Watch: AI Search for Car Dealers: Can ChatGPT Read Your Inventory?
On October 23, Cox Automotive published a new look at its AI in Auto Retail Tracker, and one gap stands out for anyone running a used-car lot: 63% of in-market shoppers say they will definitely or probably use AI on their next vehicle purchase, while only 29% of dealers say they have started adjusting for AI-powered search. For an independent lot, AI search for car dealers comes down to whether those tools can read accurate facts about your store. A shopper in your town can ask ChatGPT, Gemini, or Google’s AI Mode which lots carry a clean SUV under $15,000, and the answer gets built from whatever the assistant can find about you today.
This post covers what Cox found, how these assistants pick up facts about a used-car lot, and a checklist an independent dealer can work through this week on the lot site, the vehicle pages, and the business listings. None of it requires buying a new tool.
What Cox Automotive found about AI search and car dealers
The Cox Automotive AI in Auto Retail Tracker surveys franchise and independent dealers (483 in the second-quarter wave, fielded in May 2026) and 1,502 shoppers who plan to buy within 12 months. The October 23 write-up, published through CBT News, pulled the shopper and dealer numbers side by side. The ones that matter for a lot website:
- 63% of shoppers say they will definitely or probably use AI during their next vehicle purchase, roughly steady from 61% in the first quarter.
- AI tools (36%) now sit close to automotive sites (38%) as a research channel.
- 67% of shoppers trust AI at least some, flat from the first quarter.
- Only 29% of dealers are actively adjusting for AI-powered search, down from 32% in the first quarter.
- 32% of dealers know they need to act and have not started, up from 26%. Another 26% are not sure what AI search means for them, and 23% say it is not on their radar.
- Among shoppers who use AI, 26% use it to come up with questions to ask the dealer, 24% say it makes them feel more prepared, and only 17% use it to avoid talking to staff.
Cox also pointed out that buyers are less decided when they start. In 2020, 37% knew exactly which vehicle they wanted at the start of shopping. Today 29% do. So a shopper may open ChatGPT with a budget and a body style, not a year, make, and model, and ask the assistant to narrow it down.
For an independent lot, that last part is the opening. A question like “best used SUV under $15,000 near me” is a budget and inventory question, and independent dealers own a big share of that inventory. The lots whose prices, mileage, and stock status are readable on the open web give the assistant something concrete to cite.
How AI search finds a used-car lot
Dealers hear a lot of vendor theory about how AI picks businesses. The useful parts come from the documentation OpenAI and Google publish, and both say roughly the same thing: the assistants work from pages they can crawl, index, and read as text.
ChatGPT
When a question needs current local information, ChatGPT can search the web and show the sources it used. OpenAI’s crawler documentation names OAI-SearchBot as the crawler that surfaces websites in ChatGPT search. Sites that block it will not be shown in ChatGPT search answers, though they can still appear as plain navigational links. OpenAI recommends allowing OAI-SearchBot in robots.txt and letting its published IP ranges through your host or firewall, and it says a robots.txt change takes about 24 hours to register for search.
GPTBot is a separate crawler that collects content that may be used to train OpenAI’s models. You can block GPTBot and still allow OAI-SearchBot. Plenty of dealer sites block both because someone copied a “block all AI bots” snippet from a forum, which takes the store out of ChatGPT search answers entirely.
Google AI Overviews and AI Mode
Google’s guidance on AI features and your website says there are no extra technical requirements to appear in AI Overviews or AI Mode. A page has to be indexed and eligible to show in Google Search with a snippet. Google lists the usual basics as still worth doing, including allowing crawling in robots.txt and at the CDN, keeping important content in text form, making sure structured data matches the visible text on the page, and keeping Merchant Center and Business Profile information current. Google also says you do not need new machine-readable files, AI text files, or special schema to appear.
Since August 31, 2026, Search Console has a Search generative AI control under Settings for every site. The default includes your site in AI Overviews, AI Mode, and generative features in Discover. If someone at your store or your website vendor switched it to exclude, your pages stop appearing in those answers within a few days. It takes two minutes to check.
