Marketplaces vs manufacturers: where ChatGPT looks for car answers
In August 2026 ChatGPT cited car marketplaces and review sites far more than manufacturer sites; a comparison of the source types and what dealers and brands can do about it.
Quick verdict
In August 2026, car shoppers asking ChatGPT questions were pointed to marketplaces and review sites well ahead of manufacturers. Promptwatch's automotive citation report (August 1 to 31, 2026, ChatGPT) puts Autotrader.com first at 5% of all automotive citations and Cars.com second at 4.6%. The first manufacturer to appear is Chevrolet.com at 2.62%, in fourth place, and it is the only manufacturer site in the top eight.
The short version for a dealer or brand: the places that list and review cars are doing the talking. Your own site still earns citations, but it is not the main stage.
What was measured
The report ranks 31 domains by their share of all ChatGPT automotive citations in the month. It is one vendor's dataset, for one month, in one engine. Seven of the top eight domains are marketplaces or research and review sites. Reddit is ninth at 1.34%. At the other end of the chart, birchwoodgroup.co.uk, a single dealer group, sits at 0.28% and still makes the top 31.
That last detail is worth pausing on. A dealer group's own website can appear in the chart at all, which tells us the local and brand-owned pages are not shut out. They are just small.
Comparing the source types
| Source type | Share of all automotive citations | Notes from the report |
|---|---|---|
| Marketplaces and listings | 14.65% | Autotrader.com 5%, Cars.com 4.6% |
| Reviews and research | 12.62% | Seven of the top eight domains are marketplaces or research and review sites (the report's wording) |
| Manufacturer sites | 7.92% | Chevrolet.com 2.62%, fourth overall |
| Community and news, dealerships and service and finance combined | 2.12% | Reddit 1.34% (ninth); one dealer group at 0.28% |
Source: Promptwatch data report, August 2026, ChatGPT. Listings and reviews together hold 27.27%, about three and a half times the manufacturer group. Marketplaces on their own hold nearly twice the manufacturer share.
Reading the numbers carefully
First, a share of citations is not a share of clicks or of sales. The report counts how often a domain is cited, not what shoppers did next.
Second, the dataset cannot tell you why marketplaces win. A plausible reading is that listing sites hold structured inventory, prices and specs for many makes at once, and review sites hold comparisons that an answer can quote. That is a guess about how the engine uses sources. The report shows who was cited, not the mechanism.
Third, this is one month. Other Promptwatch reports record sharp shifts in August 2026, including a Reddit collapse in ChatGPT, so check any chart you rely on against a fresh window before committing a budget.
Fourth, it is automotive prompts in ChatGPT. It does not tell you what Perplexity or Google's AI features cite, and the report does not claim it applies in every country.
What to do without any tool
The report suggests three directions, and none of them needs software.
- Keep inventory, prices and specs accurate on the big marketplaces. If an answer quotes a listing, a wrong price or a sold car can follow your brand into the answer. Audit your feeds and listings monthly, and remove stale vehicles.
- Maintain complete profiles on review and reputation sites such as DealerRater, Birdeye and BBB, which the report names. Fill in every field, respond to reviews and keep your address, hours and categories consistent.
- Build focused pages on your own site per model and trim, per ownership cost question and per local availability question. Dealership sites sit in the smallest group in the chart, so make each page specific enough that a general marketplace page cannot answer the same question.
A fourth step is manual and cheap. Write ten questions a local car buyer might ask, such as comparing two trims or asking what a model costs to run. Run them in ChatGPT, copy every cited domain, and tally the source types. If your own domain never shows up, or a competitor's dealer page does, you have a concrete list of fixes.
Measuring it with a platform
Doing that tally weekly is tedious. In our catalog, Promptwatch is the platform that documents domain-level and page-level citation analytics, a Page Tracker for specific URLs, and Citation Trends that classify each citation by content type and source type. For this topic the useful parts are:
- Citation analytics to see which marketplace and review domains are cited for your prompts, and which competitors share those answers.
- Page Tracker to follow a particular model page or a competitor's listing page and see whether it gets cited.
- Brand Narrative, available on every plan, which pulls the claims AI makes about you and competitors out of tracked responses and scores them by theme such as pricing and reputation. For a dealer, a wrong claim about warranty or pricing is a task, and this finds it.
- Local GEO and SEO targeting, on Business, Agency Scale and Enterprise, which pins a monitor to exact coordinates so answers are checked as if asked from a given spot. Plain country-level tracking is on every paid plan, and state and city tracking starts on Business.
On cost, Business is $579 a month, Professional is $245 and Essential is $95. Crawler logs start on Professional, and Essential has none. Fact Check, the Brand Narrative step that checks a claim against your own crawled pages, needs Business, Agency Scale or Enterprise.
Other tools in our catalog may do part of this, but our entries for AthenaHQ and Scrunch AI do not record source-type classification or location pinning, so we treat those as unknown.
Who should pick which
- A single independent dealer: fix marketplace listings and review profiles first, then publish a few model pages. Run the manual ten-prompt test monthly. Move to a tool when you have more than a handful of questions to track.
- A multi-location dealer group: you need city-level checks. Promptwatch's state and city tracking and coordinate targeting start on Business, so budget for that tier if location is the point. See our guide for dealer groups across several cities.
- A manufacturer or national brand: you sit in the third of the three big groups, behind listings and reviews. Treat marketplace and review accuracy as a brand job, not a dealer job, and watch which answers name you.
- A marketplace or review site: the report is a reminder that your listings are the product the engine is quoting. Guard data quality.
For related reading, see what the best AI visibility tool for car dealerships looks like and how an automotive brand can track whether ChatGPT recommends its models.
Bottom line
For car questions in ChatGPT, one month of data shows listings and reviews ahead of manufacturer sites by a wide margin. That is a case for tidying the places you do not own before polishing the ones you do. Nothing here promises a citation, and the next window may look different, so check your own prompts.