In one sentence: AI visibility is how often, and how consistently, AI systems such as ChatGPT, Google AI Mode and AI Overviews name your business when someone asks them for a recommendation in your trade and your area.
The earlier pages explained what an answer engine is and how it builds a reply. This page is about you specifically: a plumber, a dentist, an HVAC company. What does the AI read about you, why does it name only a few businesses, and what does a “score” actually measure?
What does the AI read when someone asks about my trade?
Not your work. Not your truck. Not your reputation in the neighborhood. It reads text it can fetch in a few seconds, from a handful of sources. We know which ones because a study tracked 1.9 million citations across 200,085 local AI searches.[1]
Your Google Business Profile. The listing that appears on Google Maps. It was the single most cited source in local AI answers, at 28.63% of all citations.[1] Your name, category, services, hours, photos, service area, reviews and how you answer questions all live here. If the AI reads one thing about you, it is this.
Your website. Businesses’ own websites made up 42% of all citations and 93% of the unique domains cited.[1] This is where the AI reads the detail: which services you actually do, which towns you cover, what your process is, what you say about yourself in your own words. ChatGPT used a business’s own site 58% of the time in an earlier check.[2]
Your reviews. They live on Google, Yelp and elsewhere, and the AI reads them as evidence. A review that says “they replaced our water heater in Carmichael the same day” tells the AI three checkable facts at once.
Directories. Yelp was 9.53% of local citations, Facebook 2.23%, TripAdvisor 1.8%. Yelp appeared in 80% of ChatGPT’s local answers.[1] For a plumber that means Yelp; for a dentist it also means Healthgrades and Zocdoc; for HVAC it includes the manufacturer dealer locators and Angi. The AI uses them to confirm you exist and to see how you are described elsewhere.
Everything the AI says about you is assembled from these four. That is the whole input list.
Why does it name two, three or four businesses, and not ten?
Because it is writing a recommendation, not a directory.
Google Maps shows a list because a list is what a map interface does. A generative engine writes a paragraph. A paragraph that names ten plumbers reads like a phone book, so it does not. The measured averages: ChatGPT names 4.1 businesses per local answer, Google AI Mode 3.5, AI Overviews 2.5, with the range running from zero to seven.[1]
This matches how people decide. 72% of consumers consider three or fewer businesses before choosing one, and 75% decide in under 30 minutes.[3] The AI is compressing the list to the size people actually use.
The consequence is arithmetic. On Google Maps, a tracked business was mentioned in 66% of searches. In ChatGPT it was 33%, in AI Mode 32%, in AI Overviews 38%.[1] The AI door is roughly half as wide as the map door. It is also, per visitor, more valuable: people who arrive from an AI answer have already been told you are the pick, and AI-search visitors were measured at 4.4 times the conversion value of organic visitors.[4]
Does it name the same businesses every time?
No, and this surprises most owners.
The same prompt, run again, returned the same set of businesses only 20% to 33% of the time. Between 50% and 58% of businesses named in one run reappeared in the next.[1] And 10% to 11% of local prompts returned no business at all.
So AI visibility is not a light switch that is on or off. It is a frequency. A plumber with strong, consistent signals might be named in seven of ten runs; a plumber with a thin profile and no service pages might be named in one of ten. That frequency is what a visibility score tries to capture.
What does an “AI visibility score” actually measure?
A score is a way of turning that frequency into one number you can compare over time. There is no industry standard, so let me describe what a good one measures and what it does not.
A sound score starts with a fixed set of realistic prompts for your trade and area. For a local plumber that might be twenty prompts: “emergency plumber near me,” “who should I call for a slab leak in Natomas,” “best-reviewed plumber for tankless water heater install near Elk Grove,” and so on. It runs each prompt several times on each platform, because of the consistency problem above. Then it counts.
The main number is the share of runs in which you are named. Underneath it, a useful score also reports which competitors were named instead, which sources the AI cited when it did name you, and which of your four inputs (profile, site, reviews, directories) the AI seems to be reading or missing.
