How the check works
The free scorecard does not ask "what do you think of Company X?". That question already contains the answer. It asks what a buyer would ask: how do I pick a reliable local agent for a flat in this district, which carpet brands are worth it, who installs kitchen hood fire suppression. If the business shows up, ChatGPT brought it up on its own.
The sectors were residential real estate in one Istanbul district, carpets and kilims, industrial high-speed drills, and commercial kitchen ventilation with fire suppression. We are not naming the businesses.
Who ChatGPT named instead
Real estate: platforms, not agents
No local agency was named. The names were Sahibinden.com, Hepsiemlak and Emlakjet, plus descriptions like "local offices of corporate real estate networks". Several answers named no one and gave a checklist instead: compare the listing with the title deed, verify the company details, ask about experience in the micro-area.
Carpets: national brands
Merinos, Royal Halı and Sanat Halı each appeared in three of five answers. Stepevi, Saray Halı and Pierre Cardin Halı appeared in two. These are national brands and store chains. A small maker with its own workshop did not come up.
Drills: global makers, and a wrong turn
Makita, Milwaukee, DeWalt and Metabo came up as expected. But the phrase "high speed drill" also took ChatGPT to dentistry and surgery: Dentsply Sirona, Stryker, Bien-Air. The words used for a category are not always the words buyers and AI engines connect to it.
Kitchen fire suppression: system brands, no installers
The answers named system manufacturers such as ANSUL, Amerex and Buckeye. The rest was a description of the category: hood ventilation, duct cleaning, smoke extraction. No company that actually installs these systems was named.
What this suggests
The pattern is the same in all four sectors. When nobody names a business, ChatGPT builds its shortlist from names that are already written about widely: national brands, global manufacturers, large marketplaces. A small business is not missing because it is bad. It is missing because there is little independent text that describes it.
Three practical consequences follow from what we saw:
- A platform in the answer is a door, not a rival. If ChatGPT sends a flat hunter to Sahibinden, a complete and active profile there is how you enter that answer.
- Say plainly what you sell and to whom. "High speed drill" alone was read as surgery. "CNC drilling tools for metalworking shops" leaves less room for a wrong match.
- If you install or service a brand the answer already names, say so where others can read it: on your site, in directory listings, in customer reviews. It connects your name to a name ChatGPT already uses. This is a hypothesis to measure, not a promise.
What not to do
Do not decide from one scan. Ask the same question again and the answer changes; a single scorecard is a snapshot. Do not try to create a Wikipedia page for your business: Wikipedia has a notability rule and a company writing its own page is a conflict of interest. And do not expect a single file or tag on your site to make ChatGPT start recommending you.
Limits of this sample
This is 25 answers from one engine (ChatGPT) in one week, across 4 sectors. It is enough to see a pattern, not enough to put a percentage on the whole market. The other engines (Gemini, Claude, Perplexity, Google AI Overviews) may name different businesses, which is exactly why paid plans measure all five. We will update this post as the sample grows.