Every AutoTrader listing that matches your criteria, in a spreadsheet, refreshed while you sleep.
AutoTrader is the largest used-vehicle marketplace in the United States, carrying millions of listings from franchise dealers, independents and private sellers. It publishes asking prices, mileage, trim, dealer identity and photos on every listing page — which makes it the single richest public source of what the US used-car market is actually asking for a given car today.
Dealers use it to price their own inventory against genuine local comps instead of a gut feel. Wholesalers use it to work out which trims are thin in which markets before they bid at auction. Lenders and analysts use it to track residual values across a segment. All three need the same thing: the whole comparable set as rows, not as forty browser tabs.
AutoTrader runs bot detection and rate limiting, renders significant parts of the listing through JavaScript, and paginates search results in a way that quietly caps how deep an unauthenticated session can go. The build handles this with a real browser engine, paced requests, rotating residential exits, and search segmentation — slicing by ZIP radius, price band and model year so no single query ever hits the depth cap and nothing gets silently dropped from the middle of a result set.
Excel, CSV, JSON, a Google Sheet that refreshes on a schedule, a direct write into MySQL or Postgres, or a REST endpoint your own tools can query.
Not one that is open to buy for general market research — their data feeds are commercial arrangements aimed at partners and syndication, and most dealers cannot get access. A scraper reading the public listing pages is what fills that gap for the majority of people who ask.
Tens of thousands per run is routine. The practical constraint is search depth rather than volume, which is why the build slices the market into overlapping segments by geography, price band and model year — that reaches the full matching set instead of the first few hundred results per query.
Daily overnight is the usual choice and suits pricing work. Hourly is possible for a narrow watch list — a specific competitor, a specific model — and costs more because it means more requests and more infrastructure.
Collecting publicly visible listing data is generally lawful in the United States, and US appellate courts have repeatedly held that reading public pages is not unauthorised access. I keep request rates civil, stay off anything behind a login, and avoid personal seller data. If a specific request looks like a problem I will say so before you pay.
A single-market daily feed typically lands between $300 and $700. Nationwide coverage with price history and competitor roll-ups runs higher. You get a fixed quote within 24 hours of telling me the filters and fields you want.