The showroom moved into the chat

How do buyers use AI to shop for vehicles - and who gets recommended?

Vehicle buyers ask AI engines the questions they used to ask a salesperson: what's reliable, what's a fair price, what should I check before I buy. The engines answer with marketplaces and models far more often than with a specific dealer - because marketplaces are what the engines can actually read.

28 Labs · September 2026

Buyers are already using AI to shop for cars at scale: Cox Automotive's Car Buyer Journey Study found 25% of new-vehicle buyers engaged AI tools like ChatGPT or Copilot during their research, and a separate Cars.com survey put AI search tool usage among car shoppers at 44%. When engines answer, they name marketplaces, review sites, and specific models more often than individual dealers, because that's where the comparison data, pricing, and reviews live. CarMax became the first U.S. auto retailer with a shopping app inside ChatGPT in February 2026 - the clearest signal yet that this channel is now transactional, not just informational.

What buyers are actually asking

Four question shapes cover most of vehicle-shopping AI traffic:

Each shape pulls from a different part of the web. Comparison queries lean on owner forums, long-form reliability guides, and review aggregators. Budget and local queries lean on marketplace listings with live pricing and inventory. Stage-specific queries lean on buying guides from consumer sites and financial institutions. A brand that only shows up in one of these shapes is invisible for the other three.

What the engines do with those questions

None of the major engines answer from memory alone for a query this time-sensitive. They fan out into a handful of searches, read the pages that come back, and build the answer from what they find - the same mechanism we cover in how to navigate AEO. For vehicle queries specifically, that means reading:

Sources engines lean on

  • Marketplaces (Cars.com, CarMax, Carvana, AutoTrader)
  • Review and rating sites
  • Owner forums and enthusiast communities
  • Buying guides from consumer and financial publishers
  • Manufacturer spec pages

What individual dealers rarely offer

  • Structured, comparable listing data
  • Independent review volume
  • Content answering the buyer's actual question, not just their inventory
  • A presence outside their own homepage
The engine isn't biased against dealers. It's reading for structure and evidence, and marketplaces have more of both.

ChatGPT's shopping features are now transactional, not just informational

This shifted from research tool to storefront in early 2026. CarMax launched an app in the ChatGPT app store on February 27, 2026 - the first U.S. auto retailer to do so - letting shoppers search listings, explore vehicles, and get an instant value estimate on their current car with a direct path to CarMax's online offer tool, all inside the chat. Separately, ChatGPT's Instant Checkout lets shoppers buy directly from product cards for participating merchants - a "Buy" button that handles payment and order confirmation without leaving the conversation, currently built mainly around Shopify and Etsy catalogs.

Full vehicle checkout inside a chat window isn't the default experience yet. But the direction is clear: being readable to an engine is no longer just about getting mentioned. For a marketplace or a retailer with a real integration, it's about being the thing the buyer transacts with without ever opening a browser tab.

The buyer's path from question to purchase

1
Ask AI a budget or comparison question
2
Engine reads marketplaces, reviews, forums
3
Answer names models, marketplaces, sometimes a dealer
4
Buyer visits the cited source directly
Cox Automotive's research found that once initial questions are answered, shoppers most often go straight to the cited dealer or manufacturer site next - the AI answer is the referral, not the endpoint.

Why buyers trust the AI over the dealer's own site

A dealer's website exists to sell that dealer's inventory, and buyers have learned to discount it accordingly. An AI engine reading across marketplaces, reviews, and forums feels neutral, even though it's built from largely the same commercial content underneath. Buyers who used AI in their research reported meaningfully higher satisfaction with the buying process overall in Cox Automotive's study, and separate research found most shoppers trust AI tools to give reasonably unbiased vehicle information. The trust isn't really in the AI. It's in the fact that no single seller controls what the engine says.

Who wins these answers, structurally

Aggregators and marketplaces are built for this moment: deep comparison content, high review volume, structured and current listing data, and enough scale that engines can cite them with confidence. Individual dealers show up mostly as citations inside a marketplace answer, not as the answer itself - their visibility is inherited from the platforms they list on, not generated by their own site. This is the same asymmetry we've written about for other categories in why does ChatGPT recommend your competitor: engines recommend the entity with the clearest, most current, most comparable evidence, and that's rarely a single-location homepage.

What that means in practice differs by seat:

The honest readVehicle shopping is already one of the most AI-mediated purchase categories that exists. The winners aren't decided by ad spend or homepage design - they're decided by whether an engine can read you, compare you fairly, and recommend you with confidence. That's measurable per question, per engine, today. Most brands in this category have never checked.

What we do about it

We ask AI engines the real questions vehicle buyers ask - budget, comparison, financing, local - across ChatGPT, Claude, Gemini, and Perplexity, and read exactly what comes back: who gets named, who gets recommended first, and which sources the engine cited to get there. Every count re-derives to a specific question and a specific answer, so a marketplace or dealer group can see precisely where they're winning the comparison and where a competitor is getting cited instead.

28 Labs measures how AI engines answer real buyer questions - and what moves the counts. Every number we publish re-derives to specific questions and engines. try28labs.com