SEO visibility is a simple idea with a messy set of numbers behind it. The idea: how often are you seen when people look for what you sell? The numbers: impressions, average position, a tool's visibility score, and now, for AI assistants, how often their answers name or cite you. Each measures something different, and teams get into trouble when they blend them into one figure or read one as if it were another.
This post separates them, shows how each is calculated, and shows how we read visibility for our own site.
What does SEO visibility mean?
In classic SEO, visibility means presence in Google's results for the queries that matter to you. A site is more visible when more of its pages appear, higher up, for more of those queries.
In AI search, visibility means presence in the answers. An assistant does not show ten ranked links; it writes an answer and sometimes attaches sources. You are visible when the answer names your brand, and separately when it links to your pages.
Both describe the same underlying goal, being found when a buyer looks, but they are measured in different units and they do not move together. A site can rank well and be absent from answers, or the reverse. Answers are assembled from passages that fit the question, and those passages often come from pages that never reach Google's first page, so each layer has to be read on its own terms.
How is Google visibility measured?
Impressions
Search Console counts an impression each time someone saw a link to your site on Google. It is the closest thing to a direct measure of being seen, because it comes from Google itself. Its limit: it only covers queries where you appeared at all, so it cannot tell you about the searches where you were missing.
Average position
Where your link appeared, averaged across the queries it appeared for. Useful per query, misleading as a site average, because one page ranking first and another ranking fortieth average to a number that describes neither.
Visibility scores from SEO tools
Most SEO tools offer a visibility score or index. The usual method: take a set of keywords, look up where you rank for each, and multiply by an expected click-through rate for that position. Rank first for a keyword and you get most of its possible clicks; rank tenth and you get a fraction. Those expected rates are themselves moving: Ahrefs found the first result's average desktop CTR on informational keywords fell from 7.6% in December 2023 to 3.9% in December 2025, and to 1.6% where an AI Overview appeared (300,000 keywords, published February 2026). A score built on older weights overstates the clicks a ranking now earns.
That makes a visibility score a model of potential clicks on a keyword set the tool chose, or you chose. Change the keyword set and the score changes without anything happening in Google. Compare scores only within one tool, on one keyword set, against direct competitors, and read the trend more than the level.
How is AI visibility measured?
Here the unit is an answer. You put a fixed set of buyer questions to each assistant, repeatedly, and read every answer for two separate things.
Share of voice: are you named?
The share of answers that name your brand. At Attensira it is a rate: answers naming you divided by all successful answers in the window. Competitors are measured against the same denominator, so the rates do not add up to a whole: several brands can be named in one answer, or none.
Citation rate: are you linked?
The share of answers that attach at least one link to your own domain. A model can name you from memory without linking, or link one of your pages without naming you, so the two are counted independently.
Sample size
Every AI visibility number needs the number of answers behind it. Answers vary from run to run, and a share built on ten answers can swing widely by chance. A rate without its sample size is not a measurement.

