Why does ChatGPT recommend my competitor and not me?
An AI assistant names a nearby business based on what it can read about that business online. That footprint is something an owner can audit, and partly fix.
- AI assistants draw on public information about a business, not on personal relationships or luck
- The signals are knowable, and an owner can audit their own footprint the way a model would
- Some fixes sit fully in the owner's control, such as what their own website says
- The process takes an afternoon, not a consultant retainer
The moment lands the same way every time. An owner types a question like "best HVAC near me" into an AI assistant, and the answer comes back with a rival's name. Not theirs. The first instinct is to treat it as random, or to assume the assistant picked a favorite. Neither is what happened. The assistant worked from information it could read about local businesses, and the rival simply had more of the right kind of information sitting where a model would find it.
That is the useful part of the story. The signals a model reads are public, which means an owner can look at the same things the model looked at. A competitor showing up in an AI answer is traceable to a pattern in the competitor's online footprint, not to chance. Here is how to run that audit yourself, and where the limits sit.
What an AI assistant actually reads
An AI assistant does not have opinions about local businesses. It has information. When it names a shop or a contractor, it is drawing on what is publicly readable about that business: the business's own website content, pages elsewhere that mention the business, and the kinds of public records any reader could find. A model reads roughly what a diligent stranger with a lot of time could read, and then it summarizes.
This is why the audit works. If a rival appears and your business does not, the explanation lives somewhere in what is publicly readable. Either the rival's footprint says more, says it more clearly, or says it in a form the model could use. The task is to find which of those it is.
Audit your footprint the way a model would
The audit is a read-only exercise at first. The goal is to see your business the way an assistant sees it, before changing anything. The method is to look only at public information, in the order a reader would encounter it.
- 01Search your business by name
Type your business name into a search engine and note what the first page says about you.
- 02Search your trade and city
Search the way a customer would, such as your trade plus your town, and see whether you appear at all.
- 03Ask an AI assistant about your trade nearby
Run the same customer question through an AI assistant and record which businesses get named.
- 04Ask the assistant what it knows about your business
Query your business by name and read what comes back, including anything outdated or wrong.
- 05Read your own website as a stranger
Check whether your site plainly states what you do, where you serve it, and who you serve.
- 06List every public page that mentions you
Collect the directory listings, review pages and other mentions that turned up in the earlier searches.
- 07Write down the gaps
Note where your business is absent, thin, outdated, or described inconsistently compared with the rival that got named.
The last step matters most. An audit without a written gap list turns into browsing. The gap list is what tells you which fixes are worth the effort.
What is in your control, and what is not
Your own website is the clearest case of control. You decide what it says, how plainly it states your services and service area, and how current it stays. If the audit found your site vague or thin on the basics a customer asks about, that fix is entirely yours to make.
Public pages you do not own sit in the middle. Listings and review pages follow their own rules, and no owner controls what a customer writes or what a platform displays. What an owner can control there is accuracy where the platform allows edits, and participation, meaning reviews exist and the business responds. That is slower work, and it is honest to treat it as such.
The one thing outside any owner's control is the model itself. Assistants choose their sources and methods, and those can change without notice. The realistic goal is not to reverse engineer a model. It is to make sure that any reader, human or machine, that goes looking for a business like yours in your area finds a clear, current, consistent picture. Businesses that get named in AI answers tend to be the ones where that work is already done.
A fair way to judge progress
Run the same questions again after the fixes, a few weeks apart, and compare the answers. Treat a single result as noise and a pattern as signal. If the assistant's description of your business becomes more accurate, that is real progress even before your name starts appearing in recommendations, because it means the information the model reads is improving.
