How it works
Enter the claim you're making and the evidence block that supports it. The analyzer scores seven factors: quantification, sample size, date, source citation, first-party vs second-hand, claim ↔ evidence fit and evidence volume. The aggregate is a 0-100 evidence strength score.
Why sample size is the most-missed factor
- "4x more citations" with no sample size sounds like marketing. "4x more citations across 47 clients" sounds like data.
- AI assistants treat sample size as a proxy for rigour - it lifts quantified claims from "anecdote" to "finding".
- First-party data beats second-hand data by default because the assistant can't get it anywhere else, so it preferentially cites the primary source.
Rewrite weak evidence
For every factor under 55, add the missing piece. Combine with the Evidence Builder to wrap the result in a schema-ready block, and the Claim Support Analyzer to audit a full page.