Evidence Density
What is Evidence Density?
Evidence Density measures how much of your content is supported by:
- First-party data (benchmarks, audits, internal studies)
- Reputable third-party sources (standards bodies, peer-reviewed work)
- Clear methodology notes (“how we measured this”)
- Precise definitions and boundaries
Key Insight: AIs are risk-averse. They prefer content that reduces the chance of being wrong. Evidence-dense content gets cited; opinion-heavy content gets paraphrased (without attribution).
How Evidence Density is Scored
Score per page or per section using this simple rubric:
| Score | What It Looks Like | Typical Content Traits |
|---|---|---|
| 0 | Opinions only | No sources, no numbers |
| 1 | Light support | Vague “studies show,” unclear citations |
| 2 | Some evidence | A few sources, limited methods |
| 3 | Strong evidence | Numbers + citations near claims |
| 4 | High evidence | Methods + constraints + update dates |
| 5 | Audit-grade | Reproducible approach + primary artifacts |
Why Evidence Density Matters
Evidence Density is one of the cleanest ways to convert “good writing” into “citable writing.”
| Content Trait | Human Effect | AI Effect |
|---|---|---|
| Strong narrative | Builds trust | Often paraphrased, not cited |
| Strong evidence | Builds confidence | More likely extracted and attributed |
Related: Information Gain measures “newness.” Evidence Density measures “provability.” The best pages have both.
How to Improve Evidence Density
- Put Sources Next to Claims: Do not dump citations at the end. Place them inline.
- Add Micro-Methods: One paragraph explaining dataset size, timeframe, how you measured
- Quantify Key Statements: Replace “often” with a percentage when you can defend it
- Use Definition Blocks: A 2-3 sentence definition under an H2 gets extracted cleanly by RAG systems
- Show Constraints: “This applies when X” increases trust more than absolute claims
Evidence Density FAQs
Does higher Evidence Density always win?
No. If writing becomes unreadable, you lose humans. Aim for high evidence in “extractable” sections: definitions, tables, key findings. Keep narrative sections flowing.
Is first-party data required?
No, but it helps significantly. Third-party standards and consensus sources also work well.
How does this relate to Information Gain?
Information Gain = newness. Evidence Density = provability. You need both for maximum AI visibility.
What is the minimum score to aim for?
Score 3+ on key sections (definitions, findings). Score 2+ on supporting sections. Score 0-1 sections should be minimized or removed.