A score of zero to 100 is a convention, not a measurement standard. Two tools will give different numbers for the same business because they use different prompts and different platforms. What matters is that the same tool, with the same prompts, gives you a number today and a number in ninety days, so you can see whether the work moved it.
Commercial trackers exist at every price. Otterly, for example, charges $29 to $489 per month depending on how many prompts you track.[5] A free score from a service provider is usually a smaller version of the same thing: a handful of prompts, run once, to show you where you stand.
What a score does not measure: your ranking in the Google list, your ad performance, or the quality of your work. It measures one thing, whether AI systems name you, and that is enough, because it is the thing nobody else was measuring.
Let me do it for three businesses
A local plumber. Someone asks ChatGPT for a plumber to fix a leaking water heater in Carmichael. ChatGPT fans the question out, pulls the Google profiles of plumbers whose service area covers Carmichael, reads their websites for “water heater,” and reads reviews for the same words. It names four. The plumber whose profile lists water heater replacement, whose site has a water-heater page mentioning Carmichael, and who has fifteen reviews using those words is named in most runs. The plumber whose profile says “Plumber” and whose site says “Call for a quote” is named rarely, however good he is.
A dentist in Roseville. Someone asks Google AI Mode for a dentist that takes new patients and does same-day crowns. The AI reads profiles for “accepting new patients,” websites for “same-day crowns” or “CEREC,” and reviews for words like “crown” and “same day.” It names three. A practice with 1,000 or more reviews was mentioned about 93% of the time in one dental study, and independent practices took 72% of AI recommendations against 4.4% for large groups.[6] Size did not win. Detail did.
An HVAC company in Elk Grove. Someone asks an AI Overview who installs heat pumps in Elk Grove and whether they handle the rebate paperwork. The AI reads for “heat pump” and “rebate” across the same four inputs. The company with a page that explains the rebate process in plain words is the one with something quotable. AI Overviews name only 2.5 businesses on average,[1] so the quotable page is the difference between being one of them and not.
In all three, the AI never saw the work. It saw what was written about the work, in the four places it reads.
What this is not
AI visibility is not your Google ranking. You can be first on Google and rarely named by the AI, or the reverse. Only 19% of sources cited in AI Overviews overlapped with the organic top ten.[7]
It is not a fixed state. It is a frequency across many prompts and runs. A single “I asked ChatGPT and it didn’t say my name” proves little; ten runs across three platforms prove a lot.
It is not something you can buy from the AI. As of this writing, the businesses named in local AI answers are not paid placements.
It is not about being everywhere. It is about the four inputs saying the same, specific things.
It is not the whole picture of how customers find you. 23% of consumers used an AI tool at some point in their last local search; 71% used Google Search at some point and 52% started there.[8] AI is the third channel, and it is growing fastest: 45% of consumers used it for a local recommendation in the past year, up from 6% the year before.[9]
Where it is going
Forecasts, labelled as forecasts
Forecasts, labeled as forecasts.
Forecast one: the four inputs stay the four inputs. Profiles, websites, reviews and directories are what engines can fetch and verify. I see nothing that replaces them. What changes is how much of each the engine reads; citations per AI Overview more than doubled in fifteen months,[7] so I would expect deeper reading of your website, not shallower.
Forecast two: scores become routine. Just as owners learned to ask “where do I rank?” they will ask “how often does the AI name me?” Expect it to be a standard line on any marketing report within a few years.
Forecast three: the AI will start acting on the name it picks. Google’s AI Mode already books appointments through partners, and Yelp reservations are bookable inside ChatGPT.[11][12] When that reaches home services and dental, visibility and booking will merge: the business named is the business booked. That connects to the AI receptionist page.
Forecast four: verification stays human. 88% of people who use AI for a local recommendation double-check it.[10] I would forecast that number stays high, which means your reviews and profile keep two jobs: feeding the AI and reassuring the person the AI told.