How the score is calculated
A plain-English walkthrough of the AI Visibility score formula — how mentions, citations, ranking position, and buyer-intent queries combine into a single 0–100 number.
The AI Visibility score is a single 0–100 number, but it's built from a few simple ideas. This page opens the hood: what counts, what counts for less, and how it all rolls up. You don't need any of this to use the score — but if you want to know why yours is what it is, read on.
The 30-second version
Your score answers one question: when AI assistants help someone in your category, do they recommend you — and how prominently?
Two things drive it:
- Mentions — how often AI names your brand. This is most of the score.
- Position — when you are named, how near the top of the list you land.
We score every answer on those two signals, then roll the results up into one number that's anchored to your most important, buyer-intent questions.
Step 1 — Score each answer
For every question, on every platform, we combine the two signals:
- Mention rate is worth up to 70 points. It's how often you were named across the answers, as a percentage.
- Position bonus is worth up to 30 points. Ranking first earns roughly 27 of those points; the bonus fades to zero by about the tenth spot. No ranking data means no bonus.
So a brand named in every answer, right at the top, scores near 100. A brand named half the time, lower down, lands in the middle.
Mentions vs. citations — not all hits are equal
When we look at an AI answer, there are two very different ways your brand can show up, and they don't count the same:
| How you appear | What it means | Credit |
|---|---|---|
| Named in the answer | The assistant actually recommends or names your brand in its response | Full credit (1.0) |
| Cited as a source | Your brand wasn't named in the answer, but your own website turned up in the assistant's cited sources | Partial credit (0.2) |
| Not present | Neither named nor cited | No credit (0) |
A genuine recommendation in the answer is what actually influences a buyer, so it earns full credit. A citation-only hit means AI used your site as research but didn't recommend you to the reader — a real signal, but a much weaker one, so it earns about a fifth of the credit.
This is why a brand can have lots of citations and still post a modest score: being read by AI isn't the same as being recommended by it.
A fair denominator — what actually "counts against you"
Mention rate is your credit divided by the questions that genuinely counted. The subtlety is in that denominator — not every answer where you're absent is a fair loss:
| The answer was… | Counts as… |
|---|---|
| A real recommendation that named competitors but not you | A full miss — you lost a question you could have won |
| An answer that recommended no brands at all (generic advice, no vendors) | A partial miss (0.3) — there was no vendor slot to win, so it barely counts |
| No answer / the platform didn't respond | Ignored — it doesn't count either way |
The middle row matters: if an assistant answers "here's how to think about choosing a tool" without naming anyone, that's not really a loss to a competitor — so we down-weight it instead of penalizing you for the shape of the question.
Step 2 — Roll everything up to one number
Not every question carries the same weight. We tag each one with a reach level, from direct buyer-intent out to broad curiosity:
| Reach level | Roughly… |
|---|---|
| Core | "Best tools for X" — someone ready to choose |
| Adjacent | A neighboring need where you're a plausible answer |
| Aspirational | Broad industry and thought-leadership themes |
| Visionary | Top-of-funnel curiosity, loosest relevance |
Your headline score is the score of your core, buyer-intent questions. The other levels are still scanned and scored — you'll see them in the per-reach-level breakdown — but they don't move the headline number. That keeps the score honest about the thing that matters most: when someone is ready to buy in your category, does AI recommend you?
(If a brand has no core questions at all, the headline falls back to a blend across whatever levels did run, weighted toward the higher-intent ones.)
A worked example
Say a question set produces these four answers:
- Answer 1: names you, first in the list → full credit, top position
- Answer 2: cites your site but doesn't name you → partial credit (0.2)
- Answer 3: recommends two competitors, not you → a full miss
- Answer 4: gives generic advice, names no brands → a partial miss (0.3)
Your credit is 1.0 + 0.2 = 1.2. The questions that counted add up to 1 + 1 + 1 + 0.3 = 3.3. So your mention rate is 1.2 ÷ 3.3 ≈ 36%. Layer in the strong ranking on Answer 1, and the question score lands in the moderate range — held back mostly by the competitor loss in Answer 3.
What doesn't directly move the number
- Sentiment (whether a mention is positive, neutral, or negative) is captured for every mention and shown in your report, but it isn't baked into the 0–100 score. The score measures presence and prominence; sentiment is context you read alongside it.
- Direct-Brand Knowledge is a separate score that measures how much AI knows about you when asked by name. It's not part of AI Visibility — see the overview for the difference.
Keep going
- AI Visibility Score overview — what the score means and how a scan is built
- Recommendations — the prioritized actions to move your score