Why Doesn’t ChatGPT Recommend My Company?
There is usually more than one reason. ChatGPT may not know your business, or may know it but fail to match your products or services to the customer’s question. Or it favours competitors whose information answers the same question more clearly.
What Does ChatGPT Base A Recommendation On?
ChatGPT draws on two sources. The first is the large body of text it was trained on; the model recognises brands that appear often and consistently there. The second is live search. OpenAI added web search to ChatGPT in the last quarter of 2024; when search is on, the pages the answer relied on are shown as small links.
For a brand to be recommended, the information about it has to be present in those sources, consistent with itself, and a direct match for the customer’s question. If any of the three is missing, the model picks another brand, because it tries to give an answer it is confident in.
How An AI Answer Picks Its Sources
It May Not Know You
If a small or new business is rarely mentioned online, it takes up little room in the training data too. The model may behave as if it has never heard of you. Live search is then your only chance, and it works only as far as it can reach your site and the pages written about you.
Sometimes the model knows the business but knows it wrongly: an old address, a discontinued product line, or another firm with a similar name. Mix-ups like these can mean your name never appears in the answer, or appears in the wrong sentence.
Your Site Does Not Answer The Question
A customer asks something detailed, such as “a sateen bedding brand that ships abroad”. The model looks for sources that answer those details plainly. If your site has the products but no shipping information, fabric details or price range, you do not match that question.
This does not mean your product is bad. It means the information is not written the way a person would ask for it. In a sales call those details get asked and answered; an assistant makes do with what it finds.
Your Site May Be Unreachable
In live search, ChatGPT reads pages with its own crawlers. If your site shuts those crawlers out without you realising, you do not appear in current answers. Some infrastructure and security services can block these crawlers by default; the site looks open to visitors but is closed to the assistants.
The way to find out is not guesswork but looking at your server records. They show which crawler comes how often and which pages it reads.
Competitors May Speak More Plainly
Picture three brands that could answer the same question. One has written out its products, sizes, shipping terms and customer reviews; the others have only listed product names. The assistant naturally leans on the first.
This is less a race of quality than a race of clarity. Looking at which sources the competitors named in the answer are cited from often shows the difference plainly.
How You Are Described In The Answer
Your name appearing may not be enough on its own. If the assistant mentions you as “a more expensive option” or “a brand known for its older models”, that sentence shapes the customer’s decision. What the answer says about you matters as much as whether your name is in it.
So when measuring, it is not enough to look at present or absent; the sentence in the answer has to be recorded too. Whether that sentence changes from month to month is a signal in its own right.
What Happens When The Model Is Updated?
The models behind the assistants are renewed from time to time. A new version may have been trained on different text and may answer the same question with different brands. That is why you can be named regularly on a question yesterday and fall behind on it today.
This change is hard to know in advance, but regular measurement shows it at once. Seeing on which questions you lost ground after a version change tells you where to look.
Named On One Question, Missing On Another
Most businesses are named on some questions and not on others. That spread is not random; it usually reflects where you give clear information and where you stay silent.
That is why it helps to group the questions by type. You may be visible on product questions and invisible on comparison questions. Or you may vanish on questions asked with a need rather than a place.
Where To Start?
Measure first. Write about ten questions your customers would ask, without a brand name, record the answers, and repeat the same questions a month later.
Note who appears in your place in the answers and which sources are relied on. That list tells you far more than guessing why you are behind.
Where Do You Stand In ChatGPT And The Others?
The pre-scan puts a few customer questions to Google’s AI and sends you the result.