tmdfriction.com — AI Site Grade

Analysis

TMD Friction has an llms.txt (rare), full bot access for most AI crawlers, but the cold LLM knowledge contains a phantom class-action lawsuit that doesn't exist in any search results. The site also has a 404 at the standard /sitemap.xml despite having a full sitemap index at /en/sitemap_index.xml. Let me write the audit.

TMD Friction — AI-Visibility Audit

The cold LLM knowledge about TMD Friction contains a fabricated class-action lawsuit that does not appear in any web search results, creating a false reputational signal that AI engines may surface without live retrieval. Meanwhile, the site itself is technically well-prepared for AI crawlers but has a fragmented sitemap structure and zero FAQ or comparison schema that would help AI engines answer user queries directly.

Crawler Access

All major AI bots — GPTBot, OAI-SearchBot, ChatGPT-User, Google-Extended, PerplexityBot, Perplexity-User, Applebot-Extended — receive a full 200 response with identical content (189,669 bytes) to the browser baseline. No UA-based blocking, no JS shell, no Cloudflare challenge. ClaudeBot and anthropic-ai return connection errors (status 0), likely a network-level timeout rather than an explicit block. The robots.txt has no AI-specific directives — only a single User-Agent: * rule blocking /wp-admin/ and /wp-includes/. The site runs on nginx with a strong CSP and HSTS, hosted on a Hetzner IP (94.130.244.212).

llms.txt and Sitemap Posture

The site has a functional /llms.txt — rare for a B2B industrial manufacturer — generated by Yoast SEO v27.6, listing pages, media, careers, and locations. However, the standard /sitemap.xml returns 404, while the actual sitemap index lives at /en/sitemap_index.xml (referencing 5 sub-sitemaps with ~481 total URLs). The robots.txt correctly points to the sub-sitemaps directly (page-sitemap.xml, insights-sitemap.xml, etc.), but any crawler or tool that checks only /sitemap.xml finds nothing.

Cold-Knowledge Gap

The LLM knows TMD Friction as a global brake friction manufacturer with brands Textar, Mintex, Don, and Pagid, supplying BMW, VW, and Ford. It correctly states the company has R&D centers in Germany and the UK and a history over 100 years. The critical gap: the LLM asserts a "2023 class-action lawsuit in the U.S. over alleged defective brake pads" — no evidence of this exists across multiple web searches. This hallucination, if ingested by downstream AI systems without retrieval, could damage brand perception. The LLM also does not know about the AEQUITA acquisition (announced August 2023, closed December 2023), which is the most significant corporate event in the company's recent history and is well-documented on the site.

Schema and Content Signals

Every page uses standard WebPage, BreadcrumbList, and WebSite JSON-LD with SearchAction. No Organization, Product, FAQPage, or HowTo schema exists anywhere on the site — surprising for a manufacturer with seven distinct aftermarket brands and a racing division. The homepage has no FAQ, no comparison tables, and no structured data for the brand portfolio. The site does have a strong heading hierarchy (H1 → H2 → H3) and rich visible text (~634 words on the homepage), but the absence of answer-format signals means AI engines cannot easily extract product comparisons or specifications without crawling deep pages.

External Signals

External links are limited to LinkedIn and YouTube. No press coverage, Reddit threads, or review sites surfaced in searches. The site references a dedicated history microsite (thehistory.tmdfriction.com) and a partnership with Iberdrola for green energy, but these are not linked from external sources in a way that builds off-domain citation mass.

Findings

1. Cold LLM knowledge contains fabricated class-action lawsuit High

   The LLM asserts a 2023 class-action lawsuit in the U.S. over alleged defective brake pads, but no evidence exists across multiple web searches. This hallucination could damage brand perception if surfaced by AI systems without live retrieval.

   What to change: Publish authoritative content (e.g., a press release or FAQ) explicitly denying the lawsuit and highlighting the company's legal compliance. Monitor LLM outputs and consider submitting corrections to training data sources.

2. AEQUITA acquisition absent from cold LLM knowledge Medium

   The LLM does not know about the AEQUITA acquisition (announced August 2023, closed December 2023), which is the most significant recent corporate event and is well-documented on the site.

   What to change: Ensure the acquisition press release is prominently linked from the homepage and included in the llms.txt. Consider adding structured data (e.g., NewsArticle) to the press release.

3. Standard /sitemap.xml returns 404 Medium

   The standard sitemap URL returns a 404 error, while the actual sitemap index lives at /en/sitemap_index.xml. This confuses crawlers that only check the default location.

   What to change: Redirect /sitemap.xml to /en/sitemap_index.xml or create a sitemap index at the root that points to the sub-sitemaps.

4. No FAQPage, Product, or Organization schema on any page High

   The site lacks FAQPage, Product, Organization, and HowTo schema, which are critical for AI engines to answer user queries directly. This limits the site's ability to appear in rich results and AI-generated answers.

   What to change: Add Organization schema with logo and brand information. Add FAQPage schema to pages with common questions. Add Product schema for each brake brand (Textar, Mintex, etc.).

5. ClaudeBot and anthropic-ai receive connection errors Medium

   ClaudeBot and anthropic-ai return status 0 (connection error), likely a network-level timeout. This prevents Claude from accessing the site.

   What to change: Investigate network or firewall rules that may be blocking Anthropic's IP ranges. Ensure no rate-limiting or timeout is triggered for these bots.

6. Limited external backlinks and off-domain citations Low

   External links are limited to LinkedIn and YouTube. No press coverage, Reddit threads, or review sites surfaced in searches, reducing off-domain citation mass that AI engines use for authority.

   What to change: Develop a PR strategy to earn mentions on industry blogs, news sites, and forums. Encourage customer reviews on third-party platforms.

7. Homepage lacks comparison tables or brand portfolio overview Low

   The homepage has no FAQ, comparison tables, or structured data for the brand portfolio, making it harder for AI engines to extract product comparisons without deep crawling.

   What to change: Add a brand portfolio section with comparison tables and FAQ schema to help AI engines answer user queries directly.

What's working

• Functional /llms.txt file generated by Yoast SEO — The site has a rare and functional llms.txt listing pages, media, careers, and locations, which helps AI crawlers discover key content.
• Full access for GPTBot, OAI-SearchBot, Google-Extended, and others — All major AI bots receive a 200 response with identical content to the browser baseline, with no UA-based blocking or JS shells.
• robots.txt has no AI-specific blocking directives — The robots.txt only blocks /wp-admin/ and /wp-includes/ for all user agents, allowing AI crawlers full access to the rest of the site.
• Strong heading hierarchy and rich visible text — Pages use a clear H1→H2→H3 hierarchy with substantial visible text (e.g., 634 words on homepage), aiding content extraction by AI engines.
• WebPage and BreadcrumbList JSON-LD on every page — Every page includes standard WebPage, BreadcrumbList, and WebSite JSON-LD with SearchAction, providing basic structured data for search engines.
• Sitemap index at /en/sitemap_index.xml with 481 URLs — The site has a comprehensive sitemap index with 5 sub-sitemaps covering ~481 URLs, ensuring good crawl coverage.
• robots.txt correctly points to sub-sitemaps — The robots.txt directly references page-sitemap.xml and insights-sitemap.xml, helping crawlers find sitemaps even if the root sitemap is missing.
• Strong CSP and HSTS headers for security — The site uses a strong Content Security Policy and HSTS, which indirectly supports trust signals for AI crawlers.