Methodology
How Search Health is calculated
Version 1.0.0 — transparent, reproducible and open to challenge.
Last updated: 2026-11-05·By Kestralia editorial team
Design principle
The same data must always produce the same number, and every deduction must trace back to a specific finding. There are no hidden weights. The score is a quality measure — not a forecast of rankings, traffic or revenue.
Category weights
| technical | 28% |
| content | 22% |
| performance | 15% |
| structured Data | 10% |
| international | 7% |
| security | 10% |
| geo Readiness | 8% |
Severity weights
Each finding type has a base weight by severity: critical 0.35, high 0.15, medium 0.06, low 0.02, info 0.005.
Prevalence factor
A systemic problem is not punished once per page. The formula:
0.40 + 0.60 * (affectedPages / totalPages), capped at 1Findings are grouped by type, the worst severity in the group is used, and the prevalence factor is applied. Total deductions per category are capped at 1.0, so each category score stays between 0 and 100.
Worked example
Suppose 40 pages were crawled and "canonical points elsewhere" appears on 12 pages. Severity: high (0.15). Prevalence: 0.4 + 0.6 × (12/40) = 0.58. Technical deduction: 0.15 × 0.58 = 0.087 → roughly 8.7 points out of 100.
Overall score
The overall score is a weighted mean of the categories. AI readiness carries a smaller weight because it is an emerging capability and should not overshadow proven technical SEO.
Method limitations
- The score is based on the crawled sample; a larger site may produce a different result.
- Network latency and server load affect measured response time.
- Some findings (such as thin content) are heuristic and reported with lower confidence.
- No score can predict actual rankings.
Search Health is a reproducible quality score derived from observed crawl evidence. It is not a search engine ranking, a traffic forecast or a ranking promise.