grantthornton.de — AI Site Grade

Analysis

Grant Thornton Germany: AI-Visibility Audit

The site has an exemplary llms.txt in both German and English — one of the most complete seen — yet the homepage carries only a single WebPage schema type, no Organization, ProfessionalService, or FAQPage markup, creating a structural gap between what AI crawlers can read and what they can semantically understand.

Crawler Access

All major AI bots — GPTBot, ClaudeBot, PerplexityBot, Google-Extended, OAI-SearchBot, Bytespider, Applebot-Extended — receive a full 200 response with identical byte payload (650 KB) to a browser baseline. No UA-based blocking exists. The site runs on Cloudflare with HSTS and a strict CSP, but none of these security measures impede AI crawlers. The robots.txt is minimal: a catch-all Allow: / with only /search/ and /*GetVCard disallowed. No AI-specific directives exist — no Disallow: GPTBot, no crawl-delay. The llms.txt at both /llms.txt and /en/llms.txt is a standout: it lists all service lines, industries, locations, and key pages with descriptions, making this one of the most AI-friendly content maps in the professional services sector.

Cold-Knowledge Gap

The LLM's cold knowledge about Grant Thornton Germany is dominated by the Wirecard scandal — the model recalls that the firm faced scrutiny as auditor for Wirecard's German subsidiary, leading to regulatory investigations and reputational damage. The site itself contains zero mentions of Wirecard. A site search for "Wirecard" returns no results. The newsroom covers revenue growth (EUR 264 million in FY 2024/25), a new CFO/COO, and a strategic partnership with Cinven, but makes no reference to the scandal. This creates a stark asymmetry: AI models that retrieve the site will find a clean, forward-looking narrative of Mittelstand advisory excellence, while models relying on training data will surface the Wirecard association. The site also claims "around 2,000 employees at 10 locations" while the llms.txt says "1,900 employees at 11 locations" — a minor but detectable inconsistency across its own files.

Schema Posture

The schema implementation is thin. Every page examined carries only WebPage and sometimes BreadcrumbList — no Organization (with logo, founding date, revenue, employee count), no ProfessionalService, no FAQPage despite the homepage containing FAQ-style content. The data-facts page states "Founded in 1958, EUR 264 million revenue, 173 Partners, 441 Professionals" — all of which should be in structured data but are only in visible text. The homepage's og:site_name reads "Deutschland - Grant Thornton" while the English version reads "Grant Thornton AG" — a minor brand inconsistency across language versions.

External Signals

The newsroom shows three press releases from 2025-2026, including a revenue milestone and a strategic partnership with private equity firm Cinven. External search results for "Grant Thornton Germany Wirecard" returned zero DuckDuckGo results, suggesting either low current search visibility or regional filtering. The site links to LinkedIn (2,900+ followers), YouTube, Instagram, and Xing. The DNS uses CSC (Corporation Service Company) for nameservers and Mimecast for email security — a typical enterprise stack.

Findings

1. No Organization or ProfessionalService schema on any page High

   Every page examined carries only WebPage and sometimes BreadcrumbList schema. Key entity data (founding year, revenue, employee count, partners) is present in visible text but absent from structured data, limiting AI understanding.

   What to change: Add Organization and ProfessionalService schema markup to the homepage and about pages, including founding date, revenue, employee count, and partner numbers.

2. Site omits Wirecard scandal while LLMs recall it prominently High

   LLM cold knowledge associates Grant Thornton Germany with the Wirecard scandal, but the site contains zero mentions of Wirecard. This creates a disconnect: AI models retrieving the site see a clean narrative, while training-data-based answers surface the scandal.

   What to change: Publish a dedicated page or statement addressing the Wirecard matter to align the site narrative with external knowledge.

3. Employee count differs between llms.txt and data-facts page Medium

   The llms.txt states '1,900 employees at 11 locations' while the data-facts page says 'around 2,000 employees at 10 locations'. This inconsistency may confuse AI crawlers and reduce trust.

   What to change: Reconcile the employee and location numbers across llms.txt and the data-facts page to a single consistent figure.

4. Homepage FAQ content lacks FAQPage schema Medium

   The homepage contains FAQ-style content but no FAQPage structured data, missing an opportunity for rich results in AI and search.

   What to change: Add FAQPage schema markup to the homepage FAQ section.

5. og:site_name differs between German and English pages Low

   The German homepage uses 'Deutschland - Grant Thornton' while the English version uses 'Grant Thornton AG', creating brand inconsistency across language versions.

   What to change: Standardize og:site_name to a single brand name across all language versions.

6. robots.txt lacks AI-specific directives or crawl-delay Low

   The robots.txt is minimal with only a catch-all Allow and two disallowed paths. No AI-specific directives (e.g., Disallow for GPTBot) or crawl-delay are set, which is fine but leaves no room for fine-tuning.

   What to change: Consider adding crawl-delay or AI-specific directives if needed in the future.

What's working

• Comprehensive llms.txt in both German and English — The site provides a detailed llms.txt listing all service lines, industries, locations, and key pages with descriptions, making it one of the most AI-friendly content maps in the professional services sector.
• All major AI bots receive full 200 responses — GPTBot, ClaudeBot, PerplexityBot, Google-Extended, and others all get the same byte payload as a browser, with no UA-based blocking.
• Active newsroom with recent press releases — The newsroom features three press releases from 2025-2026, including revenue milestones and strategic partnerships, providing fresh content for AI crawlers.
• Social media profiles linked on site — The site links to LinkedIn (2,900+ followers), YouTube, Instagram, and Xing, supporting external signal generation.
• Cloudflare with HSTS and CSP provides security without blocking bots — The site uses Cloudflare, HSTS, and a strict CSP, but these measures do not impede AI crawlers.