How CommonProof examines a claim.

1. Preserve the exact claim

We record the original wording, speaker, publication date, platform, and surrounding context. A shortened quotation is never treated as the complete evidence.

2. Identify what can be tested

We separate factual statements from opinions, predictions, rhetoric, and value judgments. Only testable claims receive factual verdicts.

3. Start with primary evidence

Legislation, regulations, court filings, official datasets, transcripts, public records, and direct source material take priority. Credible secondary reporting supplies context and independent confirmation.

4. Check time, definitions, and scale

Political statistics can change meaning when the timeframe, denominator, jurisdiction, or technical definition changes. Every review examines those boundaries.

5. State uncertainty

When evidence is incomplete, disputed, or unavailable, CommonProof says so. “Not enough evidence” is more honest than a forced conclusion.

6. Conduct a Democratic perspective review

CommonProof identifies the strongest good-faith Democratic interpretation of the claim, then tests it against the same source record and names its most important evidence concern. This is an AI-assisted perspective analysis, not an endorsement or a human party representative.

7. Conduct a Republican perspective review

CommonProof applies the identical process to the strongest good-faith Republican interpretation. Neither side receives a lower evidence threshold, favorable source selection, or a guaranteed share of favorable verdicts.

8. Publish the CommonProof public review

The final review states what both perspectives can support, what remains disputed, and what the evidence means in plain English. Reviews use transparent verdicts: Supported, Mostly supported, Mixed, Misleading, Unsupported, False, or Not enough evidence. Truth is never averaged between parties.