What is appearance rate, and what is the AI Visibility Score?
Appearance rate is the share of checks in which an AI assistant names your business. Apex publishes that same number as your AI Visibility Score: one number under two names, always shown with the counts behind it.
- Check
- One tracked question, asked of one engine, at one scheduled time.
- Named
- The answer contains your business by name. Not your category, and not a link to a directory that happens to list you. Your name, in the answer text.
- Appearance rate
- How many of the checks named you, out of how many were run, over a stated window, published with both counts and never as a bare percentage.
- AI Visibility Score
- The name Apex publishes appearance rate under. The same counted ratio, the same counts beside it. If you meet "appearance rate" inside the product and "AI Visibility Score" on a report, you are looking at one number under two names.
- Window
- The period the counts cover. An appearance rate without a window is not a measurement.
Two names, one number
Apex says "AI Visibility Score" where a business owner is being introduced to the idea, because "appearance rate" tells nobody what it is for. The product, the exports and the underlying data say "appearance rate", because that is what the field is called. Nothing is converted, weighted or blended in between.
The rule that keeps the rename honest: the Score is never published without its counts. A score out of a hundred would be a different kind of object, one you cannot audit, and Apex does not have one.
Why the denominator is the whole argument
Two vendors can report the same percentage and mean entirely different things by it. One asked a broad question set, held constant, across every engine it measures. The other asked a handful of questions once, on the engine its client happens to do best on. The percentages match and the claims are not remotely comparable. The denominator is where the honesty lives, which is why it never gets dropped for a cleaner-looking slide.
What Apex publishes
- The Score with its counts beside it, never a bare percentage on a slide.
- A separate figure for every engine we measure (ChatGPT · Gemini · Perplexity · Claude), because an overall rate can hide a zero.
- The questions the number came from, so you can argue with the instrument and not only with the result.
- The businesses named instead of you, which is reliably the most actionable line on the page.
Reading it per engine
One engine can be naming you regularly while another has never named you once, and a single overall figure averages that away. The overall number is true; the zero is the useful fact. That is why the per-engine split sits beside the headline rather than behind a menu.
It is also the case for measuring more than the one engine everybody quotes. A business can be well known to one assistant and structurally absent from another, and no average will tell you which.
What appearance rate does not tell you
- Whether you were described well. Being named alongside "though they are on the pricier side" is a mention and a problem at the same time.
- Whether you were named first or fifth. Order carries weight that the count does not capture.
- How crowded the answer was, and how much of it belonged to everyone else named. That is share of voice, and it needs its own evidence window.
- Whether the person asking was in your service area. The question set controls for that; the metric does not.
Questions people ask
What counts as being named?
Your business name appearing in the answer text. A link to a directory page that happens to list you does not count, and neither does the engine describing your category without naming anyone.
Is a higher appearance rate always better?
Almost always, but read the per-engine split first. Rising on the engine you already win while sitting at zero on another is worth less than the overall number suggests.
Is the AI Visibility Score a score out of a hundred?
No. It is a counted ratio wearing a clearer name (checks that named you over checks run), and it is always published with both numbers. A blended score out of a hundred cannot be audited; counted checks can, because you see the numerator, the denominator and the window and are free to disagree with any of the three.
How often does it update?
On a schedule you do not have to run yourself, so the count moves without anyone asking it to. A single movement is usually noise rather than news; the trend is the part worth reading.