Does ChatGPT recommend my dealership? How to check in 5 minutes

Step 1: Write a buyer's question, not yours

Pick one question a real shopper in your area would type, and leave your dealership's name out of it. If you ask “Is Example Auto Sales a good dealer?”, the assistant talks about you because you asked. That tells you nothing about whether it would bring you up on its own.

Good blind questions sound like this:

• “What are the best used car dealers in [your city], TX?”
• “Which used car lots in [your city] offer in-house financing or buy here pay here?”
• “Where can I buy a used truck in [your city] with bad credit?”
• “What's a trustworthy used car dealer near [your neighborhood or ZIP code]?”

Use the city or neighborhood your buyers actually live in. If many of your customers speak Spanish, write a Spanish version too — for example, “¿Qué lotes de carros usados en [tu ciudad] ofrecen financiamiento en casa?” — and treat it as a separate question, because the answer can be different.

Step 2: Ask from a clean session

Assistants can tailor answers to the person asking. OpenAI's help center says ChatGPT may use an approximate location based on your IP address for local results, that a VPN can change that location, and that saved location details can also shape search results. If you've spent the afternoon researching your own dealership, your history may tilt the answer, too.

• Use a private or incognito browser window, signed out where the assistant allows it.
• Turn off any VPN, and either ask from your own area or put the city in the question.
• Start a new chat for every question so an earlier answer doesn't carry over.

Step 3: Ask each assistant, then ask again

Ask the same question in ChatGPT, Gemini, Claude and Perplexity. Then open a new chat and ask it twice more in each one. Three tries per assistant is enough to see a pattern without turning this into a project.

If an assistant shows a web search option or lists sources, note whether search was on. Answers with sources are easier to act on, because you can see which websites, listings or review pages the assistant used.

Step 4: Write down what you see

A screenshot is fine; a simple log is better, because you'll want to compare it next month. For every answer, record:

• Date and time, the assistant, and the mode if it showed one.
• The exact question, word for word.
• Named? Yes, no or unsure. Mark “unsure” when it names something close — a similar name, an old name, or your name in the wrong city — and check it later instead of counting it either way.
• Who was named, in order, and how many dealers it listed.
• Sources it linked to, if any.
• Anything wrong about you: hours, address, phone, a location that closed, or financing you don't offer.

Then sum it up the honest way: “Named in 1 of 3 ChatGPT answers on October 20.” That sentence describes those checks only. It isn't a verdict on your business, and it isn't a ranking.

Why the answers keep changing

Don't be surprised when the second answer differs from the first. In SparkToro's January 2026 study, volunteers ran the same brand-recommendation prompts through ChatGPT, Claude and Google's AI nearly 3,000 times. The same list came back less than 1 time in 100, and the same list in the same order less than 1 time in 1,000.

That's why position — “we're number 3!” — means very little. What holds up better is how often you're named across many tries. SparkToro's conclusion was that visibility across many runs is a reasonable measure, and a single “AI ranking position” is not.

Answers also differ between assistants because they lean on different sources. Some search the web while they answer, some rely more on what they learned in training, and Google's Gemini can use public information from Google Maps. A lot that one assistant names and another leaves out isn't unusual.

What your results tell you

• Named in most answers: keep your basics current — hours, photos, review replies — and check again every month.
• Named sometimes: the assistants know you, but other lots look like a closer match to the question. Make what you offer easy to find: the right Google category, a financing page, recent reviews with replies.
• Never named: first find out whether the assistants have your facts right. Now ask the named question — “What can you tell me about [your dealership] in [your city]?” — and compare the answer with reality. Wrong or missing facts are the first thing to fix.
• Unsure matches: look at the sources. A name close to yours may be a different business, or a sign that your name is written differently on your website, your Google profile and your sign.

What to fix first

Assistants learn about local businesses from the same places buyers look: your Google Business Profile, your website and your reviews. Start there:

• Google Business Profile fields AI assistants read — the category, description, hours and review replies worth getting right first.
• AutoDealer schema: what it is and how to add it — a small block of code that states your name, address, phone and hours in a standard format.
• A plain financing page, if you offer in-house financing or buy here pay here. Say only what's true, and link it from your main menu.

If you'd rather not run the checks by hand, AIForgotYou's free dealer check asks an AI assistant a buyer's question for your city — without your name in it — and shows you the exact answer, whether you were named, and a first fix. We ask through each provider's API, and API answers can differ from what you see in the apps, so every result is labeled with its question and date.