Sources
Where the data comes from
Five federal datasets. All public domain, no API key, no licence fee, no scraping of commercial sites.
| NHTSA complaints (bulk) Every complaint filed with the Office of Defects Investigation. The authoritative source for this site. https://static.nhtsa.gov/odi/ffdd/cmpl/FLAT_CMPL.zip | 2,233,015 records |
|---|---|
| NHTSA recalls API Recall campaigns by vehicle. Units affected are only available per campaign, from a second endpoint. https://api.nhtsa.gov/recalls/recallsByVehicle | 3,406 campaigns |
| NHTSA ODI investigations (bulk) Open and closed defect investigations. Only open investigations affect a score. https://static.nhtsa.gov/odi/ffdd/inv/FLAT_INV.zip | 154,248 rows |
| NHTSA vPIC Canonical make, model and model-year taxonomy. https://vpic.nhtsa.dot.gov/api/ | 195 car makes |
| fueleconomy.gov EPA fuel economy, engine and drivetrain specifications, 1984 to present. https://www.fueleconomy.gov/feg/download.shtml | 45,472 rows |
Refresh
Sources are re-pulled monthly and the whole dataset is recomputed from scratch; the pipeline is deterministic, so the same inputs always produce the same scores. Current vintage is August 9, 2026.
What we deliberately do not use
- Parts-retailer fitment data. It is contractually licensed. We do not scrape it.
- Manufacturer marks and press imagery. Trademarked. No badge appears on this site.
- Purchased sales figures. Not yet licensed, which is why cross-model comparison is withheld.