AutoScout24 Listings Scraper — Used Cars, Prices & Sellers

Scrape AutoScout24 used-car listings from any search URL: make, model, trim, price, mileage, fuel, gearbox, location, seller and images. HTTP-only, no account, no browser.

Input

Field	Type	Default	Description
searchUrls	array	required	Result-page URLs with your filters already applied
maxItemsPerUrl	integer	200	Stop a search after this many unique listings
maxPages	integer	20	Pages per search (site ceiling is 200 — see below)
maxConcurrency	integer	2	Searches in parallel; pages within a search stay sequential
proxyConfiguration	object	RESIDENTIAL	Recommended for sustained volume

Build the search in the browser with the filters you want, then paste the URL. Any AutoScout24 domain works. page is managed by the actor — it is rewritten rather than appended, so a URL that already carries one is handled correctly.

Output

{
  "_input": "https://www.autoscout24.com/lst/bmw/118?atype=C",
  "_source": "S1-next-data",
  "_scrapedAt": "2026-09-07T03:12:42Z",
  "_page": 1,
  "_tlsProfile": "chrome110",

  "listingId": "9b0a3d2a-…",
  "listingUrl": "https://www.autoscout24.com/offers/…",
  "make": "BMW",
  "model": "118",
  "modelVersion": "Navi WLAN PDC Lordos LED",
  "priceRaw": 11999,                 // numeric, straight from upstream
  "priceFormatted": "€ 11,999",
  "mileageInKm": "98,904 km",        // as rendered
  "mileageKm": 98904,                // parsed
  "fuel": "Gasoline",
  "transmission": "Manual",
  "offerType": "Used",
  "city": "Celle", "zip": "29223", "countryCode": "D",
  "sellerType": "Dealer", "sellerName": "…",

  // raw sub-objects, upstream field names verbatim
  "vehicle": {…}, "price": {…}, "location": {…}, "seller": {…},
  "images": […], "ratings": {…}, "statistics": {…}, "wltpValues": {…}
}

All twelve convenience fields were filled on 120 of 120 listings in testing.

How it works

Data comes from the page's __NEXT_DATA__ island (props.pageProps.listings), which carries far more than the cards show — including the seller's phone links and WLTP emission values.

The 200-page ceiling is real, and the actor says so

numberOfPages comes back as 200 whatever the search is: 200 for a query reporting 8,849 results, and 200 again for one reporting 827,638. At ~19 rows per page that is roughly 4,000 listings per search URL, full stop.

Silently returning 4,000 of 827,638 rows would look like a bug in the actor, so it warns once per URL when it detects the cap and tells you what to do instead — split the search into narrower filters (price band, registration year, region) and pass several URLs.

Two field names that lie

Worth knowing if you extend this actor, because both produce a wrong answer rather than an error:

• modelVersion does not exist. The trim level is modelVersionCustom (or modelVersionInput, or only subtitle). Reading the obvious name gave a column that was empty on all 120 test rows — which reads as "this site has no trim data" rather than "wrong key".
• mileageInKm is a string, e.g. "98,904 km". The raw value is passed through and a parsed mileageKm integer is offered beside it.

Duplicate rows across pages

Consecutive pages overlap — sponsored and rotated slots reappear. Listings are deduplicated by id, and a page that contributes nothing new ends the crawl for that URL.

Known limits

Limit	Detail
~4,000 rows per search URL	The site's own paging ceiling; narrow the filters to go deeper
List pages only	Full spec sheets and descriptions live on the detail page, not fetched here
Seller phone	Present in the raw seller object as links, not resolved to a number