On this page
Ask ChatGPT who to hire for a panel upgrade in your city. Ask it again tomorrow. You will probably get a different answer, and there is a decent chance it names nobody at all.
We know this because we have been asking, every day, for one real business.
The business is an electrical contractor in the Vancouver area. It is a customer, and it agreed to let the account run as our public demo. The numbers below are what four AI systems — ChatGPT, Gemini, Google AI Mode and Google AI Overviews — actually said when asked the questions its customers ask.
The headline number
Across one group of questions about the cost of a panel upgrade, the tool collected 43 answers. The business was named in zero of them.
That is not the interesting part. This is: in 38 of those 43 answers, the AI named no company at all. It explained what a panel upgrade costs, described what affects the price, and stopped.
Nobody won that question. It is sitting unclaimed.
Most conversations about AI search assume a race where somebody is already ahead. In a lot of categories there is no race yet, because no business has given the AI systems anything specific enough to name.
The gap between engines is enormous
The same questions, asked of four systems on the same day, produce wildly different results.
If you check one system and conclude you are invisible, you may be wrong. If you check one system and conclude you are doing fine, you may also be wrong.
Worse, Google AI Overviews frequently does not appear at all for these queries. For one city, it never appeared in three months of daily checking. That is not a score of zero — it is a different situation entirely, and treating the two as the same is how people end up chasing problems that do not exist. We explain how we separate the two on the how it works page.
One city strong, the next city invisible
In one suburb the business was named in 22 to 28 answers out of roughly 33. Strong.
Three questions about the neighbouring suburb, 80-plus answers collected: zero. A competitor appeared in about half of them, and the AI systems read that competitor's website roughly thirty times over the period.
Same company. Same trade. Same quality of work. Twenty minutes apart on the map.
The difference was not reputation. It was that one city had a page written about it and the other did not. AI systems are reading pages and repeating what those pages claim. No page, no claim, no mention.
The AI reads your site and still doesn't name you
This one took a while to spot, and it is the finding that changed how we think about the whole problem.
For emergency electrician questions, the AI systems were reading the business's own page. We can see the URL in the sources. And then they answered the question without naming the business.
The page was being used as background material. It explained what to do in an electrical emergency, in general terms, without ever making a specific, attributable claim about who to call, where, and for what.
So the AI took the explanation and left the company behind.
That pattern — read but not named — turns out to be common, and it is invisible unless you are watching the sources as well as the answers.
Why the numbers swing so much
One more thing worth knowing before you go and check your own business.
We ran the identical question, to the identical system, four times in one day. The business was named 13 times, then 25, then 18, then 20.
Nothing changed in between. That is just how these systems behave.
This has a practical consequence: a single check tells you almost nothing. If you asked ChatGPT about your business once last week and felt reassured, or alarmed, you were reading noise. Only a rate measured over many samples means anything, and it should always be reported with a range around it.
We are fairly blunt about this because the alternative is selling people a number that looks precise and isn't.
Why this matters more than it did last year
The reason any of this is worth measuring is that the answer increasingly replaces the click.
A Pew Research Center study of 900 US adults found that when an AI summary appeared on a Google results page, people clicked a traditional search result in 8% of visits, against 15% when no summary appeared. Clicks on the sources cited inside the summary ran at about 1%. Coverage of the study in The Register put the same point more bluntly: the summary satisfies the question, and the visit never happens.
If the click is disappearing, being named in the answer is the visibility that remains. That is a different thing from ranking, and it needs its own measurement.
What we take from all this
Four things, in the order we would act on them:
- Check whether anyone is being named at all. If the answers name nobody, the question is unclaimed and it is far cheaper to win than one a competitor already owns.
- Check every engine, not one. The spread between ChatGPT and Google AI Mode was consistently more than ten percentage points.
- Look at what the AI is reading. Being read without being named is a content problem with a specific fix, and you cannot see it from the answers alone.
- Measure repeatedly, or don't bother. One check is a coin flip. A rate over weeks is a fact.
None of this is theory. It is one real account, measured daily, for three months.
Questions people ask
How often should I check whether AI mentions my business?
Daily, if you want a number you can trust. The same question asked of the same system four times in one day produced four different answers in our data, so a single check is noise. A rate measured over weeks, with a range around it, is the smallest unit of information that means anything.
Does ranking on Google mean AI will name my business?
No. They are separate systems with separate inputs. In our measurements the business ranked well in ordinary search for questions where AI systems named it in none of the answers, and Google AI Overviews often showed no AI answer at all for queries where the business ranked on page one.
Why would AI read my website and not name my company?
Because the page explains a topic without making a specific claim the AI can attribute to you. General advice gets absorbed as background. A page that states plainly what you do, where you do it, and for whom gives the system something it can repeat with your name attached.
Which AI system matters most for local businesses?
In our data Google AI Mode named the business most often, at around 30% of answers, and ChatGPT least, at 13-19%. That ordering may differ in your category, which is the reason to measure all of them rather than assume.
Can I check this myself without a tool?
You can ask ChatGPT a few questions and get a rough feel, and it is worth doing once. What you cannot do by hand is ask the same questions of four systems every day, record which sources each one read, and turn that into a rate with a confidence range. That is the part we automate — you can see the whole method on the how it works page, or start with the question finder.
Harjeet Singh Bedi Founder, LLM Visibility Lens · SEO consultant, Bedi SEO
Harjeet is an SEO consultant who runs Bedi SEO and works hands-on with local businesses, online stores and agencies. He built LLM Visibility Lens after clients kept asking whether they showed up in ChatGPT and nobody could give them an honest, measured answer. Every figure in this article comes from answers his tool collected.