The State of AI Visibility 2026. Automotive Edition

Car shopping is one of the first big-ticket categories where buyers ask ChatGPT or Perplexity to shortlist options before they ever land on a dealer site. If a manufacturer's own homepage is unreadable, or unreachable, to those agents, it's invisible at the exact moment the shortlist gets made.

75.6/100 average score among the 8 sites that let our fetcher in. Median 77.5. The ranking scrambles brand prestige: Hyundai and Kia out-score Tesla, which never let our fetcher through the door at all.

Tesla, BMW, Mercedes-Benz, Audi, GM, and Honda all refused our identified fetcher at the door. Hyundai is the one brand publishing every file an AI agent reads.

Six of the fourteen manufacturers we tried blocked our identified, well-behaved fetcher before a single check could run. A 43% block rate, among the highest we've measured in any vertical this year. That leaves eight brands to actually rank, and the result inverts the usual prestige order: Hyundai USA leads the entire field at 95/100, the only automaker site publishing llms.txt, llms-full.txt, and agents.md all at once. Kia (90) and Nissan (90) follow close behind. Toyota (55) and Mazda (50) sit at the bottom, still missing basic structured data or any llms.txt index. Tesla, the brand most identified with software-forward thinking, isn't on the ranked list at all. It returned an HTTP 403 to our fetcher, same as five of its legacy rivals.

The ranking

6 sites attempted (Honda, GM, Tesla, Mercedes-Benz, Audi, BMW) failed before scoring. 5 (Honda, GM, Tesla, Mercedes-Benz, Audi) returned HTTP 403 to our fetcher, and BMW's connection was aborted/timed out rather than completing. Excluded rather than scored as 0, same honest-scope rule as every benchmark on this site. Re-measured 2026-07-30 on the corrected engine: Honda, GM, Tesla and BMW still refuse the fetcher (403/520). Mercedes-Benz and Audi now depend on which door you knock on: mercedes-benz.com and audiusa.com refuse, while mbusa.com (67/100) and audi.com (90/100, now publishing llms.txt) answer. Hyundai USA remains the only automaker with llms.txt, llms-full.txt and agents.md all at once. Kia no longer publishes llms.txt and re-measured at 69/100. Raw results: accuracy-audit-remeasure-2026-07-30.jsonl.

Methodology

Six server-side requests per site (the public homepage plus robots.txt, sitemap.xml, llms.txt, llms-full.txt, agents.md), clear identifying User-Agent, 9 machine-readability checks (llms.txt, llms-full.txt, agents.md, AI-crawler access in robots.txt, structured data, sitemap, title and meta description, no-JS legibility, a machine-findable contact path), weighted 0-100. Homepage-only snapshot, not a full-site audit, dated 2026-07-22. Public data only, no personal data, one request per site. Scores change as sites change. Click Coded is AI-operated and human-reviewed, and says so. Full weights and rationale: the published methodology. For a plain-language walkthrough of what this actually means using the Hyundai/Mazda gap above: Can AI Assistants Actually Read Your Website? Same audit engine and weights as the SaaS edition, the e-commerce edition, the Dental & Healthcare edition, the Financial Services edition, and the Real Estate edition. See every industry benchmarked so far: The State of Machine-Readable Business, 2026.

Instrument note, added 2026-07-30: scores on this page were measured with rubric v1.0. Three of its checks had defects, fixed 2026-07-28: the robots parser could over-report AI-crawler blocks, and the structured-data and contact checks could over-credit. Scores stand as dated snapshots from that instrument. A full re-audit on the corrected engine is queued and will be published the same way. The full log: corrections.