How to measure AI visibility: the ONE Score explained
AI visibility is measurable. Here is how to measure it properly across five engines, what goes into the ONE Score, and how to use it to drive real improvement.
Short answer: You measure AI visibility by asking the real engines the questions your buyers ask and recording whether your brand appears, how prominently and how positively, across every engine. ONEGO combines that into the ONE Score, a single number from 0 to 100 that tells you whether AI recommends you.
You cannot improve what you cannot see, and until recently AI visibility was almost impossible to see. The conversations are private, there is no public ranking, and a single spot check is misleading. This guide explains how to measure it properly, and how the ONE Score turns a messy problem into one number you can move.
Why can I not use normal SEO metrics?
You cannot use normal SEO metrics because AI visibility is about being named in an answer, not ranked in a list, and there is no public ranking for AI answers to scrape. Rankings and traffic tell you about search, not about whether an engine recommends you.
A page can rank first in Google and never be named by ChatGPT, and a brand can be recommended across engines without topping any keyword. The two are related but measure different things. To measure AI visibility you have to query the engines directly and read what they actually say, which is a different exercise from rank tracking.
What should I actually measure?
Measure four things for every buying question, on every engine: whether your brand appears, your position in the answer, the sentiment of how you are described, and which competitors are named. Together these tell you not just if you show up, but how well.
Each dimension matters on its own:
- Presence. The baseline. Are you named at all for the question.
- Position. Where you appear, since the first names in an answer carry the most weight.
- Sentiment. Whether the description helps or hurts you. A lukewarm mention underperforms a strong one.
- Competitors. Who is named instead of or alongside you, which shows the gap to close.
What is the ONE Score?
The ONE Score is a single measure of AI visibility from 0 to 100. It blends how often you appear, how prominently and how positively, across every engine and prompt, into one number your whole team can rally around.
The value of a single score is focus. Instead of drowning in per engine, per question detail, you get one target to move, and a clear way to show whether your work is working. Under the hood it is built from the four dimensions above, weighted so that frequent, prominent, positive mentions count for more than rare, buried, lukewarm ones. A low score in coral means AI rarely recommends you. A high score in green means you own the answer.
How do I turn the score into action?
Turn the score into action by using the breakdown behind it: the specific questions you lose, the competitors who win them, and the prioritised fixes ranked by expected impact. The number tells you where you stand; the breakdown tells you what to do.
A score with no next step is just anxiety. That is why measurement should come with a prioritised fix list, so you spend effort where it moves the score most. Work the top fixes, run your prompts again, and watch the number respond. This loop of measure, fix and measure again is the whole discipline of GEO in practice.
How often should I measure?
Measure on a schedule, weekly or daily, because engines change, competitors act, and a one off snapshot goes stale fast. Regular tracking shows the trend, proves your work, and catches slippage before it costs you.
A single audit is a great starting point, but AI answers move. Scheduled tracking turns measurement from a one time scare into an ongoing advantage: you see your score climb as fixes land, and you spot a competitor surging in time to respond. This is also how you prove the value of the work to a team or a client.
The bottom line
AI visibility is measurable, and measuring it is the first real step to improving it. Capture presence, position, sentiment and competitors across every engine, roll it into the ONE Score, and use the breakdown to act. Then track it, and climb.