What is a GEO score?
GEO stands for Generative Engine Optimization - the practice of structuring a page so AI answer engines can find it, parse it, and quote it accurately. The GEO score is a single number from 0 to 100 that summarises how well a page meets the technical, structural and semantic requirements those engines reward.
Think of it as PageSpeed for AI. One number, several diagnostic categories underneath, and an opinionated verdict on whether the page is ready to be cited by ChatGPT, Perplexity, Claude, Gemini and Grok.
What we score
- Crawlability - response status, redirects, content size, robots compliance.
- Structure - heading hierarchy, semantic HTML, schema.org markup, FAQ blocks.
- Content quality - word count, reading level, answer density, list and table usage.
- Citations - external references, expert quotes, source diversity.
- Freshness - last-modified signals, dates in copy, year-currency.
Reading your score
- 0-39 (Poor) - the page has fundamental gaps. AI engines will skip it or hallucinate around it.
- 40-69 (Needs work) - the page has the bones but is missing the structural cues AI engines reward. Most public pages land here on a first audit.
- 70-100 (AI-ready) - the page meets the bar. Iterate to push individual category scores rather than the headline number.
How GEO differs from traditional SEO
Search engines optimise for "is this the right link to send the user to?" AI engines optimise for "is this the right paragraph to quote?". Same content, different scoring. Schema markup matters more. Answer-first prose matters more. Tables and lists matter more. Anchor-text-stuffing and meta-keyword tricks matter less.
A page can rank #1 in Google and still score poorly here. The opposite is also true - a page can be invisible in search and yet quoted relentlessly by Perplexity because it has the structure AI parses cleanly.
GEO is not new SEO with a new acronym. It is a different scoring function. Pages that ranked #1 in Google for ten years can score 30 here because they were optimised for keyword density, not answer extraction. The teams who notice the gap fastest will own the AI category for their vertical.
Score band benchmarks
| Band | Verdict | What it means | Action |
|---|---|---|---|
| 0-39 | Poor | Fundamental gaps. AI engines skip the page or hallucinate. | Triage. Fix crawlability and structure first. |
| 40-54 | Weak | Page is reachable but not optimised. Below average. | Add schema, fix headings, expand answer density. |
| 55-69 | Average | Most public pages live here. Citable but not preferred. | Push two categories above 70 to break out. |
| 70-84 | Good | AI-ready. Eligible for citation and inclusion. | Iterate. Sustain freshness and citations. |
| 85-100 | Excellent | Best-in-class. Likely to be quoted verbatim. | Defend. Maintain quality and monitor. |
The five scoring pillars in detail
Crawlability
Status code, response time, redirect chain length, robots.txt rules and content size. A 200 OK in under 800ms with a clean robots policy is the baseline. Pages behind aggressive bot-fight rules score 0 here regardless of content quality.
Structure
Heading hierarchy (one H1, sequential H2-H6), semantic HTML usage, schema.org markup (Article, FAQPage, HowTo, BreadcrumbList, Product, Organization where relevant), and the presence of FAQ blocks. Schema is the single biggest lever in this category.
Content quality
Word count, reading level, answer density (declarative answers vs filler), use of lists and tables, definitional clarity, expert quotes. AI engines reward content that answers a question directly in the first 100 words.
Citations
Outbound citations to authoritative sources, source diversity, and inbound expertise signals. Pages that cite sources well are themselves cited well - the relationship is reflexive.
Freshness
Visible date, last-modified header, year-currency in body copy, sitemap lastmod. Content dated in the current year gets a meaningful boost; content dated more than three years old gets quietly demoted.
Quick wins for each category
- Crawlability - confirm
robots.txtallows AI bots, remove non-essential redirects, ensure HTTPS. - Structure - add Article schema and FAQPage schema where appropriate. Fix heading skips (H2 then H4 with no H3 is a common error).
- Content quality - add a 1-2 sentence answer capsule at the very top. Convert dense paragraphs to bulleted lists where possible.
- Citations - add 3-5 outbound links to authoritative sources. Quote a recognised expert with attribution.
- Freshness - update the visible date and
dateModifiedin your schema after any meaningful edit.