Data Methodology

Data source system

Every vehicle record is compiled from public sources and carries a record-level data_source and data_updated date. Sources are graded by tier — official and regulatory at the top, community reports at the bottom. The full tier definitions are on the Data Provenance page and the Data & Editorial Methodology page.

Since this batch, a parameter-level verification layer also records a source, verification date, model year, trim, and confidence for the six key parameters (price, range, battery, motor, acceleration, charging) on high-exposure vehicles, with a verified / mismatch / unverifiable / conflict status. Conflict values are flagged, never silently changed.

Parameter normalization rules

Fields are normalized to fixed enums rather than free text, so two vehicles can be compared on the same basis:

• Powertrain — BEV / PHEV / EREV / HEV / ICE / FCEV. Non-plug-in HEV and ICE records exist in the catalog and are labelled as such, not silently reclassified.
• Body type — Sedan / Hatchback / SUV / MPV / Wagon / Pickup / Van / Coupe / Convertible / Sports Car / Crossover / Other.
• Range — recorded against a named test standard (range_cltc_km, range_wltp_km, range_epa_km, range_nedc_km); no bare "range" field, and figures are never combined across standards.
• Units — kW for power, Nm for torque, kWh for battery, mm for dimensions, kg for weight, km/h for speed.

Where a field is absent, it is rendered as "not recorded" rather than estimated. The full field-level schema is in src/content.config.ts.

Price conversion method

Prices are kept on one of five strictly-separated bases — Official MSRP, China market reference price, China ex-factory reference price, FOB quotation, or Landed cost estimate. This catalog uses China ex-factory reference price by default, stated in USD.

The ex-factory figure is a factory-gate starting price, not a retail, FOB, or landed price. Where a figure's basis could not be confirmed it is labelled Reference (unverified). A reader comparing figures should confirm the actual quotation with a supplier — we do not publish transactional prices or quotes, and we never silently convert one basis into another.

Version matching

Where the manufacturer publishes multiple versions, the trim field records them verbatim (slash-separated). A single representative figure is recorded per standard per model, and the page states that figures apply to the base/representative configuration rather than inventing a per-trim breakdown that is not in the source data.

model_year is recorded where known; it is currently unfilled on most records and is shown as not recorded rather than guessed.

Update cadence

We review records and pages periodically rather than on a fixed schedule. Each record carries a data_updated date, and each page carries a last-updated date, so a figure's age is visible rather than hidden. Tariff and price data is revisited most often, since those move fastest.

Report a data problem

If a figure looks wrong, report it — every vehicle page has a "Report data issue" link that opens a pre-filled email with the model slug. Corrections are welcomed and, where material, recorded on the page rather than silently swapped out.