Your Business Appears in a ChatGPT Result... and Then It Doesn’t! The Problem with Tracking AI Visibility
In short
AI answers are inconsistent: the same prompt, asked twice at the same time, can recommend different businesses, and ChatGPT and Gemini rarely cite the same ones. There is no Search Console for ChatGPT or other AI tools, and the paid AI visibility tools are flawed. Focus on trends rather than specifics: make your website answer questions, meet EEAT guidelines and build reviews and social proof.

Tracking SEO rankings and traffic over the years has always been relatively easy. Rank trackers helped you to understand where you were ranking for given search queries and tools such as Google’s GA4 analytics and Search Console helped you dig deeper into the metrics to understand what traffic you were getting, from where, to what pages and what was converting.
But things have changed…
As an increasing number of users turn to tools such as ChatGPT, Claude and Gemini to search, it’s becoming much harder to really understand how your business is performing in these tools.
What the research shows
A recent study by Steady Demand analysed 14,472 citations, comparing Gemini and ChatGPT for local search queries. Both platforms only cited the same businesses 8% of the time, and the top recommended business only 4.2%.
Gemini tended to recommend actual businesses, whilst ChatGPT tended to rely more on Reddit and business directories.
However, one of the most insightful findings was that Gemini only recommended the same top business 7% of the time. In other words, repeated searches changed the order of recommended businesses.
Same prompt, different answers
Here’s an example. I’ve done the same search twice, at the same time, from the same location but in two different incognito Chrome windows. For both I was logged out of ChatGPT, to avoid any history affecting the results.
I did the same prompt for both:
“recommend a nice restaurant in marbella with parking that’s open on Wednesday”
These are the two responses generated…
Response 1:

Response 2:

The top recommendation in Response 1 does not appear in Response 2, and vice versa! It seems that being a top recommendation is not always a good thing as the other restaurants appear in both responses.
The AI visibility challenge
And this is the heart of the challenge ahead of us all in trying to understand how our businesses appear in AI searches.
There is little consistency across platforms, the results change on each prompt and as models get updated (every few months) it probably means that the results will change again, sometimes quite dramatically.
And on top of that, we don’t all use the same models, some of us are logged in (getting personalised results) and we all prompt differently. And that’s without even considering that users prompt again with follow up prompts, so a business may not appear in the first prompt, but maybe in prompt 2, 3, 4…
How do you measure all of this?
Many companies are scrabbling to try and solve this problem, but I don’t think there is an actual good solution. I’ve tested most of the AI visibility tools out there and they are all quite flawed… and expensive.
The reality is that we have to focus on trends and not specifics. Ensure that websites are set up to provide answers and meet EEAT guidelines. User reviews and social proof are huge, these play a big factor too.
There is no Search Console for ChatGPT or Claude or other AI tools. It’s unlikely there will ever be, although maybe for paying advertisers.
We are losing the deep insights we used to have but this is all part of the new era of AI search.
Just be careful paying for AI visibility tools… they won’t give you a true interpretation of how your business truly appears to users.