Measuring AI Visibility: Share of Voice, Citations, and What to Track in the AI Era

Your analytics dashboard is lying to you — not on purpose, but because it was built for a world where a visit preceded every customer. In a zero-click world, a customer can find you, choose you, and book you without your analytics ever recording a single hit. The metric that actually predicts whether you get recommended is AI share of voice — and almost nobody is measuring it yet.

Why Old Dashboards Fail

For twenty years, traffic was the scoreboard. More visits meant more visibility meant more customers. That equation held when every customer clicked through. It breaks in a world where an AI answers the question and the customer never arrives at your site — or arrives already decided, after the model recommended you. A business can lose half its traffic and grow, or hold its traffic and shrink, because the visit is no longer the moment of decision. The decision moved upstream, into the answer.

That does not mean analytics are useless. It means they measure the past behavior of convinced customers, not the upstream moment where you won or lost the recommendation. To see that moment, you need a different instrument.

AI Share of Voice, Defined

Think of it the way you used to think of share of voice in paid media: of all the times a customer could hear about your category, how often is it you? In the AI era the “ad” is the model’s answer, and your share is how often you are in it. A business with high AI share of voice gets recommended constantly, often invisibly. A business with low share is simply not in the conversation, no matter how nice its website is.

How to Track AI Citations

There are two honest ways to measure this today, and you should probably use both.

Free manual prompt audits

Write a fixed list of 10–20 prompts a customer would actually ask about your category in your area (“who is the best tattoo shop in Denver for custom sleeves,” “find a salon near me that does balayage and has openings this week”). Run them, exactly as written, in ChatGPT, Perplexity, and Gemini once a quarter. Record whether you appear, whether competitors appear, and what the model says about you. It is manual, it is imperfect, and it is still the most reliable free signal available — because you are measuring what the customer actually sees.

Paid AI-visibility tools

A growing category of tools runs prompts at scale across engines and tracks mentions over time. They are useful for volume, but they are early, expensive, and sometimes noisy. Treat them as a supplement to manual audits, not a replacement — and never as a single number to optimize blindly.

What Search Console’s AI Segments Tell You

Google is rolling out AI segments in Search Console that surface AI Overview impressions and clicks — the first native signal of how often you appear inside a generated answer. They are incomplete and evolving, but they are free, they come from Google itself, and they are the bridge between your old ranking reports and your new share-of-voice scorecard.

Read them as a directional signal: are AI Overview impressions for your queries going up or down quarter over quarter? Pair them with manual prompt audits and you have a measurement stack that finally matches the era.

A Repeatable Quarterly AI Visibility Scorecard

The goal is a number you can defend and a trend you can act on. Once a quarter, run the same instrument and record:

• Your AI Visibility Score — from our free tool, run on the same URL, same quarter. The objective, machine-readable baseline.
• Manual prompt audit results — how often you appeared across your fixed prompt list, in each engine.
• Search Console AI segments — AI Overview impressions and clicks, quarter over quarter.
• Competitor share — who else appeared, and whether your share is growing or shrinking against them.

Measurement is the closing loop of this playbook. You cannot defend a position you do not measure, and GEO only compounds if you can see whether it is working. Re-run your score quarterly; track it like you tracked rankings.