How We Verify Data Sources for Our Car Rankings

At a glance

Why source verification matters for rankings

A ranking is only as trustworthy as the weakest number behind it. If a "best resale value" list is quietly built on one dealer's guess, or a "most reliable" ranking leans on a single brand's own survey, the whole list is skewed before a reader ever sees it. Because our rankings influence real purchase decisions — sometimes involving tens of thousands of dollars — we treat sourcing as a formal step in the editorial process, not an afterthought.

The verification process, step by step

1. Identify the claim type. Safety rating, fuel economy, reliability score, resale value, ownership cost, or feature availability — each has a different set of acceptable primary sources.
2. Pull from a primary source first. Regulators (NHTSA, IIHS, EPA, Euro NCAP, DVSA), tax and registration authorities, and manufacturers' published spec sheets are treated as primary.
3. Cross-check with a second, independent source. A research firm, an insurer's loss-data report, or a second regulator's dataset must broadly agree before a figure is used.
4. Flag disagreements instead of averaging blindly. If two credible sources disagree by more than a reasonable margin, we report the range and say why, rather than inventing a false middle number.
5. Timestamp and schedule review. Every figure gets an "as of" date; the piece goes on a re-check calendar rather than sitting untouched indefinitely.

What counts as an acceptable source

How we handle estimates and ranges

Not every useful figure has a clean, published number — average total cost of ownership over five years, for instance, is always a modeled estimate rather than a fact. When we publish an estimate, we say so explicitly, name the inputs (fuel price assumptions, depreciation curve, insurance region), and give a range rather than a false-precision single number. If a claim can't be traced to a named, checkable source, it doesn't make it into a ranking, full stop.

Common mistakes this policy prevents

• Treating a manufacturer's own marketing copy as an independent reliability source.
• Using stale data (e.g., a three-year-old recall count) without flagging that newer figures exist.
• Averaging wildly different survey methodologies into one misleading number.
• Presenting a regional figure (say, UK insurance averages) as if it applies globally.

How often rankings get re-checked

Rankings tied to fast-moving data — insurance costs, EV incentives, fuel prices — are reviewed roughly every 6-12 months. Rankings built on slower-moving data — crash safety, long-term reliability trends — are reviewed every 12-18 months or whenever a model gets a significant refresh. Every article carries a visible "last updated" date so readers can judge freshness themselves.