What does an AI visibility score actually measure?
An AI visibility score measures the share of AI answers that mention your company, usually expressed as a percentage or a number out of 100. It is useful for spotting a trend, but on its own it says nothing about who is being recommended instead of you.
Most scores are built from a set of questions put to one or more AI assistants, such as ChatGPT, Gemini or Perplexity, and a count of how often your name appears. Two tools using different questions will reasonably give different numbers, which is why a score is best read as a tripwire rather than a verdict.
Why is a percentage the wrong unit for a contest?
Because B2B buying is a contest for a short list, not a climb to 100%. When a buyer asks an AI assistant which providers to consider, it typically names a handful of companies, and the buyer works from that list.
What decides whether you are on it is not your absolute score but your position against the companies your buyers compare you with. The table shows why the same score can mean opposite things.
| Company | Score | Leading peer | The Gap | What it means |
|---|---|---|---|---|
| Company A | 40 | 30 | 10 points ahead | Winning its shortlist |
| Company B | 40 | 80 | 40 points behind | Losing most shortlists |
What is the Gap?
The Gap is the leading competitor's AI visibility minus yours, in points. If AI names your leading competitor in 72% of answers to your buyers' questions and names you in 18%, the Gap is 54 points.
The leading competitor is not chosen by you. It is the direct peer that AI names most often in the answers, because that is the company your buyers hear about instead. The Gap therefore means the same thing in every market: how far you are from the place on the shortlist that someone else currently holds.
How do you measure the Gap fairly?
By removing every way the measurement could flatter you. Four rules matter most.
- Questions written blind. Write the questions from the category alone, without your name, your website's wording or your competitors. Questions built from your own keywords tend to find you.
- The whole buying committee. Ask as the CFO or CEO, the person who owns the problem, the systems lead and procurement would ask, across shortlist, evaluation and decision.
- Live answers, as a new buyer sees them. Use named platforms with their own web search on, from the market you sell into, with no login or chat history. AI assistants remember you and tend to show you yourself.
- A stated margin of error. With 36 answers the margin is about ±13 points, enough to size the Gap. With 900 answers it narrows to about ±3 points. Changes inside the margin are not progress.
Why does the measurement need repeating?
Because AI assistants change where they look without notice. Between late July and mid-August 2026, Reddit's share of ChatGPT Search citations fell from 3.83% to 0.52%, a drop of about 86%, according to provisional tracking data reported by Relevant Audience.
Obvious fixes do not always hold either. When Ahrefs tracked 1,885 pages that added structured data, it found no meaningful uplift in AI citations on Google AI Mode or ChatGPT over 30 days. A Gap measured once is a photograph; measured the same way each quarter, it shows the direction of travel.
What should a leadership team ask first?
Three questions, in this order.
- Who does AI recommend to our buyers instead of us?
- How far behind them are we, in points?
- Which of the four causes explains it: AI never found us, found us but chose someone else, confused us with another business, or knew us but found less proof?
A Gap Check answers the first two. Send your website and the market you want measured, and Genivista sends the report, at no charge and with no meeting.