Measuring AI Visibility: The Metrics That Actually Matter

One of the most common things I hear is that AI search feels like a black box — you either get recommended or you don't, and there's no way to know why. That's not true anymore. AI visibility is measurable, and once you measure it, it stops being mysterious and becomes something you can steadily improve. The trick is knowing which metrics actually mean something.

Vanity numbers will waste your time. Let me walk you through the metrics that answer real questions — the ones that tell you whether you're winning, and what to do next.

Recommendation rate: are you being named?

This is the foundational metric, and it's exactly what it sounds like. Take the questions your customers actually ask the AI engines — "best [your service] in [your town]," "who should I hire for X near me" — pose them, and track how often you're named in the answer.

Do it across the surfaces that matter and across a real range of the questions your customers use, not just your favorite phrasing. A single check is a snapshot; a consistent set of questions asked repeatedly over time is a signal. Your recommendation rate is the clearest read on whether the machines currently consider you an answer at all.

Share of voice: how do you stack up against rivals?

Being named is good. Being named more than your competitors is the real position. Share of voice measures how much of the AI conversation you own versus the other businesses that could be recommended for the same questions.

This matters because AI answers are short — only a few businesses make the cut. If a competitor is named in eight of ten answers and you're in two, you know where you stand and how much ground there is to take. Share of voice turns a vague sense of "we should show up more" into a number you can move.

Being mentioned tells you the door is open. Share of voice tells you how much of the room you actually own.

— Bodhi

Sentiment: how are you being described?

Getting named isn't the whole story — how you're named matters just as much. Sentiment tracks the character of what the AI says about you. Are you described as the trusted, established choice, or mentioned in passing with a caveat? Is the language warm and confident, or thin and hedged?

Sentiment is a mirror held up to your reputation as the machine understands it. If the description is lukewarm, that usually traces back to weak corroboration or unclear positioning in the wider web. Improve the underlying reality and the sentiment in the answers tends to follow.

Citation sources: where do the answers come from?

This is the metric I'd fight to keep if I could only have one, because it's the most actionable. When an AI answer references sources, those citations tell you which pages the model trusts enough to build its answer from. That's a direct map of where your authority lives — and where it doesn't.

• See your own pages cited? Those are working. Understand why, and build more like them.
• See competitors' sources cited but not yours? That's a precise gap — you know exactly what to close.
• See third-party sites cited — directories, review platforms, publications? Those are the places worth earning a stronger presence.
• See the same few sources over and over? Those are the trusted watering holes in your market, and being present there pays off across every answer.

Recommendation rate tells you whether you're winning. Citation sources tell you what to actually do about it. That's why I treat them as a pair.

Track the trend, not the snapshot

Any single measurement is a moment in time, and moments lie. AI engines re-read the web constantly, so the number that matters is the direction of travel. Is your recommendation rate climbing? Is your share of voice growing against a specific competitor? Is sentiment warming as your corroboration deepens?

A rising trend line is proof your work is compounding. A flat one tells you to change approach before you spend another quarter on it. Either way, you're operating on evidence instead of hope — which is the entire point of measuring.

Turn measurement into momentum

Here's how the pieces fit together in practice. Recommendation rate and share of voice tell you where you stand. Sentiment tells you how you're regarded. Citation sources tell you where to invest next. Watched over time, together they turn AI visibility from a black box into a dashboard you can steer by.

You don't need perfect instrumentation to begin — you need an honest baseline. Ask the engines your customers' questions today, write down what you find, and you'll have the first data point in a picture that gets clearer every month. Measure, improve, measure again. That loop is how a business stops guessing and starts becoming the answer on purpose.