Every meaningful change we make to data, methodology, or editorial coverage gets logged here. That includes reader-flagged errors, internal reviews that changed our minds, and methodology tweaks that affect how we score vehicles.
We keep this page because a site that ranks 16,000 vehicles by reliability is going to get some of them wrong, and the only honest move when that happens is to fix it in public. Strikethrough or follow-up notes only — we don't rewrite old entries.
If you spot something wrong on the site — a recall we missed, a complaint we mis-classified, a methodology assumption that doesn't hold up — the contact page is the fastest way to flag it.
The "still under factory" check was using a flat 3-year / 36,000-mile threshold for every make. That falsely told 4-year-old Hyundai, Kia, Genesis, and Mitsubishi owners (actual 5yr/60k) and Tesla owners (4yr/50k) that their factory coverage had expired, which inflated risk exposure and biased the verdict toward the paid affiliate. Calculator now uses make-specific periods; reasoning copy adapts to the actual number.
Vehicle hubs with fewer than 10 NHTSA complaints (~1,800 pages) were rendering "Above-average reliability for the segment" based on the calculated score. A 2024 Tesla with three complaints is a small sample, not above-average reliability. The score still displays, but the explanatory note now admits the data is provisional and points to active recalls where they exist. Threshold of 10 matches NHTSA's own minimum-sample disclaimers.
Vehicles with zero NHTSA complaints were rendering an empty AggregateRating in their JSON-LD, which Google's rich-result validator flagged. Vehicle nodes with 0 complaints but active recalls now aggregate from the recall record; vehicles with no data at all emit a single neutral rating point flagged as "limited NHTSA data." The on-page gauge value is unchanged — this only affects what search engines see.
When a reader flags an error or our internal review catches something: