SEO, GEO & Ecommerce Tools

Claim Source Mapper

Map every claim in a draft to the sources you supply, classify each as primary-source backed, weakly supported or unsupported, and export an editorial checklist.

  • Claim-to-source table
  • Support classification
  • Checklist CSV
Runs in your browser

Everything you paste, type or drop is processed in this browser tab. It is not uploaded, logged, stored or sent to analytics.

Claim sources workspace

Example:

1 Draft

2 Sources

Add (primary) or (secondary) to override the automatic type. Paste the figures you rely on into the notes: numbers are matched against them.

3 Claims and their evidence

Paste a draft and its sources, then map the claims.

What the Claim Source Mapper does

This tool finds the checkable claims in a draft - sentences with figures, comparisons, superlatives, causes or research language - and links each one to the sources you supply. Every claim is classed as primary-source backed, weakly supported or unsupported, with the reason, and the whole map downloads as an editorial checklist.

It works only on what you paste. It does not open the links or decide whether a claim is true; it shows where the evidence you have listed does and does not line up with what the draft says, which is the part of fact-checking that is easy to skip under deadline.

How to use it

  1. Paste the draft. Cite sources the way you normally do - [1], [2, 3], [1-3], (Source 4), a footnote ^5 or the URL itself.
  2. Paste the sources, one per line: number, title, URL and a note or short excerpt with the figures you rely on. Add (primary) or (secondary) to any line where the automatic type would be wrong.
  3. Select Map claims. Work down the table from unsupported claims, then weakly supported ones, reading the reason for each.
  4. Download the checklist CSV, which has an empty 'Editor checked' column, and give it to whoever signs off the piece.

Reading the results

Primary-source backed: the claim cites a primary source - the organisation that produced the data, the study itself, the official record - and every number in the claim appears in that source's notes. It is the strongest state the tool can report, and it still means only that your notes agree with your draft.

Weakly supported: there is a link, but something is missing. The source may be secondary (a blog or news article reporting someone else's figure), the match may be a suggestion based on shared words rather than a citation, or the claim states a number your notes do not contain - the most common sign of a figure that drifted during editing.

Unsupported: nothing links the claim to a source, or it cites a reference number that is not in your list. Flags add detail: absolute wording such as 'always' or 'proves', and appeals to unnamed experts or 'studies'.

Worked example: the walking article

The example draft makes eight checkable claims against five sources. Only one is primary-source backed: the NHS recommendation of 150 minutes a week is cited to nhs.uk and the figure appears in the note.

Four are weak. 'Walking 30 minutes a day' cites the WHO, but 30 is not in the note. 'Only 64% of adults' cites Sport England, whose note says 63.4% - a rounding the editor should confirm. The '2019 study' cites a DOI whose note says 2021. The 300-calorie claim cites a fitness blog, which is secondary, and 70 kg is not in its note.

Three are unsupported: 'Experts say stretching prevents injury' has no source and appeals to unnamed experts, 'always beats running' has none and uses absolute wording, and the fitness-tracker claim cites [6], which is not in the list.

What counts as a primary source here

A primary source is where a fact originates: the statistics office that ran the survey, the paper that reports the trial, the regulator's register, the company's own filing. Secondary sources report or interpret those. Editorial standards in journalism and research - and Google's guidance on helpful content, which asks whether content provides evidence of expertise and cites sources - favour going back to the primary source.

Without your tag, the tool decides by domain: government, intergovernmental, academic, DOI and standards-body addresses count as primary; everything else as secondary. That is an A2Z heuristic, and a rough one. A government press release summarising a study is secondary; a company's own sales figure is primary. Tag each line where you know better.

Limitations: what the result does not prove

  • It does not open the sources. Whether a source really says what the note says is for a person to check.
  • Claim detection is pattern-based. Opinions with a number in them are treated as claims, and factual sentences with no signal words can be missed.
  • Number matching is literal after removing thousands separators: 63.4% and 64% do not match, which is deliberate, but 'a third' and 33% do not match either.
  • A primary-source backed claim can still be wrong or out of context. The map shows gaps in the evidence trail, not truth.

Privacy: where your data goes

Everything you paste, type or drop is processed in this browser tab. It is not uploaded, logged, stored or sent to analytics. Session recording and tag-manager scripts are switched off on this page.

Standards and sources

Frequently asked questions

How do I fact-check an article before publishing?

List every claim that could be checked, link each to the source it came from, confirm the figures against that source, and prefer the original source over coverage of it. This tool does the listing and linking, and shows where figures in the draft are missing from the sources you noted.

What is the difference between a primary and a secondary source?

A primary source produced the information - a dataset, a study, an official record. A secondary source reports on it, like a news article or blog summarising the study. Secondary sources are useful for finding primaries but are weaker evidence on their own.

Why is my cited claim only weakly supported?

Because the source is secondary, the claim was matched by shared words instead of a citation, or the claim states a number that is not in the source's notes. Paste the exact figure from the source into its note, or correct the draft.

Can it detect hallucinated citations in AI-written text?

It flags citations to reference numbers that are not in your list and figures that do not appear in your notes, which catches many invented or drifted details. It cannot tell whether a listed source exists or says what the note claims; open each link.

Which citation formats does it understand?

Bracketed numbers such as [1], [1, 3] and [2-4], (Source 2), (Ref 2) or (See 2), caret footnotes such as ^3, and a source's full URL or domain written in the sentence.

Is my draft uploaded?

No. The draft and the source list are processed in your browser and are not sent anywhere. Nothing is saved unless you download the checklist.

Last reviewed by the A2Z.Tools team against the sources listed above.

Rate this tool

Was this tool useful? Your feedback helps us improve it.

No ratings yet — be the first to rate this tool.
Your rating (required)
0 / 2000

Please do not include passwords, payment details or other sensitive information.

Your feedback is sent privately to the A2Z.Tools team and will not be posted publicly.