Methodology
The scoring rulebook
Your GEO score is not an opinion. It is 67 best-practice rules, each worth a fixed number of points, applied to every page we audit. This page is the exact rulebook the engine uses.
How the score is calculated
- Each rule is measured on every audited page at a level from 0 (missing) to 5 (best practice). The site level is the average across pages.
- Points earned = level ÷ 5 × the rule's points, so partial work earns partial credit.
- Score = total points earned ÷ 418 possible points × 100.
- A level of 0 on a critical rule caps the whole score — if AI can't crawl or read a site, nothing else can rescue it.
- The grade is the band the final score lands in.
Critical rules: AI Crawler Access (caps at 59), Server-Rendered Content (caps at 59), Indexability Signals (caps at 44), Snippet Permissions (caps at 69).
Grade bands
- A+90-100AI assistants can read, trust and quote your site with almost no friction.
- A80-89Strong — AI assistants understand you well; only small gaps are left.
- B70-79Good foundation, but a few important gaps still hold you back.
- C60-69Half the basics are missing, so AI often skips you for competitors.
- D45-59Weak — AI assistants struggle to understand what you offer.
- F0-44AI assistants cannot reliably read or recommend you at all.
Crawlability & Access
11 rules • 78 of 418 points
AI Crawler Accesscritical
10 ptsrobots.txt explicitly allows GPTBot, OAI-SearchBot, PerplexityBot, ClaudeBot and Google-Extended.
If AI crawlers are blocked in robots.txt, your pages can never be cited in generated answers.
Pass at level 4/5 • Source: OpenAI GPTBot docs; Perplexity crawler docs; Google-Extended docs
llms.txt Manifest
8 ptsPublish /llms.txt with a one-line summary of the site and links to its key pages.
llms.txt gives generative engines a curated, low-noise summary of what your site covers.
Pass at level 4/5 • Source: llmstxt.org specification
XML Sitemap
8 ptsPublish an XML sitemap and reference it with a Sitemap: line in robots.txt.
Discovery is the first step of retrieval — unindexed pages are invisible to AI answers.
Pass at level 4/5 • Source: sitemaps.org; Google Search Central
Structured Data
13 rules • 80 of 418 points
Schema.org JSON-LD
10 ptsEvery page carries valid JSON-LD structured data describing its main entity.
Structured entities are far easier for a model to ground an answer in than prose alone.
Pass at level 4/5 • Source: schema.org; Google structured data guidelines
FAQ / Q&A Markup
8 ptsQuestion-and-answer content is marked up with FAQPage / Question / Answer schema.
Generative engines lift Q&A pairs almost verbatim when answering user questions.
Pass at level 4/5 • Source: schema.org FAQPage
Organization Entity
8 ptsOrganization or LocalBusiness schema includes name, url, logo and sameAs profiles.
It anchors your brand as a known entity in knowledge graphs models rely on.
Pass at level 4/5 • Source: schema.org Organization; Google Knowledge Panel guidance
Content & Answerability
18 rules • 118 of 418 points
Answer-First Structure
10 ptsThe first paragraph after the H1 answers the page's main question in 40-60 words.
Extraction models grab the earliest self-contained answer they can find.
Pass at level 4/5 • Source: Princeton GEO study (Aggarwal et al., 2023)
Semantic Heading Hierarchy
8 ptsOne H1 per page, followed by nested H2/H3 headings with no skipped levels.
Headings define the passage boundaries retrieval systems split content on.
Pass at level 4/5 • Source: WHATWG HTML; Google SEO starter guide
Content Depth
8 ptsCore pages contain at least 600 words of substantive, on-topic text.
Thin pages rarely contain the specific passage a model needs to cite.
Pass at level 4/5 • Source: Google helpful content guidance; GEO research
Authority & Trust
8 rules • 46 of 418 points
E-E-A-T Signals
10 ptsThe site shows experience, expertise, authority and trust: about page, credentials, reviews.
Engines suppress sources that show no evidence of real-world expertise.
Pass at level 4/5 • Source: Google Search Quality Rater Guidelines (E-E-A-T)
Contact & Business Details
6 ptsA physical address, phone number or contact page is easy to find.
Verifiable business identity increases the confidence weight placed on your content.
Pass at level 4/5 • Source: Google E-E-A-T guidance
Policy & Transparency Pages
4 ptsPrivacy policy and terms pages exist and are linked from the footer.
Missing policy pages are a classic low-trust signal for automated quality scoring.
Pass at level 4/5 • Source: Google E-E-A-T guidance
Entity & Brand
4 rules • 28 of 418 points
Brand & Entity Consistency
8 ptsThe same brand name appears in the title, Organization schema, Open Graph tags and footer.
Consistent naming lets engines resolve you to one confident entity instead of several fuzzy ones.
Pass at level 4/5 • Source: Google Knowledge Graph entity reconciliation
Topical Authority Signals
8 ptsRelated pages are clustered and interlinked around each core topic.
Engines pick the site that owns a topic, not the site that mentions it once.
Pass at level 4/5 • Source: Google topical authority guidance
Title & Meta Description
8 ptsEach page has a unique 50-60 character title and a 120-160 character meta description.
These strings are frequently reused verbatim as the summary of your page.
Pass at level 4/5 • Source: Google Search Central - titles and snippets
Technical Health
13 rules • 68 of 418 points
HTTPS & Secure Delivery
8 ptsThe site is served over HTTPS and plain HTTP redirects to it.
Insecure pages are down-weighted or skipped entirely by many crawlers.
Pass at level 4/5 • Source: Google Search Central - HTTPS as a ranking signal
Server Response Speed
6 ptsThe HTML document responds in under 800 ms.
Crawlers work to a time budget — slow responses mean fewer pages retrieved.
Pass at level 4/5 • Source: Google Core Web Vitals - Time to First Byte
HTML Weight & Signal Ratio
6 ptsThe HTML document is under 300 KB.
A low text-to-code ratio dilutes the signal a retriever extracts.
Pass at level 4/5 • Source: web.dev performance guidance
Live rulebook — updated as best practices change.