An AI search for car dealers checklist you can run this week
Work through these in order. The first three cover most of the problems we see on independent lot sites, and each one also helps regular Google rankings and marketplace leads.
1. Make sure the crawlers can get in
Open yoursite.com/robots.txt in a browser. Look for lines that disallow OAI-SearchBot, Googlebot, Bingbot, or a blanket rule for every user agent. A setup that keeps search access open while opting out of model training looks like this:
User-agent: OAI-SearchBot Allow: / User-agent: GPTBot Disallow: / User-agent: * Allow: /
Then ask your website provider, in writing, whether a CDN or bot-protection service challenges automated visitors. A robots.txt file that allows OAI-SearchBot does nothing if a firewall blocks OpenAI’s IP ranges before the crawler reaches the page.
2. Put the vehicle facts in plain page text
Each car needs its own URL with the year, make, model, trim, price, mileage, VIN, stock number, and lot address in the HTML, not baked into a photo or loaded by a script after the page opens. A December 2024 analysis by Vercel, The rise of the AI crawler, found that OpenAI’s crawlers did not render JavaScript at the time. Inventory widgets that build the page in the browser can show a nice listing to a person and a blank shell to a crawler.
A quick test: open a vehicle detail page, right-click, choose View Page Source, and search for the VIN. If the VIN is not in the source, assume an AI crawler cannot see it either. Our explainer on what makes a vehicle page crawlable walks through the rest of the page basics.
3. Show one price everywhere, with mandatory fees included
Assistants pull from your site, your Google listing, and marketplaces like CarGurus and Cars.com. If the website says $14,995, the feed says $13,995, and the description adds a $699 dealer fee “not included,” the answer a shopper gets can quote any of those numbers. That is a bad first impression, and under the FTC’s September guidance it is also a compliance problem. The advertised price has to be the price any buyer can pay, with only government charges left out. Our breakdown of the FTC price transparency FAQs for dealers covers the fee rules, and the post on feed and vehicle page price mismatches covers keeping the feed and the page in sync.
4. Add structured data that matches the page
Google says no special schema is required for its AI features, but structured data that matches the visible text still helps search systems read a page without guessing. On a used-car site that means AutoDealer markup on the homepage and contact page (name, address, phone, hours) and Vehicle or Car markup on each vehicle page (VIN, mileage, price, availability). The prices and mileage in the markup have to match what the page shows. If you want to check what you have, our vehicle and AutoDealer schema generator builds the JSON-LD from your own facts and flags missing fields in markup you paste in.
5. Make the store details match on every listing
Use the same business name, address, phone number, and hours on your website, Google Business Profile, Bing Places, Apple Business Connect, Facebook Page, Yelp, and your dealer profiles on each marketplace. Pay attention to old phone numbers from a tracking vendor you dropped, a second lot you closed, or a “Sunday by appointment” line that lives on one listing and nowhere else. When an assistant finds two versions of your hours, it has to pick one, and it may pick the wrong one. Our guide to local SEO for car dealerships covers the Business Profile side in more depth.
6. Answer the questions shoppers bring
Among shoppers who use AI, Cox found the top way it helps with the dealer relationship is generating questions to ask the dealer. Those questions are predictable on a used lot: Can I trade in a car I still owe money on? What do I need to bring for financing? Do you hold a car with a deposit, and is it refundable? Do you take out-of-state buyers? Is there a warranty on this car or is it sold as-is?
Put short, honest answers to those on your own site, one question per page section, in your store’s words and with your store’s real policies. A “Financing” page that lists the documents you ask for (proof of income, proof of residence, a valid license, references) gives an assistant a specific answer to cite. A page that says “we work with all credit types” gives it nothing.
Budget pages help too. If your site can show a live, crawlable page of every car under $15,000, or every third-row SUV on the lot, that page matches the way Cox says shoppers now start: with a budget and a use, not a model. Keep these pages tied to live inventory so they never show sold cars.
