is the number right, and does it mean what the sentence says? We check the figure, the sample, the time period and the comparison.

The original paper or dataset itself.

EXAMPLERight number, wrong measure: 13% more calls handled means 13% more calls, which is narrower than “more productive”.

Did it happen when the page says? We check publication years, event dates, deadlines and “since” or “as of” statements.

The publication record, official announcement or archive.

EXAMPLECiting the year a study was reported in the press instead of when it was published.

Did this person say these words, in this context? We check the exact wording and who said it.

The original interview, transcript, speech or publication.

EXAMPLEA paraphrase presented as a direct quote, or a quote attributed to the wrong person.

Is the term explained correctly? We check what a product, law, condition or technical term actually means.

The regulator, statute, official guidance or standard reference.

EXAMPLEMixing up two similar things: a Roth IRA is taxed going in and tax-free coming out.

Does the evidence show one thing causes another, or only that they happen together? We check the claimed link.

A study designed to test cause.

EXAMPLETurning “X and Y rose together” into “X caused Y”.

Statements like “We think hybrid is the future” are matters of judgement, so they are left as written. Your report lists every sentence we skipped, so you can see the full coverage. If an opinion rests on a factual claim: “hybrid is the future because 80% of firms now offer it”: the factual part is checked.

Where one wrong fact actually costs you something

Questions about fact-checking

The short answer first, then the detail.

Statistics (“13% more productive”), dates (“a 2020 study”), quotes and who said them, definitions (“a Roth IRA is…”) and cause-and-effect statements (“because they skip the commute…”). Opinions, predictions and clearly labelled commentary are left alone: we note each one we skipped, so you see the full coverage.

The study itself, the official dataset, the government table or the original interview. If only a weaker source exists, we say so in the verdict, so you can decide whether to keep the claim.

Holds: the source supports it as written. Wrong: the source says something different: we give the correct version. Unsupported: a reliable source is still to be found, so we suggest removing or softening it. Update: it was true once but a newer source replaces it.

A specialist in the field: an economist for an economics article, a clinician for a health page. Their identity and qualifications are verified, their name appears on your page’s record, and their past accepted work is public on their profile.

For each claim that needs a fix you receive the current text, the exact replacement, the reason and the source. You accept or decline each one and make the edits yourself. Each fix goes live when you accept it.

We confirm timing when we scope your order. Pages with many claims, or in specialist fields such as law or medicine, can take longer: you are told before work starts.