Skip to content
Back to writing
Method

AI visibility score: what it measures and what it hides

An AI visibility score squeezes mentions, position, citations and sentiment into one number. Here is what goes into it, what falls out, and what to measure instead.

10 Nov 2026 - 7 min read
On this pageShow 8Hide

Many AI visibility tools lead with a number. Your brand is a 42, or a 67, up three points since last week. It looks like a credit score for AI search, and it drops neatly into a board deck.

The trouble is what the number is made of. An AI visibility score blends several different measurements into one figure, and the thing you most need to know, which competitor the engines name ahead of you and why, is easy to lose. This guide covers what goes into a score, what it hides and how to measure AI visibility in a way you can act on.

What an AI visibility score is

An AI visibility score is a single figure, usually from 0 to 100, that summarizes how a brand shows up in AI answers across a set of prompts and engines. A tool puts prompts to engines such as ChatGPT, Gemini, Perplexity and Google AI Overviews, reads the answers, and turns what it finds into points.

No tool sees inside the engines. Google says in its AI optimization guide that no outside tool has access to its internal ranking or AI systems. So every score is built from the same raw material you could collect yourself: the answers.

Free scan

Check your AI visibility free

See which buying questions ChatGPT, Perplexity, Gemini and Google's AI answers name you for, and who they name instead. No call, no card.

What goes into the number

Vendors weight things differently, and many do not publish the formula. The ones that describe their scores show the range of what gets blended. On their own product pages, read on 31 October 2026:

  • Yoast describes the AI visibility index in Yoast SEO AI+ as "a score out of one hundred that blends mentions, ranking, citations, and sentiment."
  • Surfer reports a Visibility Score, which it describes as "how often and how prominently you appear," beside a separate mention rate and average position on its AI tracker page.
  • HubSpot's free AI Search Grader gives a total out of 100 across five dimensions: sentiment (up to 40 points), presence quality (20), brand recognition (20), share of voice (10) and market competition (10). HubSpot says the grader asks ChatGPT, Perplexity and Gemini how they characterize your brand based on their training data.

Across these, a score is some mix of five inputs:

  1. Mentions: how often the answers name you.
  2. Position: how early you are named when you are.
  3. Citations: how often the answers link to your pages.
  4. Sentiment: whether the wording about you is positive, neutral or negative.
  5. Competitive share: how you compare with the other brands named.

Each of those is worth measuring. The problem starts when they are added together.

What one number hides

The competitor named ahead of you

A score tells you how you did. It rarely tells you who beat you. If a rival is named first in eight of ten answers and you are named third, a blended score can still look healthy, because you were mentioned and the tone was kind. In HubSpot's published weighting, share of voice is worth 10 of the 100 points and sentiment up to 40, so a brand described warmly but named behind a competitor can still post a strong total.

The engine you are missing from

Engines behave differently, even inside Google. Google says AI Overviews and AI Mode "may use different models and techniques, so the set of responses and links they show will vary" (Google Search Central). An average across four engines can hide a brand that is named in nearly every ChatGPT answer and in no Gemini answer. That split points to where the work is, and a blended number erases it.

Mentions versus citations

A mention means the answer names you. A citation means it links to one of your pages as a source. They can move independently. A brand can be named often because roundups and reviews on other sites describe it, while its own pages are never cited. Blend the two and you cannot tell which one moved.

Why the number moved

The models are not deterministic. OpenAI's API documentation says outputs "may differ from request to request," and the Gemini app lets you regenerate a response and switch between versions. AI Overviews also often do not appear at all: Google says they are shown only when its systems decide they add something to classic Search. A three point drop might be a competitor winning a roundup, an AI Overview that did not trigger that day, or ordinary variation. The score cannot tell you which.

The sources

The most useful part of an AI answer is the list of pages it drew on. When ChatGPT searches the web, its answers may include citations and a Sources panel, according to OpenAI's help center. Perplexity says each answer includes numbered citations linking to the original sources (Perplexity). Those pages explain why the answer came out the way it did. A score throws them away.

