AI visibility is measured by whether ChatGPT, Google AI Overviews, Gemini, and Perplexity cite a business by name when someone asks a relevant question, and whether what they say is correct. Traffic and rankings can't tell you that. Seven specific metrics make up that picture, and none of them are "position."
Traditional SEO reporting was built for a search era where every result was a ranked, numbered link. AI answer engines don't work that way. They generate one answer and either mention a business or they don't, which means the old reporting toolkit doesn't measure the thing that matters anymore.
A business new to AI visibility work typically starts around a 10 to 20% citation rate on its core queries, with 50%+ considered strong. Here's what to track instead of a vanity traffic number.
Citation rate
The percentage of target queries where a business gets mentioned when someone asks an AI tool about it. This is the foundational metric, the one everything else sits on top of. If a business isn't showing up at all, nothing else on this list matters yet. New-to-GEO businesses typically start around 10-20%, moderate performers land 30-50%, and a strong position is 50% or higher.
Citation accuracy
Getting cited isn't automatically good news. This metric tracks whether the AI's citation is correct: right hours, right services, right location, right contact information. A citation with wrong details can cost a business the call just as easily as no citation at all, and it's a distinct failure mode that a simple "were we mentioned, yes or no" metric completely misses.
Share of AI voice (SOAIV)
How often a business gets cited relative to its named competitors on the exact same queries. A citation rate on its own doesn't tell a business whether it's winning or losing in its category, only that it shows up sometimes. Share of AI Voice answers the more useful question: out of everyone who could get mentioned for this query, how much of that attention is going to this business.
A citation rate tells you if you're in the room. Share of voice tells you who's actually being listened to.
Citation position
Whether a business is the primary source in an AI's answer or a secondary, afterthought mention. AI answers don't have "page one" the way search results used to, but they do have a structure, and being the first thing an AI mentions carries more weight than being buried in a list of five alternatives at the end.
AI referral traffic and traffic quality
Sessions that arrive from an AI platform, and, more importantly, what that traffic does once it lands: bounce rate, pages per session, whether it converts. AI referral traffic is often lower in raw volume than organic search traffic, since a citation frequently answers the question without requiring a click at all, but the traffic that does arrive tends to be higher-intent.
Content citability score
A composite score, built from five factors scored 0-10 each (specificity, structure, author or organization credentials, sourcing, and freshness), that predicts how likely a specific page is to get cited. This is the metric used to prioritize which existing content to fix first, rather than guessing.
Citation coverage
The percentage of a business's total published content that gets cited at all, across every platform tracked. This is the metric that flags a genuine content gap: a topic the business should own but hasn't published anything citable about yet.
Why this replaces "traffic went up"
Here's the direct comparison between what an old-style report shows and what each number misses.
| Old metric | What it tells you | What it misses |
|---|---|---|
| Page views | Someone visited a page | Whether they were the right person, in the right market, asking the right question |
| Keyword ranking | A position in Google's organic results | Whether AI tools mention the business at all, since AI answers don't map to ranked positions |
| "Engagement" | Time on page, scroll depth | Whether the business is winning or losing against named competitors for the same queries |
How do you measure AI visibility in each AI tool?
Ask each tool the same set of real buyer questions, then write down three things: whether the business was named, whether the details were correct, and where in the answer it showed up. Run each question more than once, because AI answers change from one ask to the next.
Do this separately for each tool. ChatGPT, Gemini, and Perplexity pull from different sources, so a business can show up in one and be missing from another. Averaging them into one number hides exactly the gap that needs fixing.
How do you track AI visibility metrics in Gemini?
Track the same core numbers in Gemini as anywhere else: citation rate, citation accuracy, and share of AI voice. The difference is where Gemini gets its details. For local questions, it pulls business information from Google Maps, so your Google Business Profile shapes most of what Gemini says about you.
Gemini is worth its own test. The Gemini app passed 1 billion monthly users in August 2026, and that count covers the app only, not AI Overviews or AI Mode inside Google Search. Showing up in one of those doesn't tell you whether you show up in the other, so check them separately.
Where Gemini gets local answers
Since December 2025, Gemini has shown local results as visual cards with photos, ratings, and business details pulled from Google Maps. That changes how I read citation accuracy in Gemini. If it lists the wrong hours or the wrong services, I check the Google Business Profile before I touch the website, because that profile is the data Gemini is showing.
How to run a Gemini check
- Write 10 questions a real customer would ask, in their words. Mix broad ones like "best [service] near [city]" with specific ones like "who does [service] on Saturdays in [neighborhood]."
- Name the city or neighborhood in the question, or run the check from the area you serve. Local answers depend on location.
- If you can, use a Google account that isn't tied to your business, so you see what a stranger sees.
- Ask each question three times, in a new chat each time. Write down whether you were named, whether the details were right, and whether you came first or showed up further down the list.
- Compare every detail Gemini gave against your Google Business Profile: hours, categories, services, phone number, address.
- Run the same questions once a month. One check is a snapshot. The trend across months is the number that matters.
What a Gemini gap usually points to
When a business isn't named at all, I look first at whether its Google Business Profile is complete and whether its categories match what people are actually asking for. When it's named with wrong details, the profile is usually out of date somewhere. When it's named but buried under competitors, I compare reviews and photos side by side, because that's what Gemini puts on the card.
The bottom line
None of these seven metrics replace the others. Citation rate tells a business if it's showing up. Accuracy tells it whether what's being said is true. Share of voice tells it how it stacks up. Position tells it how prominently. Referral traffic and citability score tell it where the content is and isn't working. Coverage tells it what's still missing entirely. Together, they're the actual scoreboard for AI visibility, built for how AI search works now instead of borrowed from how Google search used to work.
Last updated: October 2026.
Sources: AI Overview search prevalence (48% of Google queries, March 2026): The STACC / Advanced Web Ranking, Digital Applied. Gemini app passing 1 billion monthly users (August 11, 2026): iPhone in Canada, reporting Sundar Pichai's announcement. Gemini local results from Google Maps (December 13, 2025): PPC Land.