Scoring methodology
Zitatix reports two branded metrics for every scan: the Citation Readiness Score™ (CRS) and the AI Visibility Index™ (AVI). They measure different things, are computed in different ways, and are deliberately never blended into a single number. This page explains exactly how each is calculated, so the scores are verifiable rather than vague.
Current methodology version: 1.2, last updated . Whenever the checks or weights change, the version number changes and every report records the version it was scored with.
Citation Readiness Score™ (CRS)
The CRS grades what is under your control: the page itself. It is fully deterministic — the score is produced by roughly twenty-five programmatic checks with fixed point values, with no AI model involved in scoring. Run the same scan on the same page twice and you get the same number. That reproducibility is what makes the score citable.
The 100 points are split evenly across four pillars, 25 points each:
Findable (25 points)
- HTTPS with a healthy response status
- robots.txt access for AI crawlers such as GPTBot, ClaudeBot, PerplexityBot, and Google-Extended, evaluated with a group-aware robots.txt parser
- An XML sitemap that is reachable and referenced
- Indexability (no accidental noindex) and a canonical URL
Quotable (25 points)
- Exactly one clear H1 and a logical subheading structure
- Question-and-answer content that maps to how people ask AI assistants
- Lists and tables that can be lifted as structured answers
- Substantial content depth and concise, standalone paragraphs an engine can quote verbatim
Understandable (25 points)
- A descriptive title tag and meta description
- Semantic HTML landmarks (header, main, nav, footer) and image alt text
- A declared document language and Open Graph metadata
- A healthy content-to-code ratio — pages that need JavaScript to show their main content fail this pillar hardest
Trustworthy (25 points)
- JSON-LD structured data (Organization, FAQPage, Article, and similar types)
- Visible organization or author identity, plus about and contact pages
- Machine-readable publish and modification dates
- Outbound citations to authoritative sources, a favicon, and linked social profiles
Each check has a fixed weight. A failed check subtracts its full weight from the pillar; there is no partial credit unless the check itself is graded in steps (content depth, for example, is). The action plan in every report shows the exact points recoverable per pillar — computed from the failed checks, not estimated by an AI.
Score bands
| CRS | Interpretation |
|---|---|
| 80–100 | Strong — AI engines can find, parse, and quote the page reliably |
| 50–79 | Needs work — key signals are missing; citations will be inconsistent |
| 0–49 | At risk — the page is hard for answer engines to use at all |
AI Visibility Index™ (AVI)
The AVI measures your reputation off-site: how well AI answer engines already recognize and corroborate your brand through independent sources. It cannot be computed from your page alone, so Zitatix gathers evidence through live web search at scan time and grades it on a 0–100 scale.
Evidence classes, in rough order of weight:
- Knowledge-graph presence — Wikipedia, Wikidata, and equivalent structured registries
- Independent press coverage — articles about the entity, not by it
- Directory and review presence — industry directories, review platforms, ratings volume
- Consistent entity data — the same name, description, and contact details appearing across sources
- Social corroboration — active, linked official profiles
Because the AVI depends on the live state of the web, it can change between scans without any change to your site — that is by design. It answers a different question than the CRS: not "can AI quote this page?" but "does AI already believe this entity is real and notable?"
Why two scores instead of one
Blending on-page readiness with off-site reputation into one number would hide the most important information: what to do next. A site with CRS 90 and AVI 30 needs PR and directory presence, not more markup. A site with CRS 40 and AVI 80 is a well-known brand whose pages are invisible to crawlers — a fixable technical problem. Keeping the metrics separate keeps the advice concrete.
What a scan collects
A scan fetches the page (or, for website audits, the homepage plus key pages discovered from the sitemap), robots.txt, and sitemap.xml, then runs the deterministic checks server-side. The off-site analysis issues web searches about the entity and evaluates the results. Fetched content is treated strictly as data — it is never executed, and it cannot alter how the scan behaves.
Limitations
- The CRS evaluates rendered-free HTML — exactly what most AI crawlers see. Content that only appears after JavaScript execution is scored as absent, because for citation purposes it effectively is.
- The AVI reflects the evidence discoverable by web search at scan time; very new companies will score low until independent coverage exists.
- Neither score predicts search-engine rankings. AEO and SEO overlap but are not the same discipline — see What is AEO?
Frequently asked questions
Is the Citation Readiness Score deterministic?
Yes. The same page always produces the same score. All ~25 checks run server-side on the fetched HTML with fixed point values, so score changes always reflect changes on your site — never randomness in the tool.
Why is my AI Visibility Index low even though my score is high?
The two metrics measure different things. The Citation Readiness Score evaluates your own pages; the AI Visibility Index measures whether independent external sources corroborate your brand. A technically perfect site with no third-party mentions will score high on-page and low off-site.
How many pages does a website audit analyze?
A standard audit covers up to 10 pages including the start page, discovered via sitemap.xml or homepage links. Agency plans can extend an audit with additional pages from their monthly page pool.
How often should I re-scan?
After every meaningful content or markup change, and otherwise every few weeks. On-page fixes register as soon as engines re-crawl; off-site reputation moves more slowly, so monitoring over months is the realistic horizon.
Sources and further reading
The crawler behavior and markup requirements described on this page are based on the following primary sources: