
AEO Armory's AI visibility score is the average of points earned in each AI answer, scaled to 100. You get full marks when an answer names you first, fewer points when it names you lower down, a small amount when it cites your site without naming you, and nothing when it ignores you. There is no universal "good" number: a score means most next to your competitors' results and your own earlier reports. Here is exactly how it is worked out, since it is our tool and you should know what the number means.
Step 1: collect the answers
The audit asks ChatGPT and Google AI Mode 8 questions your customers would ask, which gives up to 16 answers. Each answer is one unit of the score. If an engine never returns an answer for a question, that cell shows as "no answer" and is left out rather than counted as zero.
Questions that mention your business by name are also left out. If someone asks "is Acme Roofing any good?", Acme being named says nothing about whether AI recommends it to strangers, so those rows are shown but not scored.
Step 2: decide whether you were named
For each answer, the audit looks for your business name as a whole word, your domain in the text, or, for longer names, the name with spaces and punctuation ignored, so "ignitevisibility" still matches "Ignite Visibility". In ChatGPT's local business listings it also checks the names and domains of the listed places. Separately, it checks whether your domain is among the sources the answer cited.
Want to see your own results?
AEO Armory runs the full audit free: real ChatGPT and Google AI Mode answers, 9 readiness checks and a prioritized fix list.
Run your free AEO audit →Step 3: score each answer
Your position is worked out among you and the competitors you entered, by where each name first appears in the answer. Then each answer earns points:
| What happened in one answer | Points |
|---|---|
| Named, first among you and your competitors | 1.00 |
| Named second | 0.85 |
| Named third | 0.70 |
| Named fourth or lower | 0.55 |
| Not named, but your site cited as a source | 0.30 |
| Not named, not cited | 0 |
One consequence is worth knowing: position is relative to the competitors you listed, not to every business in the answer. If you enter no competitors, any mention counts as first. Adding the one or two competitors you actually lose to makes the score stricter and more honest.
Step 4: average and scale
The score is the total points divided by the number of scored answers, times 100, rounded. A worked example with 16 scored answers:
- Named first in 2 answers: 2.0 points
- Named second in 2 answers: 1.7 points
- Cited but not named in 1 answer: 0.3 points
- Not present in the other 11: 0 points
That is 4.0 points out of 16, so the score is 25. Each engine score is the same calculation using only that engine's answers, which shows whether you are stronger in ChatGPT or in Google AI Mode.
The other numbers next to the score
- Answers that name you, for example 5 of 16. The plainest number in the report.
- Share of voice: the answers that named you, divided by all the times you or any of your entered competitors were named. If you were named 5 times and two competitors 9 and 6 times, your share of voice is 5 out of 20, or 25%.
- Site readiness: a separate score from 9 site checks. It measures whether AI can read and trust your pages, not whether it names you.
The report also puts a plain label on the score: 70 and up reads "AI recommends you often", 40 to 69 "sometimes", below 40 "rarely", and zero mentions "isn't recommending you yet". Those labels describe the number; they are not an industry benchmark.
So what is a good AI visibility score?
We do not publish a benchmark, because a fair one does not exist yet. The score depends on the questions, your market, the competitors you enter and the engines tested. A local plumber and a national software company are not playing the same game. Judge it two ways instead:
- Against competitors. If the businesses you lose to are named twice as often, that gap matters more than the absolute number. Share of voice and the competitor table show it directly.
- Against yourself over time. Run the same report, with the same inputs, after each batch of fixes. A trend from 20 to 35 across several runs means something; one run at 35 does not.
Why the score changes between runs
AI assistants do not give the same answer every time. Wording, the sources they pull in and the businesses they list vary from run to run, even minutes apart. With 16 answers, one answer naming you or not moves the score by up to about 6 points. That is why the audit tests several questions on two engines and why you should read the pattern rather than one result.
What actually moves it
- Being where AI already looks. The cited sources in your report are the directories, publications and threads the answers lean on. Getting listed or mentioned there is often the fastest route into the answer.
- Pages that answer the question. A page that directly answers "how much does X cost in Y" gives AI something to cite. The prompt plan lists them.
- Letting AI read your site. Blocked crawlers or JavaScript-only pages mean nothing else counts. These are usually at the top of the fix list.
- Your inputs. Changing the service wording, location or competitors changes the questions and the score. Keep them fixed when you compare runs.
How to track it sensibly
- Write down the inputs and keep them the same.
- Run a report before you start, then after each batch of fixes, and at least monthly.
- Look at the question table first and the score second: which questions moved from not named to named?
- Average two runs close together if you need a steadier baseline.
- Report the trend and the share of voice, not a single number.
Our guide to measuring AEO results covers tying this to leads and revenue.
Frequently asked questions
How is the AI visibility score calculated?
Each AI answer earns points: 1 for being named first among you and your competitors, 0.85 second, 0.7 third, 0.55 lower, 0.3 for being cited without being named. The score is the average across answers, times 100.
What is a good AI visibility score?
There is no fair universal benchmark. Compare your score with the competitors you lose to and with your own earlier reports, using the same inputs.
Why did my AI visibility score change when nothing changed?
AI answers vary from run to run. With 16 answers, one answer changing can move the score by several points, so look at the trend across runs.
Is the AI visibility score the same as site readiness?
No. Visibility measures whether AI answers name or cite you. Site readiness measures whether your site is easy for AI to read and trust.