Two brands, one score

Here is a made up example to show how much a single number can flatten. Suppose a tool scores two brands at 50, each on the same 20 answers: five prompts on four engines.

  • Brand A is named in 10 of the 20 answers, spread evenly across the four engines, usually third or fourth in a list, and never cited.
  • Brand B is named in all five ChatGPT answers and all five Perplexity answers, first each time, with its own pages cited, and in none of the Gemini or AI Overview answers.

Same score, opposite problems. Brand A needs to get onto the pages the engines cite so it is named earlier and more often. Brand B is strong on two engines and has a specific gap on the two Google surfaces, which points at the pages Google ranks for that topic. One number would have sent both teams into the same meeting with nothing to act on.

How to measure AI visibility instead

If you want numbers, keep them separate and keep the answers behind them.

  1. Prompt level answers. For every prompt and engine, keep the full answer text. That is where you judge accuracy and tone in context, rather than as a percentage.
  2. Mention share and citation share, by engine. Count how often each brand is named, and how often each brand's own pages are cited, engine by engine. Our guide to AI share of voice walks through the arithmetic.
  3. Who is named instead. For every answer that leaves you out, record the brands it names. That list is your real competitive set in AI search, and it may not be the one in your sales deck.
  4. Cited sources. Record the pages behind each answer and sort them by type: roundups, comparisons, reviews, forum threads, competitors' own pages. Those are the pages to get onto.
  5. Your Google ranking beside the answer. Google says its generative AI features rely on its core Search ranking systems to retrieve pages from its index (Google). Seeing your position for the keyword behind a prompt next to the AI answer tells you whether the gap is in the rankings or in the answer.

For your own site, Search Console has a Generative AI performance report with impressions from AI Overviews and AI Mode (Google). It does not show competitors or other engines, but it is Google's own data and it is free.

If you still want a score in your AI visibility report

Scores have one honest use: a quick trend line in a monthly report. If you report one, do three things.

  • Ask what it blends and how each input is weighted. If the vendor will not say, treat the trend as a rough signal only.
  • Report it per engine, never as one average across all of them.
  • Keep the answers one click away, so anyone who asks why it dropped can read what the engines actually said.

Read the answers, not the number

We built our free checker around this view, so weigh this section accordingly: it deliberately shows answers, not a score. You enter your domain, confirm five buyer prompts drafted from your site, and about two minutes later you can see the answers behind your score: 20 answers across Google AI Overviews, ChatGPT, Gemini and Perplexity, the pages each engine cited, the brands named instead of you and your Google position for the keyword behind the prompts. No card and no email.

If you want the same reading every day, the alwaystracked plan checks 50 prompts and 10 keywords daily on those four engines, from $129 a month with a 14 day free trial. It keeps every answer with its sources, with citation and ranking side by side and never averaged into one figure.

Questions

What is a good AI visibility score?
There is no standard scale. Each vendor blends different inputs with different weights, so a 60 in one tool is not a 60 in another. Compare yourself with the brands you lose deals to, on the same prompts, engine by engine, and watch the direction over several weeks.
How do I measure AI visibility without a score?
Run a fixed set of buyer prompts across ChatGPT, Gemini, Perplexity and Google AI Overviews, keep every answer, and count two things per engine: how often you are named and how often your own pages are cited. Then record which brands are named when you are not, and which pages the answers drew on.
What is an AI visibility index?
It is a name some vendors give their composite score. Yoast, for example, describes the AI visibility index in its AI+ plan as a score out of one hundred that blends mentions, ranking, citations and sentiment. Like any composite, it is only as useful as your access to the answers behind it.
Can Google tell me my AI visibility?
For Google only, and for your own site only. Search Console's Generative AI performance report shows impressions for your pages in AI Overviews and AI Mode. It does not cover ChatGPT, Gemini or Perplexity, and it does not show competitors.
Free scan

Check your AI visibility free

See which buying questions ChatGPT, Perplexity, Gemini and Google's AI answers name you for, and who they name instead. No call, no card.