Fact-checking protects publications from errors that damage credibility. The process involves systematically verifying claims, checking sources, and documenting verification. Strong fact-checking workflows catch mistakes before publication while respecting deadline pressures. Not every sentence needs independent verification—but certain claim types always do.
7 Claim Types That ALWAYS Require Verification
What to Verify
| # | Claim Type | What to Check | Example Error Caught |
|---|---|---|---|
| 1 | Statistics/numbers | Source document, calculation | "37%" was actually 34% |
| 2 | Direct quotes | Recording, transcript | "We will never raise taxes" vs. "no current plans" |
| 3 | Titles/credentials | Official website, directory | "Stanford Professor" was actually "Adjunct Lecturer" |
| 4 | Dates | Official records, calendars | Event was Tuesday, not Wednesday |
| 5 | Causal claims | Research, expert confirmation | Correlation stated as causation |
| 6 | Historical facts | Archives, reference sources | Law passed in 2019, not 2018 |
| 7 | Technical/scientific assertions | Published research, experts | Study actually showed opposite result |
Source Verification Hierarchy
Start with most authoritative source:
Source Reliability Ranking
| Tier | Source Type | Examples |
|---|---|---|
| 1 (Best) | Primary documents | Official meeting minutes, court filings, original study |
| 2 | Official databases | Census data, SEC filings, government records |
| 3 | Direct confirmation | Calling source to verify quote or claim |
| 4 | Institutional websites | Company "About" page, university directories |
| 5 | Secondary news sources | Previous news coverage (verify independently) |
Key insight: Secondary news sources can propagate errors. One fact-checker found a "unanimous vote" reported in multiple outlets—official minutes showed one abstention. Always prefer primary sources.
Fact-Check Comment Format
**Verified claims:** [Verified: City Budget Document 2026, page 47] [Confirmed via phone with mayor's spokesperson, 11/26/25] [Matches published study: DOI 10.1234/example] **Unverified claims:** [Cannot independently verify. Consider attribution: "according to Smith"] [Source document requested; not yet received] **Corrections needed:** [Correction: Article states 37%, but source document shows 34%. See attached screenshot.] [Quote check: Recording shows slightly different wording. See transcript.]
Handling Disputed Information
When Sources Disagree
| Situation | Approach | Example |
|---|---|---|
| Competing official estimates | Report both with attribution | "City estimates $50M; analysts project $75M" |
| Source makes unverifiable claim | Add attribution | "Officials estimate..." not "The project will cost..." |
| Claim contradicts public record | Note discrepancy | "Smith said X, though records show Y" |
| Technical claim beyond verification | Seek expert review | "According to researchers at [institution]" |
Quick Verification Checklist
- ☐ All statistics have documented sources
- ☐ Quotes checked against recordings/transcripts
- ☐ Titles and credentials verified with official sources
- ☐ Dates confirmed with calendars/records
- ☐ Causal claims supported by evidence
- ☐ Historical facts checked against archives
- ☐ Technical claims reviewed by experts
- ☐ Unverifiable claims clearly attributed
Frequently Asked Questions About Fact-Checking
How do I fact-check under deadline pressure?
Prioritize high-risk claims. Focus on statistics, direct quotes, and accusations first—these cause the most damage if wrong. Verify lower-risk claims (routine titles, common facts) as time allows. Note what couldn't be verified.
What if I can't verify something before deadline?
Add attribution to make claim's source clear. "According to the mayor..." instead of stating as fact. Note in comments what needs post-publication verification. Update online version when confirmation arrives.
Should I call sources to verify quotes?
Best practice is to record interviews and check against recordings. If no recording exists, calling to confirm is appropriate—but note that sources sometimes "clarify" to soften statements. Original recording trumps later clarification.
How do I handle "off the record" claims I can't verify?
If you can't independently verify, don't publish as fact. Attribute clearly ("sources say") or seek on-record confirmation. Anonymous claims require higher verification standards, not lower.
Can AI help with fact-checking?
Yes, AI tools like River's Fact-Check Assistant help identify claims requiring verification. It flags statistics, quotes, dates, and causal claims, searches for supporting sources, and tracks verification status. Always have humans make final verification decisions.
Systematic fact-checking protects your publication's most valuable asset: credibility. Use River's Fact-Check Assistant to build verification into your workflow without slowing publication.