How Ranketic calculates the score
This page is generated directly from the analysis scoring model. Every number here is exactly what each analysis computes with (model version 2). Each analysis also shows its own calculation under “How the score is calculated”.
Overall score
The overall score (0–100) combines technical (35%), content (35%) and GEO – readiness for AI search systems (30%). All partial values are integers; the calculation uses the displayed values and rounds half up.
Overall = technical × 35% + content × 35% + GEO × 30%
If the AI assessment for GEO fails, GEO does not count as 0: technical and content are then weighted half each, and the report says so.
Caps
If Google cannot index the analysed page – “noindex” in the robots meta tag or X-Robots-Tag header, Googlebot blocked via robots.txt, or an HTTP error – the overall score is capped at 25. A page nobody can find should not score well.
If robots.txt locks out OAI-SearchBot (ChatGPT search) or PerplexityBot, the GEO score is capped at 40: the page cannot be cited there. Blocking pure training crawlers (e.g. GPTBot, ClaudeBot, Google-Extended) is a legitimate choice and costs no points.
Where the information comes from
Every value in the report carries a provenance label:
- MeasuredRead directly from the website (HTML, HTTP headers, robots.txt, sitemap). No AI judgement.
- Rule-basedFixed, published rules applied to measured facts. Same website → same result.
- AI-assistedA language model supplies partial measurements that are combined with fixed formulas.
- AI assessmentJudged by a language model based on the measured facts. An assessment, not a measurement – values may vary slightly on repeat runs.
- AI with web searchA language model researches the web and summarises. Please spot-check the details.
- AI-generatedWritten by a language model based on the findings of this analysis.
- Third-party estimateProjection by an external data provider (SpyFu), not a measurement on your website.
What Ranketic does not measure
Ranketic does not check whether your website is actually mentioned in ChatGPT, Perplexity or Google AI Overviews. The GEO score assesses readiness for it: crawler access, structure, clarity and citability.
Technical
Rule-based checks on measured facts. Each check row in the report shows exactly the severity that is deducted from the score.
Technical = 100 − 15 × critical findings − 5 × warnings − 1 × hints (minimum 0)
Checks include: HTTP status, indexability (noindex, nofollow, nosnippet), robots.txt and access for search and AI crawlers, sitemap, title, meta description, canonical, structured data, Open Graph, viewport, HTTPS, response time, image alt texts and language attribute. Site-wide checks (robots.txt, sitemap, crawler access) are judged once, on the main page.
Content
Points per criterion. Criteria that do not apply to the page type or could not be measured are left out; the remaining points are scaled to 100.
Content = earned points ÷ applicable points × 100
| Criterion | Points |
|---|---|
| Length of the main text (excluding navigation) | 15 |
| Content utility (AUI, AI-assisted): structure, redundancy, flow | 25 |
| Keyword density (no keyword stuffing above 2.5%) | 15 |
| Heading structure (one H1, gap-free hierarchy) | 25 |
| Linking (internal and external links) | 20 |
Text-length guidance per page type
A legal notice does not need 800 words, a how-to article does. The values are guidance, not a ranking factor in themselves.
| Page type | critical below | warning below | full points from |
|---|---|---|---|
| Homepage | 150 | 300 | 500 |
| Article | 300 | 600 | 800 |
| Product | 100 | 200 | 300 |
| Category | 50 | 150 | 250 |
| About | 100 | 200 | 300 |
| Contact | not scored | ||
| Legal page | not scored | ||
| Login/account | not scored | ||
| Cart/checkout | not scored | ||
| Search | not scored | ||
| Service/landing page | 150 | 300 | 500 |
Page types
The page type is derived from the URL, structured data (e.g. Product, Article), the title and login forms. Legal and functional pages – imprint, privacy policy, terms, login, account, cart, checkout, search – are not scored as subpages.
Subpages (full analysis)
Technical and content combine the main page (60%) with the average of the scored subpages (40%). Legal and functional pages, “noindex” pages and pages with HTTP errors are not scored. Subpage GEO values are shown but do not enter the overall score.
Technical or content = main page × 60% + avg. subpages × 40%
GEO – readiness for AI search systems
The GEO score combines an AI assessment of the page (76%, 70% for documentation and SaaS sites) with a rule-based llms.txt check (24% or 30%). The language model receives the measured facts (page type, indexing, crawler access, structured data) and rates on a fixed scale.
AI assessment scale: 0–20 barely usable for AI · 21–40 major gaps · 41–60 solid base with clear gaps · 61–80 well prepared · 81–100 exemplary.
Failures
If an AI-assisted evaluation fails, it is marked “not available” and does not enter any score as 0. The report names the affected parts.
Model changes
October 2026 (version 2): indexability (noindex, X-Robots-Tag, robots.txt per crawler) with caps; page types with their own text-length guidance; legal and functional pages no longer scored as subpages; partial scores show the values that enter the overall score; provenance labels; “not available” instead of 0 when AI fails; llms.txt only in the GEO score; missing robots.txt is a warning instead of critical.