AI Visibility Tracking: The 6 Metrics That Matter

Most teams that start AI visibility tracking check one thing: did ChatGPT mention us. They run a few prompts, see the brand name show up, and call it tracked. Then a model update lands and the number they never wrote down moves by half, and nobody notices for a month.

Being mentioned is the shallowest of six metrics. It tells you that you exist in the answer. It tells you nothing about whether you won it, whether the engine trusts your domain, or which competitor the model put next to your name.

This guide covers what AI visibility tracking actually measures, why it is a different job from tracking Google rankings, the six metrics worth a weekly look, and how to start without buying anything.

What is AI visibility tracking?

AI visibility tracking is the practice of measuring how often, and how well, AI answer engines like ChatGPT, Gemini, Perplexity, and Google AI Overviews mention and cite your brand. It runs your buyer prompts across each engine on a fixed cadence, scores where you appear and who appears beside you, and flags when that position moves.

The reason it needs its own discipline is that the answer changes constantly, and it does not change in step with Google. A rank tracker watches a results page that updates on a known rhythm. AI visibility tracking watches a generated answer that can rewrite itself the day a model ships.

Why tracking AI visibility is not the same as tracking Google rankings

The instinct is to assume your SEO dashboard already covers this. If you rank well, surely the AI cites you. The data says otherwise, and it now says it from two independent sources.

Profound launched the Profound Index in June 2026, built on more than 1.5 billion real-user prompts across 50-plus industries. Its headline finding: only about 19% of ChatGPT's answers overlap with Google's top results. Separately, GEO firm Brandlight found the overlap between top Google links and AI-cited sources has fallen from roughly 70% to under 20%. Two unrelated vendors, the same number.

Winning Google and winning AI are now two different jobs. Roughly four out of five AI answers do not match the page you optimized for.

Tracking Google rankings tells you:

• •Where your page sits in the SERP this week.
• •How many keywords moved up or down.
• •Whether your backlinks are growing.

Tracking AI visibility tells you:

• •Which buyer prompts name you, and which name a competitor.
• •Whether the engine cites your domain or just paraphrases you.
• •Which rival brands the model groups you with.

The two dashboards answer different questions. A rising rank chart can sit next to a falling citation share, and most teams only own the first chart.

The 6 metrics that matter in AI visibility tracking

A single visibility number hides more than it shows. The Profound Index moved past one score to six diagnostic metrics, and that breakdown is the most useful framework published this year for deciding what to actually watch. Here they are, from the shallowest signal to the one that maps your competitive set.

Metric #1: Visibility score tells you how often AI mentions you at all

This is the base rate: across your tracked prompts and engines, how often does your brand appear in the answer in any form. It is the first thing to measure and the easiest to over-trust. A high visibility score with nothing underneath it means you are named but not necessarily chosen or cited. Treat it as the floor, not the goal.

Metric #2: Share of voice tells you how often AI picks you over rivals

Share of voice is your visibility as a percentage of the whole category: of every answer in your space, how many feature you versus your competitors. It is the AI equivalent of market share, and it is the number an executive team understands without a translation layer. Our own first-party AI search data, drawn from more than 34,000 AI answers, shows the top brand in a category averages 76% share of voice. The gap between first and second is wide, and it is the gap worth closing.

Metric #3: Mention position tells you whether you lead the answer or trail it

Where you land inside the response matters as much as whether you land at all. A brand named first in a recommendation gets read; a brand buried in position seven rarely does. Mention position separates the answers where you are the lead recommendation from the ones where you are filler the model added to look thorough. Two brands can share an identical visibility score and have completely different mention positions.

Metric #4: Citation share tells you whether AI links your domain as a source

Mentioned and cited are not the same thing. Semrush found that on Gemini, the overlap between a brand being mentioned and being cited can be as low as 30%. You can be named in the prose with no link back to you. Citation share measures how often the engine actually sources your domain, which is the closest read you get on whether the model treats you as an authority rather than a passing reference.

Metric #5: Co-citation share tells you which domains AI trusts beside you

When an engine cites you, it usually cites others in the same breath. Co-citation share is the set of domains that show up alongside yours: the Reddit threads, review platforms, and publications the model pulls from to build the answer. This is a map of the source pool you are competing inside. If the same three domains appear next to your category every week and you are not on any of them, you have just found your off-page target list.

