Short answer: AI answers aren't a ranking you can look up. The same question can return a different list of brands each time, so one screenshot proves nothing. Measure by repeated samples over time, log the results, and combine them with what your own analytics can actually see.

Why a single check misleads

SparkToro's Rand Fishkin and Gumshoe.ai's Patrick O'Donnell studied how consistent AI brand recommendations are. As reported by Search Engine Journal in January 2026, they ran 12 brand-recommendation prompts many times through ChatGPT, Claude and Google's AI in Search (AI Overviews, with AI Mode used when an Overview didn't appear), collecting 2,961 prompts across November and December.

The headline finding: ChatGPT and Google's AI in Search returned the same brand list less than 1% of the time, and the same list in the same order less than 0.1% of the time. In Fishkin's words: "If you ask an AI tool for brand/product recommendations a hundred times nearly every response will be unique."

Two cautions matter. The report is "not peer-reviewed academic research," and Gumshoe.ai sells AI tracking tools, a conflict Fishkin disclosed. Treat it as a strong signal, not settled science. It also found some brands recurred across many responses, and that visibility across many prompts was more stable than ranking position.

What this means for you: "Are we number one in ChatGPT?" is the wrong question. "How often do we appear across many realistic questions, and is that trending up?" is the right one.

A method you can run yourself

Step 1: Build a question set.

Write 15 to 25 questions your real customers would ask, in their words, not yours. Mix types: "best [service] in [city]", "how do I choose a [service]", "[problem] fix", and a few using your name. Keep the set fixed so you can compare month to month.

Step 2: Sample repeatedly.

Run each question several times on each tool you care about, on different days. Use logged-out or clean sessions where you can, because personalisation changes results.

Step 3: Log what you see.

For each run, record: tool, date, question, whether you were mentioned, whether your site was linked, which competitors appeared, and which sources were cited. A spreadsheet is enough.

Step 4: Report rates, not anecdotes.

"Mentioned in 6 of 40 runs (15%)" is a measurement. "We showed up" is not. Track the rate and the cited sources by month.

Step 5: Read the sources.

The pages AI answers cite tell you where to invest: your own pages, directory listings, review sites, articles that mention you.

What your own analytics can and can't show

  • Google's AI features: Google says traffic from AI Overviews and AI Mode is included in overall Search traffic and reported under the "Web" search type in Search Console's Performance report. It isn't broken out separately.
  • Other AI tools: referrals from some assistants may appear as a source in your analytics, but practices vary and aren't uniformly documented. Check your own source and medium reports for the assistants you care about, and treat any count as a floor, because some visits arrive without a referrer.
  • Brand searches and direct visits: a rise in people searching your name or arriving directly can be a sign of exposure you can't attribute precisely.

How to read a vendor's AI visibility report

Ask before you pay:

  • How many prompts, and how were they chosen?
  • How many runs per prompt, and over what period?
  • Are results from the live product or an API, and does that differ?
  • Do they show variation, or one score?
  • Can you see the raw responses?

Fishkin's own advice was to "make sure your provider answers the questions we've surfaced here and shows their math." If a tool gives you a single score without any of this, treat the score with suspicion.

What this won't tell you

Even a careful sample can't tell you whether a mention produced a customer, and AI results vary by person, location and day. Use visibility rates as one input beside enquiries, search traffic and sales, not as the goal itself. For what Google itself requires, see What Google Actually Requires to Appear in AI Overviews and AI Mode; for the crawler decision, Should You Let AI Crawlers Onto Your Website?; and for the numbers to pair this with, Five Numbers That Tell You Whether Your Marketing Is Working. For how AI search relates to ordinary SEO, see What Is Answer Engine Optimization? and Why Your Business Needs to Show Up in AI Search.

Common questions

Can I track my "ranking" in ChatGPT?

Not meaningfully. The study above found brand lists and their order almost never repeated, so a single rank isn't stable.

How many runs are enough?

Nobody has established that. Fishkin lists it as an open question. Start with several runs per question and watch for stability.

Will Search Console show AI Overview clicks separately?

No. Google says they're included in the "Web" search type.

Is paying for an AI visibility tool worth it?

Only if it can show its method and variation. A spreadsheet and an hour a month can answer the basic question for most small businesses.

Velocity Branding builds SEO and AI search visibility for small businesses and startups across the US and South Africa.

Want a baseline of where your business appears in AI answers today? Run our free AI visibility check or book a free strategy call.