What Is an AI Visibility Score? How Businesses Can Measure AI Search Presence
An AI visibility score summarises how often, and how prominently, AI search answers name your business. Here is what goes into a well-designed score, how to calculate one, and the limits you should keep in mind.
VisiScout Team6 min read

As more customers ask AI assistants for recommendations, businesses want a way to answer a simple question: are we showing up? An AI visibility score is one way to turn many individual AI answers into a single number you can track over time.
This article explains what an AI visibility score is, what a well-designed one should measure, how a composite score is calculated, and how to read it without overstating what it means.
What an AI visibility score is
An AI visibility score summarises how your business appears in answers from an AI search tool, such as ChatGPT Search, for a defined set of questions in a defined market.
It is not a universal ranking. There is no official "AI rank" published by AI providers, and no score can tell you how every possible customer question is answered. A useful score is always tied to:
- A specific AI search tool, because different assistants produce different answers.
- A fixed set of prompts, the questions you choose to monitor.
- A market, because answers depend on location.
- A clear formula, so you know exactly what moves the number.
If a vendor cannot tell you those four things, treat the score with caution.
Why a single score is useful
Raw AI answers are messy. One prompt might name you first, the next might not mention you at all, and a third might cite your website without listing you. Looking at dozens of answers individually makes it hard to see whether things are improving.
A score helps you:
- Track a trend across weeks or months.
- Compare segments, such as local versus commercial prompts.
- Communicate clearly with a team, a client or an agency.
- Spot changes after website updates, new content or new reviews.
The score should be a summary, not a replacement for the underlying detail. You should always be able to see which prompts and answers produced it.
The three things worth measuring
Most of the useful signal in an AI answer comes from three questions.
1. Are you mentioned? (brand visibility)
The most basic measure is the share of answers that name your business. If you monitor five prompts and check each three times, you have 15 answers. If you are named in 9 of them, your brand visibility is 60%.
Good matching matters here. Your business may be known by several names ("Lakeview Physical Therapy", "Lakeview PT"), and a score should count all of them while avoiding false matches on common words.
2. How prominently are you recommended? (prominence)
Being named first in a list of five is not the same as being named fifth, or mentioned in passing after the list. A prominence measure gives more credit for higher positions. For example:
- First position earns full credit.
- Each lower position earns somewhat less.
- A mention without any numbered list earns partial credit.
- A mention that appears outside a ranked list earns less again.
The exact credits are a design choice. What matters is that they are fixed and published, so the number means the same thing every time.
3. Is your website cited? (citations)
AI search answers often link to the sources they used. A citation to your own website shows that the answer drew on your content, and it can send visitors to your site. The citation measure is the share of answers that cite your domain.
Branded versus unbranded prompts
A critical design decision is what to do with branded prompts, questions that already contain your business name, such as "Is Lakeview Physical Therapy good for back pain?".
These almost always mention you, so including them would inflate the score without telling you anything about discovery. A sound score:
- Uses unbranded prompts for the headline number.
- Reports branded prompts separately, as a check on how you are described.
- Treats a prompt that names your business or domain as branded even if it was labelled otherwise, so a labelling mistake cannot inflate the result.
How a composite score is calculated
A composite score combines the components with fixed weights. As an example, this is the formula behind the AI Visibility Index that VisiScout is preparing:
| Component | What it measures | Weight |
|---|---|---|
| Brand visibility | Share of unbranded answers that name you | 50% |
| Recommendation prominence | Position-weighted credit when you are named | 30% |
| Website citations | Share of unbranded answers that cite your website | 20% |
Each component is scored from 0 to 100, then combined:
Index = 0.5 × visibility + 0.3 × prominence + 0.2 × citations
A worked example
A physical therapy clinic monitors five unbranded prompts, each checked three times, for 15 answers:
- It is named in 9 of 15 answers, so visibility = 60.
- Its position-weighted credit across those answers averages 42.7 out of 100.
- Its website is cited in 5 of 15 answers, so citations = 33.3.
Index = 0.5 × 60 + 0.3 × 42.7 + 0.2 × 33.3 = 30 + 12.8 + 6.7 = 49.5
A separate branded prompt mentioned the clinic in 3 of 3 answers. That is reported on its own and does not change the 49.5.
When a component cannot be measured
If a business has not provided its website domain, citations cannot be measured. Rather than scoring citations as zero, which would unfairly lower the result, a well-designed score redistributes that weight across the other components and makes clear that citations were not measured.
Handling variability
AI answers change from one request to the next. A score built on one answer per prompt can swing widely for no real reason. Good practice is to:
- Check each prompt several times per measurement and score all of the observations.
- Exclude failed or empty responses instead of counting them as "not mentioned".
- Label small samples as preliminary until there are enough successful observations to be meaningful.
- Version the formula. If the scoring method changes, results from the old method should be labelled and not compared directly with the new one.
How to read your score
Score bands make a number easier to interpret. VisiScout's upcoming AI Visibility feature will use these descriptive bands:
- 0–30 Low: your business rarely appears.
- 31–60 Emerging: you appear in some answers.
- 61–80 Strong: you appear in most answers.
- 81–100 Leading: you appear in nearly every answer.
These are guides, not industry benchmarks. A score of 49.5 in a crowded city market may be a solid starting position; the same score in a niche with few competitors may suggest room to improve. The trend over time, and the comparison with competitors in the same answers, are usually more informative than the absolute number.
What an AI visibility score cannot tell you
Be clear about the limits:
- It does not measure how many people saw an answer or clicked through.
- It covers only the prompts you monitor, not every question customers ask.
- It reflects one AI search tool at a time.
- It does not guarantee future answers. AI systems change, and results can shift without anything changing on your side.
Used with those limits in mind, a score is a practical way to track a channel that traditional analytics does not show.
Improving the components
Each component points to different actions:
- Low visibility: make sure your website clearly describes your services, locations and specialties, and that your business details are consistent across directories and your Google Business Profile.
- Low prominence: look at which competitors appear above you and what their websites and reviews communicate that yours does not. Competitor research tools such as Competitor Intelligence help here.
- Low citations: publish pages that directly answer the questions in your prompts, and keep your site technically healthy so it can be crawled and understood. A free website audit is a quick first check.
None of these steps guarantees inclusion, but they improve the information AI search tools can draw on, and they help your customers and traditional search at the same time.
Getting started
You can calculate a basic score yourself with a spreadsheet: pick five to ten realistic prompts, check each one three times in your market, and record mentions, positions and citations. Our guide on how to track your business's visibility in ChatGPT Search walks through the process step by step.
VisiScout is also preparing an AI Visibility feature that automates this for ChatGPT Search using the formula above. We will share more when it is available.
Related articles
Organic SearchGoogle Search Console Metrics Explained: Clicks, Impressions, CTR and Average Position
What clicks, impressions, click-through rate and average position in Google Search Console really mean, why they behave the way they do, and how small businesses can use them to find realistic SEO opportunities.
Organic SearchHow to Track Your Business's Visibility in ChatGPT Search
Customers increasingly ask ChatGPT for recommendations. Here is a practical, repeatable way to check whether your business appears in those answers, what to record, and how to read the results without over-reacting to a single response.
Organic SearchBest Ways for SEO Consultants to Report Results to Small Businesses
How SEO consultants can report results that small-business owners understand: which Search Console and analytics metrics matter, why rankings alone mislead, a simple monthly report structure and common reporting mistakes to avoid.