Metric #6: Co-mention share tells you the competitive set AI built for you

The model decides which brands belong in a comparison, and that set is not always the one your sales team uses. Co-mention share shows which competitors get named with you most often. Sometimes it surfaces a rival you do not track. Sometimes it reveals the model has filed you next to the wrong tier of the market entirely, which is a positioning problem you can only fix once you can see it.

How to start AI visibility tracking in 4 steps

You do not need a platform to begin. You need a prompt set, a spreadsheet, and a standing hour each week. Here is the loop that gets you a real baseline.

Step 1: Pick the 20 buyer prompts that decide your deals

List the questions a buyer types before they shortlist: "best [category] tool for [use case]," "alternatives to [competitor]," "is [your brand] good for [job]." Twenty is enough to start. These are the prompts where a citation changes a deal, so they are the only ones worth tracking first.

Step 2: Run them across every engine your buyers use

ChatGPT, Gemini, Perplexity, Google AI Overviews, and Copilot pull from different sources and weight freshness differently. Run the same prompts through each and log the raw answers. An agency that reports a single "AI visibility" number is flattening five answers that disagree with each other.

Step 3: Score every answer on all six metrics

For each answer, record more than a yes or no. Note whether you were named, where you landed, whether your domain was cited, which other domains were cited, and which competitors appeared. The extra columns are the difference between knowing you exist and knowing whether you are winning.

Step 4: Re-run weekly and watch the drift

The first run is a snapshot. The value is in the second, third, and tenth. Run the same set weekly, and the movement tells you what a single number never could. Our first-party data shows the category leader changes in 24% of weekly editions. One week in four, the brand on top is no longer on top.

Manual tracking vs a tool vs a managed service

Once you have a baseline, the question is who runs the loop. There are three honest options, and the right one depends on how many prompts and engines you watch and whether anyone has the standing time.

The tools are worth knowing before you buy. We break down how they work and when you need one in our guide to AI rank trackers, and we cover the broader practice in AI brand monitoring. When the loop is too relentless to run on the side, a managed GEO agency can own the measurement, the passage work, and the weekly tracking together.

This is where the volatility argument earns its keep. GPT-5.5 became ChatGPT's default model in 2026 and, by Profound's measurement, moved major brands by 35% to 56% in some categories on the update alone. A single model update can rewrite your share of AI answers in a week, with no changelog you would ever see. That is the case for weekly cadence, whoever runs it.

FAQ

What is AI visibility tracking?

AI visibility tracking measures how often and how well AI answer engines like ChatGPT, Gemini, and Perplexity mention and cite your brand. It runs a fixed set of buyer prompts across each engine on a regular cadence, scores where you appear and who appears beside you, and flags when that position changes. It is the AI-search equivalent of rank tracking, built for generated answers instead of a results page.

What metrics should you track for AI visibility?

Track six: visibility score (how often you appear), share of voice (your slice of category visibility), mention position (where you land in the answer), citation share (whether your domain is sourced), co-citation share (which domains are cited beside you), and co-mention share (which competitors are named with you). Most teams track only the first and stop, which hides whether they are actually winning answers.

How do you track AI visibility without a tool?

Pick 20 buyer prompts, run them across ChatGPT, Gemini, Perplexity, AI Overviews, and Copilot, and log six things per answer in a spreadsheet: whether you were named, your position, whether your domain was cited, the other cited domains, and the competitors mentioned. Re-run the same set weekly. The drift between runs is the signal a single check never gives you.

How often should you track AI visibility?

Weekly is the practical floor. AI answers move faster than Google results because a model update can rewrite them overnight, and our first-party data shows the category leader changes in 24% of weekly editions. Monthly tracking misses the swings that matter; quarterly tracking means the drift outruns your ability to respond.

Is AI visibility tracking the same as an AI rank tracker?

An AI rank tracker is one way to do AI visibility tracking, not the whole of it. The tool automates the scoring across engines. AI visibility tracking is the broader practice, which also covers choosing the right prompts, reading the source pool, and acting on what moves. The tool gives you the numbers; the practice decides what to do with them.

The bottom line

You cannot defend a position you never measured. AI visibility tracking exists because the answer your buyers read is generated fresh, changes weekly, and overlaps with your Google rankings less than one time in five.

Start with the 20 prompts that decide your deals, score all six metrics instead of just the mention, and re-run the set every week. Once you can name your share of voice and your citation share this week, you are finally measuring the game your buyers are actually playing. If you would rather not run the loop by hand, an AI visibility audit gives you the baseline and tells you whether the next move is a tool or a team.