AI Visibility Score: how to read your brand's AI visibility score
An AI visibility score is only useful if it says what it measures and what to do about it. Here is how the readiness score, the three axes and the mention rate work.
Why one number is not enough
"AI visibility" mixes two different questions:
- Is the site ready to be cited? That depends on what you control: crawler access, content clarity, proof.
- Is the brand being cited? That also depends on competitors, the sources each assistant uses and how answers vary from one run to the next.
A score that tries to answer both at once does not tell you what to fix. So PlaceFirstAI keeps them apart: the readiness score answers the first question, and the mention rate answers the second.
How the readiness score is built
The score comes from a set of signals checked on the site — for example, whether robots.txt allows AI crawlers, whether content is readable without JavaScript, whether there is an llms.txt, whether the page states clearly what the company does and whether there is evidence to support a recommendation. Each signal is rated pass, attention or fail.
The signals are grouped into three axes:
| Axis | The question it answers | Example signals |
|---|---|---|
| Access | Can the AI read the site? | robots.txt, content without JavaScript, llms.txt |
| Clarity | Does it understand what you sell? | Explicit offer, audience and region, answered questions |
| Preference | Does it have a reason to choose you? | Proof, reviews, concrete differentiators |
Looking at the axes matters more than looking at the average. A brand at 92 on Access, 65 on Clarity and 68 on Preference is not "mostly fine": it has one weak side, and that is where the recommendation stalls.
Signals that could not be read stay out of the calculation instead of counting as zero. A score that penalizes what it could not measure invents a problem.
How to read the bands
| Band | Score | What it usually means |
|---|---|---|
| Critical | 0–39 | Something stops assistants from reading or understanding the site. It is the most likely reason the brand is not cited. |
| Needs attention | 40–59 | The site is read but says little. There is a lot to gain from clarity and proof. |
| Good | 60–79 | A solid base. The contest shifts to competitors on comparison questions. |
| Strong | 80–100 | The site is ready. The work now is to keep it that way and win the questions where the brand still disappears. |
The other half: mention rate
The mention rate is the percentage of buying questions — tested on each assistant — in which the brand appears in the answer. Three things to keep in mind when reading it:
- Questions that already name the brand do not count. "Is [your brand] reliable?" always mentions the brand and would inflate the result.
- Answers vary. The same question can cite the brand in one run and not in the next. That is why each question is repeated, and unstable answers are flagged.
- Small swings are noise. A change of a few points between cycles may be sampling, not a trend.
The two numbers together
| Readiness | Mention rate | Reading |
|---|---|---|
| Low | Low | Start with the site: there are blockers or a lack of clarity. |
| High | Low | The site is ready; competitors own the answers. The work is content and presence beyond your site. |
| Low | High | The brand shows up in spite of the site, probably thanks to reputation. There is a risk of losing ground. |
| High | High | A strong position. Keep tracking it so you do not lose it. |
Where to see yours
The free diagnosis shows the readiness score across the three axes, the signal count and the highest-leverage step. The Growth plan adds the mention rate on ChatGPT, Gemini, Claude and Perplexity, measured every 15 days, with the competitors that show up instead. See how it looks in the sample reports.