What AI visibility is, and how to track it

Measure	What it records	Why it counts
Mention rate	The share of answers that name you	The headline number
Position	Where in the answer you appear	Buyers act on the first two or three
Share of voice	Your mentions against each competitor's	Shows who the assistants prefer
Cited sources	The pages and publications the answer drew on	The one measure you can act on directly
Accuracy	Whether what is said matches your own facts	A wrong answer can be worse than none

What good AI visibility looks like

The questions that count are the ones buyers ask while choosing: which tools in a category are best, how two vendors compare, what something costs. If ChatGPT lists five vendors for "best PR software for B2B teams" and you're not one of them, that buyer never hears of you, whatever your Google ranking. Being described correctly counts as much as being named, and the assistants often get it wrong. Studies that check AI answers against companies' own facts keep finding errors in most of them, such as an old price, a product that was renamed or a customer that left.

Where an answer comes from

ChatGPT, Perplexity and Google's AI Mode search the web before answering most buying questions, pick a handful of pages they trust and write the answer from those, with citations. Gemini and Google's AI Overviews draw on Google's index. What they read is lopsided. Most citations point to companies' own websites. Media are a small share of the rest, and on buying questions the media they cite are mostly trade and specialist titles, not the national press.

Which assistants to watch

For a B2B brand, watch ChatGPT, Gemini and Perplexity first. Every tracker we've compared covers them at entry level, and the citation research above draws mostly on them. Google's AI Mode and AI Overviews come next, because they sit in front of every Google search. Copilot, Claude and Grok are worth a look if your buyers live in them. Start with three and add others when the data gives you a reason.

Tracking it

Write 25 to 50 prompts the way buyers type them, covering category questions, best-of lists, a comparison with each competitor and price. Then fix the set, because a trend only means something if the questions don't move.
Pick three to five competitors, the ones you lose deals to. Share of voice is only useful against names your buyers already know.
Run them weekly. Assistants vary their answers from run to run, so daily movement is mostly noise, and four weekly runs show a trend. Go daily only during a launch or a crisis.
In every answer, record whether you were named, your position, who else was named, which sources were cited and anything said about you that's wrong.
Treat the cited sources as a work plan. They tell you which review sites, comparison articles and trade titles each assistant trusts for your category, and those are the places to be named.

A spreadsheet does this for a few prompts, and past that a tracker saves the hours.

Ours is AI Visibility Core, which runs 40 prompts a week on ChatGPT, Gemini and Perplexity for $139/month and checks every answer against the facts you've recorded.

How to improve your AI visibility

Start with your own site, written plainly, because most citations go to company websites. A homepage that says what you are, for whom and at what price does more than any campaign, and descriptions that conflict across your site, LinkedIn and review profiles lose, because the assistants merge them.

Then write a comparison page of your own. The page that answers the buyer's question, rivals included, is the one most likely to be cited. One UK agency using Agentcy went from being named in 9 of 496 AI answers to 175 of 493 in two weeks in September 2026, mostly on the strength of one such page; the weekly index keeps the diary.

Third parties give it staying power. Strip out the answers that read that agency's own site and it ranked far lower that week. Your own pages earn the jump. Making it last takes trade press, directories and reviews, which is PR work.

Questions

How is AI visibility different from SEO?

AI visibility measures whether an assistant names your brand in a written answer and what it says about you, while SEO measures where your pages rank in a list of links. The two overlap, because ChatGPT, Perplexity and Google's AI Mode search the web before answering most buying questions. A page-one ranking still doesn't guarantee a mention, since the assistant reads a handful of pages and writes from those. Accuracy is a new problem as well, because a search result shows your own words while an assistant paraphrases you, sometimes with an old price or a product name you've dropped.

What do GEO, AEO and LLM visibility mean?

GEO, AEO and LLM visibility are other names for AI visibility and the work of improving it. Generative engine optimisation (GEO) and answer engine optimisation (AEO) describe the work, while AI search visibility and LLM visibility describe the measure, which is how often and how accurately the assistants name you. The search trade tends to say GEO, marketers say AEO and tool vendors say LLM visibility. Buyers search for "AI visibility" most, so that's the term we use.

How do I check my AI visibility for free?

You can check your AI visibility for free by running the free checker on your domain, with no sign-up for the first result. It asks the assistants the questions a buyer in your category would ask and shows whether you're named, which competitors are named instead and which sources the answers cite. One run is a snapshot, because assistants vary their answers from run to run. For a trend, ask the same 25 to 50 questions each week and log the answers in a spreadsheet, or use a tracker once that gets slow.

What is a good AI visibility score?

There's no published benchmark for a good AI visibility score, because mention rates depend on the category, the questions asked and how many vendors compete for them. Judge your own trend over at least four weekly runs, and your share of voice against the three to five competitors you lose deals to. A mention rate that rises while a rival's falls is the result to aim for. Position counts too, since buyers act on the first two or three names, so fewer mentions near the top can be worth more than many at the bottom of a list.

Can I see AI visibility in Google Analytics?

Google Analytics shows the traffic that follows AI visibility, and nothing about the answers themselves. Referrals from chatgpt.com, perplexity.ai and gemini.google.com show up as sessions, and they're worth a channel group of their own so you can watch the trend. Analytics can't tell you how often you were named, who was named beside you or what was said. A buyer who reads an answer and searches for you by name later arrives as search or direct traffic, so treat the referral count as a floor and measure the answers directly.