How we score AI visibility.

The rubric

26 scored checks across eight dimensions, weighted to 40.5 points and normalised to a 0–100 score. Each check returns pass, warning or fail; a warning earns half credit. Two more checks are added only when they fire — a site that refuses our reader outright, and one serving a broken certificate chain. Seven further results are reported on every scan and never scored; they are listed below the bands, with the reason for each.

One rule sits outside the points. If robots.txt blocks OAI-SearchBot, Claude-SearchBot, PerplexityBot or Googlebot entirely, that block goes to the top of the fix list, the summary names it, and the score is held at 79 or below until it is fixed. At half a point an agent the arithmetic cannot say what a block costs, and the top band claims engines can find you, which a block makes untrue.

Dimension	Weight	Share	What it measures
Structured data	8	20%	Whether an engine can read your business as data rather than inferring it from prose — and whether you use a specific type (Dentist, LegalService, Restaurant) rather than a generic one. FAQPage markup is reported, not scored.
Identity & entity clarity	8	20%	Whether the page states who this business is, unambiguously, in both the visible copy and the markup.
NAP completeness	8	20%	Phone, a complete postal address, and opening hours — as structured data, not just text. The most common gap we find in the field.
Answerability & content	6	15%	Whether any page answers the literal question a customer types, in a form an engine can lift.
Authority signals	4	10%	Linked, claimed profiles — and enough of them to corroborate the business across independent sources.
Technical foundation	3.5	9%	HTTPS, canonical, viewport, Open Graph, sitemap. Genuinely table stakes, and weighted accordingly. llms.txt is reported, not scored.
AI crawler access	2	5%	Whether robots.txt blocks the four agents that decide whether an engine can use your pages in a search answer: OAI-SearchBot for ChatGPT search, Claude-SearchBot for Claude, PerplexityBot for Perplexity, and Googlebot for Google Search, AI Overviews and AI Mode included. The training controls, GPTBot, ClaudeBot and Google-Extended, are reported and not scored. Half a point an agent, because almost every site passes; a full block is handled outside the points, as described above.
Reviews & reputation	1	2%	Whether the page links the review profiles where your reputation actually lives: Google, Yelp, an industry platform. That reputation sits off your site, where this scan cannot read it, so the dimension is small. Rating and review markup are reported, not scored.

What the bands mean

Score	Tier	Meaning
80–100	Recommendable	Engines can find, read and trust the business well enough to name it confidently.
60–79	Findable	Engines can identify the business, but gaps cost it recommendations against better-described competitors.
40–59	Unconvincing	Engines can see the site, but too much of what a recommendation needs is missing or unreadable.
20–39	Unreadable	Engines cannot describe the business well enough to name it.
0–19	Undescribed	Nothing on the page gives an answer engine enough to recommend the business by name.

What the scan reports but does not score

Each of these used to cost points or was the wrong thing to test. They stayed in the report because they are worth knowing; they left the score because the providers' own documentation does not support grading a site on them.

• GPTBot, ClaudeBot and Google-Extended. GPTBot and ClaudeBot are OpenAI's and Anthropic's training crawlers. Google-Extended is a robots.txt token, not a crawler, that limits use of your pages for training Gemini and the models behind Search's generative features, and for grounding in Gemini Apps. Blocking any of them is a decision about model training. It does not remove you from ChatGPT search, Claude search or Google Search, so the scan says what the setting does and leaves the choice with you.
• llms.txt. Google says Google Search ignores the file and that publishing one neither helps nor harms. OpenAI, Anthropic and Perplexity document their crawler controls in robots.txt and say nothing about reading it.
• FAQPage markup. Google stopped showing FAQ rich results for all sites on May 7, 2026, and says its AI features need no special schema. Microsoft says schema of every kind, FAQ included, helps AI systems understand a page, but no engine has published that FAQ markup decides what it quotes. Questions answered in visible text are scored, under Answerability; the markup is not.
• AggregateRating and Review markup. Google shows no review stars on a business's own pages for reviews the business controls, and says not to mark up ratings gathered from other sites, such as a Google rating copied into your schema. A scan that failed you for lacking the markup was pushing toward the one thing the guidelines rule out.

What this scan cannot tell you

A methodology page that admits its limits is more useful than one that claims completeness — and this list is the part we would want to read first if someone else had written it.

• It does not ask an answer engine whether you are named. The scan reads the signals that decide whether you CAN be recommended; it does not run live queries against ChatGPT, Perplexity or Google AI Overviews. That check is part of our human work, and we say so on the audit page rather than implying the tool does it.
• It reads one page — the URL you give it — plus robots.txt, your sitemap and llms.txt. A strong interior page will not rescue a weak homepage score, and vice versa.
• It reads robots.txt, not your CDN or firewall. A bot rule there can turn away a search agent without appearing in the file, and the scan will still report that agent as allowed.
• It cannot see your Google Business Profile directly. It infers profile health from what your site publishes and links to, which is a proxy and sometimes a poor one.
• It does not judge whether your copy is any good, whether your prices are right, or whether your reviews are deserved.
• A high score is not a promise of being recommended. It means the obstacles we can measure are cleared.
• It is not the instrument behind the Front Range AI Visibility Index. That research graded businesses on whether answer engines named them, across seven batches in June 2026, on a different seven-tier scale. This rubric scores a site's readiness and queries nothing. The two share a section of this site and measure different things, so do not read a tier on one as a tier on the other.

Why we rebuilt this scale

An earlier version of this rubric was wrong in a way worth publishing. It weighted AI crawler access at more than a fifth of the grade, and because almost no site blocks those crawlers, nearly everyone collected those points for free. Combined with checks that could only ever return "warning" rather than "fail", the arithmetic floor for any site on HTTPS was 46 out of 100 — the bottom two tiers were literally unreachable, and a practice with no structured data at all could score 68 and be told it had "a solid base".

It also failed sites for being more specific than required: a dental practice correctly marked up as Dentist failed a check for LocalBusiness, because the comparison was a string match rather than a type hierarchy. And its contact check passed on the mere presence of the word PostalAddress anywhere in the HTML — which meant our own site, with no phone number on it at all, passed a check for having a phone number.

The current scale weights what a recommendation actually depends on, and most checks can now fail outright. Scores went down across the board, including our own. That is the point of publishing the rubric rather than the score.

A second correction followed on 13 September 2026. The crawler check tested GPTBot, PerplexityBot, ClaudeBot and Google-Extended, and told a site that blocked GPTBot it was invisible to ChatGPT. That was wrong: three of those four govern model training, not search answers, and our own crawler guide already said so. It now tests the search agents and reports the training controls without scoring them. The same revision stopped failing sites for missing llms.txt, FAQPage and rating markup, and stopped recommending the deprecated Attorney type. Scores rose for most sites that lacked that markup. Those points should never have been withheld.