The Metrics That Define AI Search Visibility (And How to Track Them)

"AI Visibility" Is Vague Until You Measure It

"AI visibility" sounds like a feeling, not a number, which makes it easy to nod along with and hard to actually act on. It stops being vague the moment you define it as a handful of trackable metrics — the same way "SEO" stopped being abstract once businesses started tracking rankings, impressions, and click-through rate instead of just hoping.

Here are the metrics that actually define it, and how each one gets measured in practice.

1. Mention Rate (Share of Voice)

The core metric. Out of a fixed set of realistic prompts a customer might actually type — "best plumber in [city]," "who should I call for emergency AC repair near me" — what percentage of the time does an AI assistant name your business at all?

This is measured by running the same prompt set on a schedule (weekly is enough to catch real movement) across ChatGPT, Perplexity, and Google's AI Overviews, and logging whether you appear. A business mentioned in 6 of 10 relevant prompts has a mention rate of 60%. Most businesses that have never checked this are surprised by how low their number actually is.

2. Position and Framing

Being named isn't the only variable — how you're named matters almost as much. Are you the first business mentioned or the third? Does the assistant describe you accurately ("family-owned HVAC company serving [region] since 2010") or vaguely ("a local option")? A specific, accurate description reads as the assistant having real confidence in the facts about you. A vague one suggests it's working with thin or unclear information.

3. Source Citation Rate

When an AI assistant does name you, does it cite your website as a source, or is it pulling from a directory listing, a review aggregator, or a competitor's comparison page that happens to mention you? Being the cited source matters because it means the assistant trusts your own site as the authority on your business, not a secondhand summary of it — and it's a strong signal for whether your content is actually answer-shaped enough to quote directly.

4. Crawler Access Rate

A technical, binary-per-page metric: can the AI crawlers that matter (GPTBot, PerplexityBot, Google-Extended, and similar) actually reach and read each important page on your site? This is checkable directly against `robots.txt` and server logs, and it's worth auditing per page, not just once at the domain level — it's common for a site to allow crawler access generally but accidentally block it on the pages that matter most (service pages behind a slow JavaScript framework, for instance).

5. Structured Data Coverage

What percentage of your key pages carry correct, complete schema markup — `LocalBusiness`, `Service`, `Review` — versus how many are relying on an AI system to infer that information from unstructured prose? This is a straightforward audit: list your important pages, check each one against Google's structured data testing tools, and track the percentage that pass clean.

6. Consistency Score

How many of your core facts (name, address, phone, hours, services offered) match exactly across your website, your Google Business Profile, and your top directory listings? Score it as a simple ratio: matching fields over total fields checked, across your top five to ten listed sources. This one compounds — a single mismatched phone number can be the reason an AI system treats every other claim about your business with more caution.

7. Competitive Share

All of the above matter more in context. If you're mentioned in 40% of relevant prompts but your closest competitor is mentioned in 90%, the raw number was never the real problem — the gap was. Running the same prompt set against two or three actual competitors alongside your own business turns a flat percentage into an actual read on where you stand in your market.

How to Actually Track These Without Expensive Tooling

You don't need enterprise software to start. A spreadsheet, a fixed list of 10 to 15 realistic prompts, and a recurring calendar reminder to run them covers metrics 1 through 3 by hand. Metrics 4 through 6 are one-time technical checks that only need re-running after a site change. How to check if ChatGPT recommends your business walks through the exact prompt-running process for the manual version of this.

The tradeoff with doing it manually is consistency — it's easy to check once, get discouraged or distracted, and never check again, which defeats the point of a metric that's supposed to move over time.

Where This Fits

This is exactly what the optional Monitor plan automates: weekly AI-visibility checks across these metrics, a monthly re-grade, and an alert when something slips, so the tracking happens whether or not anyone remembers to run it manually. It's monitoring only, not a requirement to buy anything else.

If you haven't measured any of this yet, the free instant check scores your current AI visibility as part of a full grade, in about the time it takes to read this post.