How Do Fact-Checkers Work, and Would Your Coverage Survive One?

What does the code actually require?

Five commitments, each with operative detail rather than aspiration. The IFCN Code of Principles governs organisations that publish non-partisan reports on the accuracy of claims by public figures and institutions.

Non-partisanship and fairness comes first: signatories “fact-check using the same standard for every fact check”, avoid concentrating on one side, follow consistent processes, let evidence drive conclusions, and do not advocate policy positions on the issues they examine.

Standards and transparency of sources is the commitment most relevant to anyone producing coverage. Organisations must “identify the source of all significant evidence” in sufficient detail for readers to verify it, prioritise primary over secondary sources, verify claims against multiple sources, and disclose relevant interests of sources that could affect accuracy.

The remaining three cover funding and organisational transparency, published methodology for how claims are selected and researched, and an open corrections policy. Together they describe a discipline that is unusual mainly in being written down.

The methodology commitment deserves separate mention because it governs what gets checked at all. Signatories publish how they select claims, basing selection on “reach and importance”, present evidence supporting and undermining a claim even-handedly, and clarify what counts as fact-checkable in the first place. A claim about the future, a matter of opinion, or a statement too vague to test falls outside the exercise regardless of how contestable it sounds.

How is compliance established?

Externally, which distinguishes this from most standards examined in this series.

Assessment is carried out by independent assessors against 31 criteria, with the results reviewed by the IFCN advisory board. A signatory is not self-certified; someone outside the organisation examined whether it does what it claims.

That is a meaningfully higher bar than voluntary frameworks where adoption is self-applied, and it is worth noting the contrast. Journalism's trust indicators, the wire services' editorial standards and the awards programmes examined elsewhere in this series all rest on the organisation's own word.

It remains a compliance assessment rather than a guarantee of individual outputs. A signatory can satisfy 31 criteria and still publish a fact check somebody disagrees with, because the criteria govern process and transparency rather than the correctness of any particular conclusion.

The corrections commitment is the one most worth borrowing. Signatories must maintain a visible policy covering how major mistakes are handled, correct transparently, and where possible ensure users see the corrected version. Most organisations publishing claims about themselves have no equivalent, which means an inaccurate figure in a press release has no defined path to being fixed and typically stays in circulation through every republication.

Would promotional coverage survive this treatment?

Most of it would not be checked at all, and the parts that were would fail on sourcing.

Fact-checkers select claims by “reach and importance”, so a sponsored article about a mid-sized company is unlikely to attract attention. The exercise is still worth performing privately, because the criteria are published and anyone can apply them.

Apply the sourcing commitment to a typical press release and the problem is immediate. Claims to be the leading provider, to have transformed a category, to be trusted by customers, generally have no identified source, no primary evidence, and no way for a reader to verify them. Under the code's standard they are not weakly sourced; they are unsourced.

The interests requirement adds a second difficulty. Fact-checkers are asked to disclose relevant interests of sources that could influence accuracy, and the source of a promotional claim is the party that benefits from it — a disclosure that, once made, does most of the work of the assessment.

It is worth distinguishing two failure modes here, because they carry different weight. A claim that is unverifiable — being the leading provider, delivering exceptional service — fails because there is nothing to check. A claim that is verifiable but unsourced — a market share figure, a growth rate, a customer count — fails only because the source was omitted, and that one is fixable in a sentence.

Why does the funding commitment matter here?

Because it is the clearest statement anywhere in this series of what disclosed interest is supposed to look like.

Signatories publish pages detailing funding sources accounting for 5% or more of revenue, their organisational structure, how editorial control operates, and professional biographies of key figures. They must ensure funders have no influence over conclusions.

Set that beside the placement market as documented across these reports. Sponsored articles must be labelled but rarely name what was paid; awards publish fees but not their judging economics; vendor research is published by parties selling what it validates, usually without disclosure; and several sources cited in this series could not be obtained in primary form at all.

The IFCN standard is not a realistic demand on commercial publishing. It is a useful benchmark for reading it: when a claim arrives, knowing who paid for the work that produced it is the first question, and the fact that it is rarely answerable is itself informative.

The 5% threshold is worth borrowing as a rule of thumb. It sets a level below which an interest is treated as immaterial, which is a more workable standard than demanding every relationship be listed.

What can a company take from this?

A test to apply to your own material before anyone else does.

For every significant claim in a release, an article or a media kit, ask what the source is, whether a reader could check it, and whether it is primary. A claim that survives those three questions is defensible anywhere — to a journalist, an adjudicator, an encyclopaedia editor, or a customer.

Then ask what a corrections policy would require. The code asks signatories to correct transparently and to ensure users see corrected versions. Most companies have no equivalent practice for their own published claims, and a statement that quietly changes is a different object from one that is corrected.

The discipline is not onerous. It amounts to being able to say where each number came from, which is the practice this series has tried to follow, including where the answer was that a primary source could not be obtained.

Keep the working file too. The organisations held to this standard can show where a figure came from years later because they recorded it at the time, and a company that cannot reconstruct the basis of its own published claim is in a weaker position than one that simply got it wrong.

What this means for anyone buying placement

Write to a standard you could defend. A release built on identified, checkable, primary evidence reads better, clears wire editorial review more easily, and is more useful to any journalist who picks it up.

Be careful with claims that cannot be sourced. Superlatives are the default register of promotional copy and are precisely what fails every verification standard, including the ones applied by publications deciding whether to run something.

Disclose interests in your own material, including funding of any research you publish. A study a company commissioned about its own market is legitimate and useful when labelled, and worthless when presented as independent.

Media Morpheus sells placement. What a placement cannot do is make an unsourced claim true, and a page repeating unverifiable superlatives on a recognisable masthead has purchased distribution for a statement that does not survive being checked. The cheapest improvement available to most buyers is not a better outlet; it is a claim with a source attached.

Methodology and limitations

This report is based on the International Fact-Checking Network's published Code of Principles and its commitments, quoted for the five principles, the sourcing and funding requirements, and the assessment process. Four limits apply. First, the code governs fact-checking organisations; applying its sourcing standard to promotional content is our own analytical use, not an IFCN position. Second, the description of assessment against 31 criteria by independent assessors reflects the code's published account of its own process, which we did not independently verify. Third, no data is reported on how many organisations are signatories or on fact-checking outcomes, because no reliable figure was retrieved. Fourth, the code is revised periodically and this describes the position at the date of publication.

Conclusion

Fact-checking has a written standard, assessed by outsiders against 31 criteria, and its sourcing commitment is the most useful sentence in it for anyone producing corporate communications: identify the source of all significant evidence, in enough detail that a reader can check it, preferring primary sources. Very little promotional material survives that test, not because it is dishonest but because it was never built to be checked. The practical value is not that a fact-checker will examine your press release — almost certainly nobody will. It is that the standard is published, anyone can apply it, and a claim with a verifiable primary source behind it works everywhere while a superlative works nowhere that matters.

Frequently Asked Questions

Sources

1. trade body International Fact-Checking Network. Code of Principles: the commitments, for the five principles, sourcing and funding requirements, and assessment process.