7. Pull sold cars quickly
A car that sold Saturday and still shows “available” on Tuesday is a wasted lead and a bad review waiting to happen. When an AI answer cites that listing, the shopper calls about a car you no longer have. Mark units sold or remove them the same day, and make sure the marketplace feeds update on the same cycle. Our FTC FAQ breakdown also covers why a sold unit should never stay advertised and why off-site or in-transit units need to be labeled on the listing.
How to test what ChatGPT and Google say about your lot
Do this once now and once a month. Open a fresh ChatGPT chat with search on, then do the same in Gemini and Google’s AI Mode. Ask the questions a local buyer would ask, using your town by name:
- Which used-car dealers in [your town] have a good reputation?
- Where can I buy a used SUV under $15,000 near [your town]?
- Does [your dealership name] have any pickups under 100,000 miles?
- What are [your dealership name]’s hours on Saturday?
- Can I trade in a car with a loan at [your dealership name]?
For each answer, write down the date, which assistant, whether your store was named, which URLs were cited, and anything wrong. Ask each question two or three times, because answers change from run to run.
The sources list matters most. If the assistant quotes old hours from a directory, fix the directory. If it cites a marketplace listing instead of your vehicle page, compare the two pages and figure out what the marketplace page shows that yours does not, which is usually price, mileage, or VIN in text. If your store never shows up on the budget question, look at whether your site has a crawlable page that answers it.
Vendor research suggests budget and inventory questions are where most lots disappear. VIN Index, which sells dealer AI visibility services, published a September 29 sample of 216 saved answers from ChatGPT, Gemini, and Perplexity in Ottawa. The two “where should I buy” questions named a reviewed local dealership in every answer, a used SUV under CAD $25,000 question named one in 17%, and “best used seven-passenger SUV under CAD $30,000” named one in 6%. It is one market and one vendor’s sample, but it matches what you would expect: general questions get general answers unless a dealer page gives the assistant real inventory to point to.
How to measure AI search traffic
OpenAI’s publisher FAQ says ChatGPT adds utm_source=chatgpt.com to referral links, so in Google Analytics you can filter sessions by that source and see which pages people land on. Expect small numbers at first. A shopper who reads an answer that names your store and then searches your name on Google shows up as a branded search or a phone call, not as a ChatGPT session.
For Google, clicks from AI Overviews and AI Mode are counted inside the regular Web search numbers in the Search Console Performance report, according to Google’s documentation. The Search Console help page for the generative AI control also points to a Generative AI performance report. Check it after you make changes to robots.txt or vehicle pages.
The simplest tracking for a small lot is still the sales desk. Add “found us through ChatGPT or Google AI” as a lead source option in your CRM or on the deal sheet, and ask every walk-in how they heard about the car.
What not to pay for
ChatGPT, Gemini, and Google pick the sources for each answer themselves. Allowing a crawler only makes a page available to them. No vendor can sell you a guaranteed spot in an AI answer. Be careful with pitches that lead with a special AI file, a proprietary “AI schema,” or a monthly fee to “submit” your dealership to ChatGPT. Google says outright that its AI features need none of those things.
Spend the time on the work above instead. Every item on the checklist also improves classic Google rankings, Google Vehicle Ads quality, and marketplace lead quality, so none of it is wasted if AI search grows slower than Cox’s survey suggests.
Where AutoFire fits
AutoFire gives every published vehicle its own detail page on your lot site, and the JSON-LD structured data and meta tags are generated from the listing fields. Update the price or mileage on the vehicle and the page updates with it, so the markup and the visible text stay in step. Your hours and contact details sit on the same site with consistent branding. AutoFire does not promise rankings or AI citations, but it handles the page structure covered above. You can see how the vehicle page SEO and structured data work, or look around at goautofire.com. If you also want ChatGPT or Claude to help manage inventory from the back office, we covered that in our guide to running your dealership with MCP.
Whatever platform you use, start with the first three checks this week: robots.txt and firewall access, vehicle facts in the page source, and one fee-inclusive price on every channel. Then run the five test questions and save the sources list. That gives you a baseline to compare against when Cox publishes its third-quarter wave.
This post is general marketing guidance, not legal advice. Check state advertising and fee rules with your attorney or state dealer association.
Leave a Reply