Encar Used Car Collector

Korea's largest used-car marketplace, as a structured dataset — every listing under the manufacturer/model tree, with the history that actually decides a price.

Accident records, insurance history, inspection results, factory options, dealer and region. The fields a buyer squints at, in columns you can query.

What a row looks like

{
  "recordType": "encarVehicle",
  "manufacturer": "BMW",
  "modelGroup": "3시리즈",
  "modelName": "320i",
  "gradeName": "M Sport",
  "yearMonth": "2021-08",
  "priceManwon": 3980,
  "originPriceManwon": 6120,
  "mileage": 41200,
  "fuelName": "가솔린",
  "transmissionName": "오토",
  "colorName": "화이트",
  "region": "경기",
  "sellStatus": "판매중",
  "optionsStandard": ["헤드업디스플레이", "어라운드뷰"],
  "record": { "accidentCount": 0, "ownerChanges": 1, "repairedParts": [] },
  "photoUrls": ["https://ci.encar.com/..."],
  "detailUrl": "https://fem.encar.com/cars/detail/40852960",
  "scrapedAt": "2026-08-26T04:33:18.550Z"
}

Prices are in 만원 (10,000 KRW) exactly as Encar quotes them — no lossy conversion.

How it finds listings

Encar's search is a faceted tree, not a flat list. This actor walks manufacturer → model group → model, and reuses the query string the API itself returns for each facet rather than reconstructing Encar's bespoke query grammar. That is why stacked filters compose correctly instead of silently dropping.

Three stages: build the taxonomy, list each group, then fetch detail per vehicle.

Input

{
  "carType": "for",
  "manufacturers": ["BMW"],
  "modelGroups": ["3시리즈"],
  "taxonomyDepth": "model",
  "yearRange": [2020, null],
  "priceRange": [null, 4000],
  "accidentTypes": ["N"],
  "sortBy": "PriceAsc",
  "maxListingsPerGroup": 20,
  "maxTotalListings": 200,
  "collectDetail": true,
  "collectRecord": true,
  "collectInspection": true
}

28 filters are available — fuel, transmission, body type, seats, colour, region, service history, seller type, options and more. Leave them empty to sweep broadly, or stack them to carve out exactly one segment.

Field	Meaning
carType	for (imported) or kor (domestic)
taxonomyDepth	modelGroup for breadth, model for precision
collectRecord / collectInspection / collectDiagnosis	The history data — an extra request each
maxListingsPerGroup / maxTotalListings	Budget per facet and overall
proxyConfiguration	Residential by default — required, see below

Two things to know before running

A residential proxy is not optional. Encar refuses datacenter IP addresses at the network layer — the connection fails outright, and no amount of retrying from a cloud server gets through. proxyConfiguration defaults to Apify's residential group for exactly that reason. Residential traffic is billed by Apify on top of compute.

The same car gets listed twice. Dealers re-post vehicles under new ad IDs, so de-duplicating on the ad ID alone leaves the same physical car in your results more than once — in a 200-row sample, 21% were repeats. Rows are de-duplicated on vehicleId instead, which means a run can return fewer rows than you asked for: request 200 and you may get 144 unique cars. You are billed for what you receive. Set allowDuplicateVehicles: true if you are tracking listings rather than vehicles.

Good for

• Price modelling against mileage, year and accident history
• Dealer inventory and regional supply analysis
• Spotting underpriced listings in a segment
• Tracking how long specific models sit on the market

Notes

• Taxonomy rows go to a separate encar-taxonomy dataset so they do not count as results. taxonomyToDefaultDataset: true restores the old mixed shape.
• writeMarkdown emits a readable per-vehicle summary file alongside the dataset.