Read your AI Visibility score
What does my AI Visibility score actually measure?
AI Visibility is a rank score. Synup asks ChatGPT, Gemini and Perplexity a set of discovery questions about your category and city, then scores each answer on whether you were named and how near the top. Nothing else feeds it — not your rating, not your review count, not your directory listings.
That is the single most useful thing to know about this page, because the Visibility factors panel sitting right underneath the score is measuring completely different things.
What the Overview is showing you
Three boxes, and they are the whole page in miniature: the metrics row, the score with its label, and how that score breaks down by engine. Everything further down is evidence for those numbers.
Why an 81 can sit next to a Reviews score of 0
Here is that same agency's Visibility factors panel, sitting a few centimetres below its Excellent gauge.
Only one of those five is inside the score.
| Factor, and what it measures | Feeds your AI Visibility score? |
|---|---|
| Relevance — how visible you are across the discovery questions | Yes — this is essentially the score itself |
| Reviews — your rating and review-volume signal | No |
| Engagement — the share of answers that mention you | No |
| Freshness — how recently the listing was updated | No |
| Citations — directory citation coverage | No |
A score of 81 alongside Reviews 0 and Citations 12 is not a bug and not a contradiction. The assistants are recommending this agency anyway — they are leaning on property blogs and agent directories rather than on Google Business or Yelp, neither of which cites it.
Read the factors as reasons your position could change, not as inputs to today's number. Citations at 12 is why a competitor with a stronger directory footprint could overtake you next month; it is not why your score is 81 today.
The five numbers across the top
The KPI row is the fastest read on the page, and each number answers a different question.
| Metric | On this location | What it tells you |
|---|---|---|
| AI Visibility | 81/100 | How often and how highly you're named |
| Mention rate | 94% | The share of answers you appear in at all |
| Avg position | 2.5 | Where you land when you appear — lower is better |
| Prompts covered | 6/6 | Distinct questions where at least one engine named you |
| Sources cited | 98 | Distinct sources the engines cited in these answers |
Mention rate and Avg position are the pair worth watching together. A high mention rate with a poor average position means you are always in the conversation and never the recommendation. This location has the opposite problem to solve — it is named in 94% of answers at an average position of 2.5, so the remaining work is the one prompt it misses entirely, not its standing.
Sources cited is not a count of sources citing you. It counts every distinct source the engines cited while answering — which is why a location with no visibility at all can still show a healthy number.
This Fairbanks bed & breakfast scores 0/100 — Low, is named in no answer on any engine, and still reports 36 sources cited. The assistants read 36 things while answering questions about bed & breakfasts in Fairbanks. None of them was about this one. Its chips say the rest: Absent on 6 prompts, #7 of 7.
The bands, and what they're called
| Score | Label the product shows |
|---|---|
| 81–100 | Excellent — "Strong visibility — well positioned for AI-driven discovery." |
| 61–80 | Good — "Good visibility with room for improvement." |
| 41–60 | Average |
| 21–40 | Below average |
| 0–20 | Low — "Limited visibility — requires significant improvement." |
The boundaries are tighter than they look. A Texas auto-repair shop scoring 79 reads Good; the Miami agency at 81 reads Excellent. Two points, different words. Don't let the label do more work than the number.
Sydekick's read, and when to trust it
Above the page title you'll see a Sydekick's read bar, or Write one if no read exists yet. It generates a one-sentence summary of what it sees.
Treat it as a prompt to look, not as a finding. On the same 81-scoring location above, it produced "…misses key Miami prompts while competitors dominate search results." That sentence is anchored on one genuinely absent prompt, but it describes a location that is named in 94% of answers at position 2.5 as though it were losing. The numbers on the page are the authority.
Where to go next
- How do I read the prompt coverage grid? — the grid behind the score, one prompt and one engine at a time
- Where does this report come from? — why there's no Generate button, and when the numbers change