How the AI Visibility Score works
The score is not a number an AI model makes up. It is computed from 45 explicit checks across 8 dimensions. Every check records the evidence it used, and the same evidence always produces the same score.
We read your site the way most AI crawlers do: raw HTML, without running JavaScript, starting from your homepage and the pages that explain your business (pricing, services, about, contact, FAQ). An AI model helps extract facts, like what you sell and who you serve, and must quote your pages word for word. We verify every quote against your pages and lower confidence when a quote can't be found.
The 8 dimensions
Whether AI crawlers and assistants are allowed to fetch your pages and can read them without running JavaScript.
- Homepage responds quickly20 pts
- AI crawlers aren't blocked by bot protectionCaps the score at 30 when failing20 pts
- AI answer engines may read your site (robots.txt)Caps at 40 if all are blocked20 pts
- Content readable without JavaScriptCaps at 35 for JavaScript-only pages20 pts
- Homepage can be indexedCaps at 55 when noindex10 pts
- XML sitemap available5 pts
- Served over HTTPS5 pts
Whether a machine can state your name, what you do and what kind of business you are, consistently.
- AI knows your business name15 pts
- AI can say what you do30 pts
- Descriptive title and meta description15 pts
- Homepage H1 describes the business10 pts
- Consistent story across pages15 pts
- AI can categorize your business15 pts
Whether your products or services are named, described and matched to a clear customer.
- AI can name what you sell30 pts
- Offerings are described20 pts
- Dedicated products or services page15 pts
- Clear target customerNot applied to restaurants and hotels20 pts
- Offerings in structured data15 pts
Whether prices or the pricing model can be found and read. Expectations adapt to your category.
- Prices are readableProfessional services get credit for a stated pricing model45 pts
- Pricing page exists20 pts
- Prices in structured data20 pts
- Pricing model is clear15 pts
Schema.org JSON-LD, Open Graph and llms.txt: facts machines can read without guessing.
- Structured data (JSON-LD) present25 pts
- Organization identity in structured data25 pts
- Structured data is valid10 pts
- Category-relevant markup20 pts
- Open Graph summary card10 pts
- llms.txt guide for AIEmerging, optional convention; weighted lightly10 pts
Proof, identity, people, profiles and policies that let an assistant recommend you with confidence.
- About page15 pts
- Verifiable business details15 pts
- Social proof20 pts
- Linked official profiles15 pts
- Privacy policy and terms10 pts
- People behind the business10 pts
- No attempts to manipulate AI15 pts
- Your numbers are readable without JavaScriptOnly applied when counters read as 0 in the raw HTML10 pts
Whether an assistant or agent can contact, book, buy or start with you without a human in the loop.
- Machine-readable contact methods25 pts
- Clear primary action for your category25 pts
- Self-serve completion (book, buy, sign up online)20 pts
- Forms agents can fill10 pts
- Hours and locationLocal businesses only10 pts
- Machine-executable actions10 pts
Differentiation, comparable facts, service area and the questions your site leaves unanswered.
- Clear differentiation30 pts
- Concrete, comparable facts20 pts
- Clear service area20 pts
- Common questions are answered30 pts
Weights adapt to your category
A consulting firm shouldn't be judged on published prices the way an online store is. We detect your category and weight the dimensions accordingly.
| Category | Crawl | Business | Offering | Pricing | Structured | Trust | Agent | Recommendation |
|---|---|---|---|---|---|---|---|---|
| Default | 12% | 16% | 15% | 9% | 13% | 11% | 13% | 11% |
| Software / SaaS | 12% | 14% | 15% | 12% | 12% | 10% | 14% | 11% |
| Online store | 12% | 12% | 14% | 14% | 16% | 9% | 15% | 8% |
| Local business | 12% | 14% | 13% | 8% | 12% | 14% | 17% | 10% |
| Restaurant or food | 12% | 12% | 12% | 12% | 12% | 12% | 18% | 10% |
| Hotel or hospitality | 12% | 12% | 12% | 13% | 12% | 12% | 18% | 9% |
| Professional services | 12% | 17% | 16% | 5% | 12% | 15% | 12% | 11% |
| Creator, media or education | 14% | 20% | 14% | 6% | 12% | 14% | 8% | 12% |
| Nonprofit | 14% | 20% | 14% | 2% | 12% | 16% | 12% | 10% |
Caps: some problems outweigh everything else
If AI can't get in, nothing else matters much. These cap the overall score, and the report always says so:
- Bot protection blocks automated visitors: max 30
- Content only appears after JavaScript runs: max 35
- robots.txt blocks every major AI answer engine: max 40
- Homepage marked noindex: max 55
What we don't claim
- We can't see inside ChatGPT, Claude, Gemini or Perplexity. Nobody outside those companies can. We measure what is in your control: whether machines can read, understand and act on your site.
- A high score doesn't guarantee recommendations, and llms.txt is an optional convention, not a ranking switch.
- We measure what your website communicates on its own. Famous brands may be known to AI through other sources.
Bands
- 85 to 100: AI-ready
- 70 to 84: Strong, with gaps
- 50 to 69: Partially understood
- 30 to 49: Hard for AI to read
- 0 to 29: Nearly invisible to AI