Vehicle marketplace data becomes comparable only after identity, condition, currency, mileage unit, seller type, geography, and observation time are normalized. Preserve the raw public listing and source URL because titles alone rarely identify a vehicle configuration precisely.
All Vehicle Scraper is the AgentX multi-source route across 52 marketplaces in the current catalog description. Dedicated Actors include AutoTrader US, AutoTrader UK, Cars.com, and SGCarMart.
What should a normalized vehicle record contain?
| Group | Core fields |
|---|---|
| Provenance | source, listing ID, canonical URL, collected time |
| Identity | make, model, trim, model year, body type, VIN where legitimately public |
| Condition | new/used, mileage, mileage unit, accident or history fields when provided |
| Offer | price, currency, displayed finance context, availability/status |
| Seller | dealer/private classification, displayed seller, source seller ID |
| Location | city, region, country, coordinates where available |
Source-specific equipment, engine, fuel, transmission, and warranty fields should remain in an extension layer when they cannot be standardized safely.
How should the same vehicle be matched?
A verified VIN is a strong identifier, but it may be missing, masked, or reused incorrectly upstream. Without it, matching can combine make, model, trim, year, mileage, seller, location, photos, and title details.
Keep listing identity separate from vehicle identity. The same vehicle may be relisted, transferred between dealers, or appear on several marketplaces.
How are mileage and price normalized?
Store the displayed value and unit. Convert miles and kilometers into a comparison field only with a documented factor. For price, retain currency and separate cash price from finance examples, deposits, monthly payments, taxes, and fees.
A low displayed monthly payment is not the vehicle price. Filter such values before computing market distributions.
Which metrics are useful?
Defensible observed-listing metrics include:
- inventory count by make/model/year and market;
- asking-price distribution for comparable configurations;
- mileage distribution;
- newly observed and no-longer-observed listings;
- visible days between first and last observation;
- dealer or marketplace coverage;
- share of listings with required fields.
Do not label an unobserved listing “sold” without confirmation.
Multi-source or marketplace-specific collection?
Use the all-vehicle Actor for discovery, cross-market dashboards, and a common schema. Use focused Actors when a marketplace-specific filter, seller field, or regional convention is important.
For a comparison test, run equivalent vehicle criteria on two sources and evaluate valid-match rate, unique listing count, field coverage, unit parsing, and source URL quality.
What should be checked before automation?
Review a sample in every country or marketplace included. Validate model-year parsing, price type, mileage unit, seller classification, duplicate groups, and inaccessible URLs. Add alerts for schema drift and extreme value changes.
The live Actor page remains the authority for current coverage. Browse the BestCrawler complete Actor catalog or the market-signals topic hub for related property and commerce workflows.
Continue in the directory
Turn the guide into a real sample run.
Open the current AgentX contract, check pricing and fields, then validate a narrow output.