What the score measures
GeoScore estimates whether a public page exposes the technical signals needed for crawler access, machine interpretation, evidence assessment, and stable rendering.
The result is a heuristic readiness estimate. It is not a measurement of real answer-engine visibility. GeoScore does not query ChatGPT, Perplexity, Gemini, Copilot, or Claude during this audit and does not claim access to their proprietary ranking systems.
What GeoScore collects during a scan
A user submits one public URL. GeoScore requests robots.txt before the page and checks redirect destinations before fetching them. The scan then reads the final HTML response, response headers, robots.txt, and the origin sitemap endpoint.
- Titles, descriptions, canonical tags, headings, links, images, and visible body text.
- Meta robots directives and the X-Robots-Tag response header.
- JSON-LD types, author signals, publication dates, modification dates, and Last-Modified.
- Script and interactive-control counts used for render-readiness estimates.
- Path-specific robots rules for selected search and answer-engine discovery agents.
The 12 factors and general-score weights
Weights total 100 and apply to the general score. They reflect GeoScore editorial engineering judgment about the relative importance of observable page readiness signals. They are not market-share weights and do not describe private engine algorithms.
| Factor | Group | Weight | Observed signal |
|---|---|---|---|
| Crawler access | Access and delivery | 12% | robots.txt rules for the analyzed path and relevant discovery agents. |
| Indexability | Access and delivery | 11% | Meta robots, X-Robots-Tag, title, description, and visible HTML content. |
| Canonical clarity | Access and delivery | 6% | Presence and quality of the canonical URL signal. |
| Sitemap signals | Access and delivery | 6% | Reachable sitemap.xml or a sitemap reference in robots.txt. |
| Render readiness | Access and delivery | 8% | Whether important content appears in HTML without excessive client-only dependence. |
| Structured data coverage | Meaning and structure | 10% | Count and breadth of valid JSON-LD entity types. |
| Schema fit | Meaning and structure | 8% | How well schema types match the visible page purpose and content. |
| Content and metadata alignment | Meaning and structure | 10% | Alignment between title, description, headings, and primary page text. |
| Multimodal access | Meaning and structure | 5% | Image alternatives and descriptive signals for non-text content. |
| Citation grounding | Trust and evidence | 10% | Visible external evidence and source-oriented linking around important claims. |
| Authorship signals | Trust and evidence | 8% | Visible or machine-readable author and publisher information. |
| Freshness signals | Trust and evidence | 6% | Publication, modification, and Last-Modified indicators. |
Overall calculation and critical gate
Each factor receives a score from 0 to 100. The general score is their weighted mean, rounded and clamped to the same range.
Access is a prerequisite. If crawler access or indexability is below 40, the weighted result is multiplied by 0.78. This prevents strong schema or copy from masking a page that discovery systems cannot reliably reach or index.
Model-oriented views
OpenAI, Perplexity, Gemini, Copilot, and Claude tabs reuse the same 12 observed factors with different emphasis. For example, citation and authorship signals receive more emphasis in source-heavy views, while structured entities and rendering receive more emphasis in Google-oriented views.
These values are planning lenses, not simulations of private systems. They should be used to compare which signal groups need work, not as claimed probabilities of being cited.
Bot access status
GeoScore evaluates the analyzed path against robots.txt for OAI-SearchBot, PerplexityBot, Googlebot, bingbot, and Claude-SearchBot. The result is intentionally limited to published robots policy.
Allowed does not mean indexed, cited, or recommended. Firewalls and bot protection can also block a request even when robots.txt allows it.
How to read score bands
Known limitations
- A single-page scan does not represent every template or URL on a domain.
- GeoScore does not execute a full browser rendering pipeline or measure every network request.
- Robots analysis describes published rules, not verified crawler identity or actual indexing.
- Content quality, factual correctness, and subject-matter expertise cannot be proven from markup alone.
- Real mentions and citations require a separate, repeated answer-observation dataset; GeoScore keeps this evidence outside the readiness score.
Methodology version history
Questions about the score
Does a score of 100 guarantee AI citations?
No. GeoScore measures technical readiness signals from the page and crawler policy. It does not prove that an answer engine indexed, mentioned, cited, ranked, or recommended the page.
Are the model-oriented scores based on private engine algorithms?
No. They are transparent heuristic views that emphasize different observable signal groups. GeoScore has no access to proprietary ranking or answer-selection algorithms.
Why can a score change after a methodology update?
Weights and thresholds can improve as the audit becomes more defensible. GeoScore includes a methodology version so comparisons can distinguish current and legacy calculations.