“Are we number one in ChatGPT?” is the wrong question. Here is what to ask instead — and what local businesses should actually measure.
A lot of local business owners are starting to ask the same question: when someone asks ChatGPT, Gemini, or Claude for a recommendation in my category, does my business come up?
That is a fair question. The honest answer is complicated.
The old way of measuring search performance, “what is our rank,” does not work the same way in AI search. The numbers move too fast. They change by user. They change by prompt. They change by model update. Anyone selling you a “you are number one in ChatGPT” report is selling you a measurement that breaks the moment a new model ships.
If you are a local business trying to figure out what to actually track, here is what is going on and what to do about it.
For most of the history of search, rank tracking was the only game in town.
You picked your target keywords, you watched where your site showed up on Google, and you worked to climb. The number was imperfect, but it was useful. Most of the time, the same keyword produced roughly the same result for the same person in the same place.
We have already written about why that mental model is breaking down on Google itself. The short version: Google’s search results are getting squeezed by AI Overviews, ads, and other answer surfaces, and your ranking position matters less than you think.
AI search takes the same problem further. It does not work the way rank tracking assumes.
When a customer asks an AI for a recommendation, the answer is generated on the spot. It depends on:
That last point matters most. AI systems do not “rank” businesses the way Google ranks pages. They decide who to mention, who to skip, and who to recommend based on whether they can stand behind the claim. Position is not a number. It is a judgment.
A few weeks ago, every AI visibility tool on the market showed the same thing: a sudden drop in citations.
Customers panicked. Agencies panicked. The drop was not real. ChatGPT had quietly stopped showing as many citation links in its HTML response. The tool that scrapes ChatGPT to count your brand mentions was suddenly blind. The brand’s actual visibility had not changed. The reporting had.
This is not a one-time event. It is the new normal.
The same kind of thing happens every time a model changes. Every time a lab ships a new version. Every time a major source updates its content. The output of the AI is volatile by design, and traditional rank tracking, which assumes a stable result you can count, cannot keep up.
Even on a quiet week, the data is noisy. One tool will say your business is cited in 4 prompts. Another will say 200. Both can be correct for what they measured. Neither tells you the full story.
If “rank” is the wrong frame, the right question is: how stable and how well-represented is my business across AI answers, over time, across a range of real customer questions?
Two measurements matter most.
Stability. Is your business mentioned roughly the same way, by the same kinds of sources, for the same kinds of prompts, week after week? Big swings are a red flag. Steady, consistent presence is the goal.
Representation. When AI answers do mention your business, do they get you right? Do they spell the name correctly? Do they describe what you actually do? Do they put you in the right category, in the right location, for the right customer? A single mention that is wrong can do more harm than ten right ones.
The goal is not to be at the top. The goal is to be cited accurately, consistently, and in context — across ChatGPT, Gemini, Claude, and Perplexity — for the questions your real customers actually ask.
That is a different kind of work than chasing rank. It is closer to reputation management than to SEO. It is closer to making sure your business is unambiguous to a machine than to gaming a leaderboard.
If you run a plumbing company, a dental office, a med spa, a law firm, or any other local service business, the practical advice is straightforward.
This is slower than the “rank in 30 days” pitch. It is also more durable. The model update that wipes out a position-based report does not wipe out a consistent, well-grounded presence. The court ruling that puts pressure on AI to verify what it says rewards the businesses that are easiest to verify.
The right metric is not “where do I rank.” The right metric is “how clearly and consistently does AI understand my business, and how often does it get me right.”
That is a different game. It rewards different work. It is the work Proof Signal is built to do.
If you want to see what an AI audit actually looks like for a local business, the free score at https://proofsignal.biz/score runs the same kind of measurement across ChatGPT, Gemini, Claude, and Perplexity in about two minutes. If you want a deeper audit, the strategy call is 15 minutes, no pressure, and we will tell you honestly whether we can help.
Start with the Free Score to see what AI systems are actually saying about your business, where the gaps are, and what to fix first.