ChatGPT, Perplexity, Google AI Overviews, Gemini, and Copilot now account for a substantial share of product-discovery interactions among B2B buyers, by some measures over 40%.
Your visibility in those answers is not a function of your search ranking. You can hold the first organic position for a term and never be named when a buyer asks an assistant the question that term represents. Most companies have never checked whether that is happening to them, because the tooling to check is new.
Why the two diverge
Step 1
Write the twenty questions a buyer asks, in their own words
Step 2
Ask each across two or three assistants and record the answer
Step 3
Note whether you appear, and whether the description is accurate
Step 4
Fix the stale page causing any wrong characterisation first
Step 5
Repeat monthly. An hour beats a subscription.
Ranking is about matching a page to a query. Citation is about a system deciding your content is the most reliable available basis for a claim it is about to make.
Those reward different things. Ranking rewards structure, links, and relevance signals accumulated over time. Citation rewards specificity, verifiable numbers, named sources, and clear declarative answers. A page can be excellent at the first and unremarkable at the second.
What the tooling now measures
“Almost everyone who runs this exercise finds they are described inaccurately rather than omitted. That is a much easier problem, and it is caused by a stale page.”
A category of AI visibility trackers appeared over the past year: Profound, Otterly, Peec, SE Ranking's visibility product, Ahrefs Brand Radar, and others. They monitor how often and in what context a brand appears in AI-generated answers, reporting mention rate, citation frequency, share of voice, sentiment, and which of your pages earn the citations.
The genuinely useful features are narrower than the feature lists suggest. Seeing which exact prompts trigger your brand tells you what to write next. Seeing competitors named beside you tells you what the model considers your category. Seeing which page earned a citation tells you which format works.
Almost everyone who runs this exercise finds they are described inaccurately rather than omitted.
The free version of this
Turn what you know into what you own.
Vibepreneur builds structured ventures from professional expertise, with positioning, launch assets, and growth systems included.
Join the WaitlistYou do not need to buy a tool to start. Write down the twenty questions a buyer asks before purchasing something like yours, in the words they would use speaking to an assistant. Ask each one, across two or three assistants, and record whether you appear, how you are described, and whether the description is accurate.
That takes about an hour. Repeat it monthly. For most small companies the manual version surfaces the same three or four actionable findings a subscription would, and the discipline of doing it matters more than the automation.
The finding that usually comes first
Almost everyone who runs this exercise discovers the same thing initially: they are described inaccurately rather than omitted. The assistant knows the company exists and characterises it as something adjacent to what it is, usually because the clearest public description of the product is old.
That is a much easier problem than invisibility and it is fixed by publishing an unambiguous, current description of what the product does and who it is for. A stale pricing page or an out-of-date llms.txt actively propagates the wrong answer.
See generative engine optimisation for solo founders for the practice, and the citation economy for what actually earns a mention.