AI Visibility Score™
Does AI recommend you?

It measures how your business shows up when ChatGPT, Claude, Gemini, and Perplexity are asked twenty real buyer questions about your service and your area — whether you were named, whether a competitor was named instead, and which sources the engines read to decide. The result is a composite score from 0 to 100, weighted by engine, plus four pillar scores that group the same responses by theme: Brand Visibility, Local Authority, Content Authority, and Personal Authoritativeness. Every mention is counted from an archived response, so the report is a record of what the engines said on that date, not an estimate of what they might say. It is the first reading in the measurement that entity work — starting with listings that agree, the subject of directory optimization — is designed to move.

ChatGPT, Claude, Gemini, and Perplexity, queried in parallel with the same twenty prompts. They are weighted differently in the composite: Claude, ChatGPT, and Gemini carry 30% each and Perplexity carries 10%, as stated on the report itself. They also reach your website differently — each vendor documents separate user agents for training, for user-triggered visits, and for search, and each can be allowed or blocked independently in your robots.txt file, which Google describes as the file that tells crawlers which URLs they can access on your site. A site that blocks one engine's search agent while allowing another's will show it on the platform-by-platform breakdown — and because a robots rule can change silently in a plugin update, it is one of the regressions website maintenance checks for on a schedule. The engine-differences table above links each vendor's own documentation, and Google's introduction to robots.txt covers the mechanism.

Yes. The score is free to run and the findings are yours whether or not you ever speak to us. The form asks for your name, business, website, service category, and a description of what you do, because the prompt set is generated from those details and the Personal Authoritativeness pillar tests whether the engines recognize you, the owner or leader, as an expert. SMS consent is required because results are delivered by text, and the consent language on the form spells out what you are agreeing to. There is no card, no trial, and no obligation. What is not free is the work of moving the score, which is the engagement the report is designed to scope honestly, drawn from the service portfolio rather than from a package.

Twenty prompts, written the way a buyer in your area would ask them, are run against all four engines. The archived responses are scored two ways. The composite weights each platform's performance — Claude 30%, ChatGPT 30%, Gemini 30%, Perplexity 10% — into one number. The four pillars regroup the same responses by theme, each scored 0 to 100, so the pillar average can differ from the composite, and the report tells you so. Separately, two live checks run against your site: whether AI crawlers can reach it, and whether it carries structured data the engines can parse. Google describes structured data as a standardized format for providing information about a page and classifying its content, which is why the second check matters for citation. Both checks feed Technical AI Readiness, the pillar built to clear them. Source: Google Search Central, Introduction to structured data markup.

There is no published benchmark, and we will not invent one. A "good" score is one that is higher than the competitors the engines named instead of you, on the prompts that represent real revenue, and that holds or rises on the next reading. The report gives you the comparison directly: your share of AI voice against the market leader's share, and the leaderboard of who the engines recommended. A business with a low composite but a high Content Authority pillar has a different problem from one with the reverse, and the plan differs accordingly. Treat the first reading as a baseline. The number that matters is the second one. A low Content Authority pillar, for instance, points at the pages themselves — before it points at anything technical; and a score that rises while leads do not is a conversion problem, not a visibility one.

Because one of the four pillars, Personal Authoritativeness, tests whether the engines recognize the owner or leader of the business as a subject-matter expert in the industry. Answer engines recommend people as well as companies, and in many service categories a named, findable expert attached to a business is a stronger signal than the business alone. The prompt set includes questions where that matters, and the pillar reports whether your name came back. Your name is used to generate those prompts and to deliver your results; the form's privacy language states how the information is used and how to request deletion. Reviews that name the owner are one of the places that presence shows up, which is why a reputation management program is part of moving this pillar.

It depends on which agent you block. The vendors publish separate user agents for model training, for user-triggered page visits, and for search, and they document that each control is independent. OpenAI, for example, states that a site can allow OAI-SearchBot to appear in search results while disallowing GPTBot for training. Perplexity states that PerplexityBot is used to surface and link websites in its results and is not used to crawl content for training. Google documents the same split for its own products — crawlers and fetchers that act either automatically or when triggered by a user request. A rule that blocks every agent from a vendor removes you from that engine's answers; a rule that blocks only the training agent generally does not. The report's live crawler check shows which agents your site currently refuses; the vendor statements are linked in the engine table above. Source: Google Search Central, Overview of Google crawlers and fetchers; robots rules are set at build, which is why crawler access is part of the launch checklist under website design.

A single reading is a baseline. Engine answers shift between runs — the report itself notes that results reflect the run's date — so a score that is checked once tells you where you stood on one date and nothing about direction. Monthly is the cadence we use in engagements, because it is frequent enough to catch a lost recommendation the month it happens and slow enough that a change is a trend rather than noise. Between readings, the work that moves the score is the entity, citation, and content marketing work described above — none of it changes the number overnight.

No. A ranking is a position on a list; an answer engine does not return a list, it returns a recommendation, and there is no second page. The score measures whether you are the recommendation, across four engines, for prompts that matter, and it measures it from archived answers rather than from an index. The two are related — Google states that its automated ranking systems are designed to prioritize helpful, reliable information created to benefit people, and the engine table above records that the same fundamentals govern its AI features — so strong ranking work still feeds the score, particularly on Gemini. But a business can rank well and never be named by an answer engine, which is the exact gap this instrument was built to expose. Source: Google Search Central, Creating helpful, reliable, people-first content; a business can also own the map pack through Google Business Profile optimization and still be absent from the answer.

You decide. The report gives you two self-serve actions targeted at checks you failed, and it names the competitors and sources you are up against; a business with an in-house team can act on that alone. If you want the full diagnosis, the A.R.C. Report covers the rest — the complete competitor leaderboard with receipts, every prompt and who wins it, the named source list in the order to pursue it, and a 30/60/90-day plan — and it is free to request for qualified partners. Nothing about the score obligates you to the report, and nothing about the report obligates you to an engagement. The score is the honest first step, and the honest second step is deciding whether the gap it shows is worth closing — the standard of evidence you should expect is the one in our case studies: the archived answer before, and the archived answer after.