- Crawl & index 30%
- On-page 25%
- Structure & links 20%
- Performance & accessibility 15%
- Structured data 10%
Audit methodology
Understand how website findings are verified and prioritized
See what SearchProof AI reviews, how findings are tied to public-page evidence, and how results should be interpreted.
Core
Scoring model
How the score is calculated
Every finding passes through evidence validation, deduction limits, category weights, and critical-rule caps.
- 01
Verified evidence
HTTP, DOM, links, structured data, source URLs, and quotes are linked by evidence ID.
- 02
Deduction candidates
Rules and semantic AI review can propose findings, but unsupported proposals are excluded.
- 03
Server-enforced cap
Deductions stay within the rule-derived category cap based on severity and affected-page coverage.
- 04
Area score
Weighted category scores are combined before any critical P0 cap is applied.
Scoring formula
Core formulas
rule deduction = min(policy cap, severity base × (0.35 + 0.65 × affected-page ratio))applied deduction = min(100, AI recommendation × 1.30, rule-derived category cap × 1.30)category score = clamp(round(100 − applied deduction), 0, 100)area score = round(Σ(category score × category weight)) → critical-rule capSeverity bases are P0 55, P1 25, P2 12, and P3 5. Per-check policy caps and affected-page coverage can change the applied deduction.
Weight
Category weights by area
- Direct answers 30%
- Question coverage 25%
- Entity & terms 20%
- Evidence & trust 15%
- Semantic structure 10%
- Entity consistency 25%
- Author & source 25%
- Factual verification 20%
- Citability 20%
- Freshness & access 10%
Measurement states
How measurement states affect scoring
| State | Meaning | Score treatment |
|---|---|---|
| Measured | Evidence was collected and the rule could be evaluated. | Only verified failures deduct points |
| Unmeasured | Access, integration, or evidence was unavailable. | Not converted to zero or failure |
| External observation failed | A connected search or AI channel returned no usable response. | Kept separate from readiness scores |
| Review required | A new check or policy decision still needs approval. | Excluded until approved |
AI mention rate, recommendation rate, and official-domain citation rate are observation metrics calculated from successful external responses. They are not added to SEO, AEO, or GEO readiness scores.
01
Who it helps
When this assessment is useful
evidence behind each reported issue
For teams that need to evaluate evidence behind each reported issue with evidence instead of assumptions.
priorities for technical and content work
For teams that need to evaluate priorities for technical and content work with evidence instead of assumptions.
measured, unavailable, and external-observation states
For teams that need to evaluate measured, unavailable, and external-observation states with evidence instead of assumptions.
02
Analysis scope
What the Audit methodology assessment checks
The checks use observable public-page evidence and keep unavailable data separate.
SEO, AEO, and GEO readiness signals
Checks SEO, AEO, and GEO readiness signals on accessible public pages and links each finding to its source.
source URLs and page-level evidence
Checks source URLs and page-level evidence on accessible public pages and links each finding to its source.
issue impact and remediation priorities
Checks issue impact and remediation priorities on accessible public pages and links each finding to its source.
AI observations kept separate from readiness
Checks AI observations kept separate from readiness on accessible public pages and links each finding to its source.
03
How it works
How the assessment works
- 01
Define the target
Confirm the website, market, language, and question being assessed.
- 02
Collect public evidence
Inspect accessible pages and record the exact URLs used as evidence.
- 03
Verify and prioritize
Separate verified findings from unavailable observations and assign practical next steps.
04
Result example
Audit methodology result example
Sample values only; no customer data or performance claims are used.
| Priority | Example finding | Evidence | Recommended action |
|---|---|---|---|
| P1 | A reported issue has no verifiable source URL | Public page evidence | Keep it out of the confirmed findings until evidence is available |
| P2 | A required integration is unavailable | Public page evidence | Mark the result unmeasured instead of treating it as a failure |
Interpretation notes
Limitations to review with the result
- The result does not guarantee rankings, AI citations, traffic, or revenue.
- Private, login-only, blocked, and unavailable pages are outside the measured scope.
- Compare results under the same website scope, market, language, questions, and observation conditions.
05
FAQ
Frequently asked questions
01What does the Audit methodology assessment measure?
See what SearchProof AI reviews, how findings are tied to public-page evidence, and how results should be interpreted.
02Does this guarantee better visibility?
No. It establishes a measurable baseline and identifies changes, but search and AI systems remain outside SearchProof AI’s control.
03How do I run this assessment for my site?
Enter your domain on the audit screen to record the current baseline and identify the highest-priority fixes.
06
Related guides
Next checks
Start with a baseline
Audit my site with this method Start by recording a baseline for your public pages.
Enter your domain to save a baseline score and identify the highest-priority fixes.