turcanconnell.com — AI Site Grade

Analysis

Turcan Connell's site serves full HTML to GPTBot and Google-Extended but returns 403 Forbidden to ClaudeBot and Bytespider — a selective AI-crawler blockade with no corresponding robots.txt rules, creating an opaque visibility gap for Anthropic and ByteDance models.

Crawler Access

robots.txt at https://www.turcanconnell.com/robots.txt contains a single User-agent: * rule disallowing only /cpresources/, /vendor/, /.env, and /cache/. No AI-specific directives exist for any crawler. Despite this, compare_bot_access on the homepage reveals ClaudeBot and Bytespider receive HTTP 403 while GPTBot, OAI-SearchBot, ChatGPT-User, Google-Extended, PerplexityBot, Perplexity-User, Applebot-Extended, and anthropic-ai all return 200 with identical byte-size content to a browser. The blocking is server-side (nginx, hosted at DigitalOcean on IP 138.68.147.253) and not declared in any standard crawler protocol. No llms.txt exists (404).

Cold-Knowledge Gap

When queried cold, the LLM described Turcan Connell as operating from a "single Edinburgh office" with a "Turcan Connell Asset Management" arm offering investment management. Neither claim matches the live site. The site lists three offices (Edinburgh, Glasgow, London) and makes zero mention of asset management, wealth management, or investment services across any fetched page. The LLM also stated the firm was founded in 1997 — the site does not state a founding year anywhere on the homepage, careers page, or CSR page. The site describes a 200-strong team serving wealthy individuals, landowners, entrepreneurs, and charities across legal and tax services only.

Schema Posture

Every page uses WebPage + BreadcrumbList + Organization schema via SEOmatic. However, the Organization schema is minimal — no @id beyond #identity and #creator, no logo, sameAs, address, telephone, foundingDate, numberOfEmployees, or areaServed properties. The 41+ staff profile pages (e.g., gillian-crandles) use WebPage schema but not Person schema, meaning AI engines cannot extract structured data about individual lawyers' roles, specialisms, or contact details. No LocalBusiness or LegalService schema type is used anywhere.

Content Signals

The site runs on Craft CMS 3 with SEOmatic, serves server-rendered HTML (no JS-rendering risk), and publishes regular news articles (most recent dated April 2026). Content is well-written but lacks FAQ, comparison tables, or structured answer-format signals that AI engines use for featured snippets. The homepage has no H1 tag — only H2 and H3 headings. The llms.txt 404 means no AI-friendly content map exists. The humans.txt file exists but is mostly empty (no creator name or URL filled in).

Findings

1. ClaudeBot and Bytespider receive 403 Forbidden High

   ClaudeBot and Bytespider are blocked server-side with HTTP 403 on the homepage, while GPTBot, Google-Extended, and other AI crawlers get full HTML. No robots.txt rules declare this block, making it opaque and unreviewable.

   What to change: Remove the server-side block for ClaudeBot and Bytespider, or declare the restriction in robots.txt with a clear disallow directive.

2. No llms.txt file published Medium

   The site returns 404 for llms.txt, meaning no AI-friendly content map is available for LLMs to discover key pages or context.

   What to change: Create an llms.txt file listing important pages (services, offices, team) to guide AI crawlers.

3. LLM cold knowledge fabricates asset management and single office High

   When queried cold, an LLM described Turcan Connell as having a single Edinburgh office and an asset management arm, but the site lists three offices and makes no mention of asset management. The founding year (1997) is also unverified on the site.

   What to change: Add a clear 'About' page with founding year, office locations, and service scope. Include structured data (Organization schema with foundingDate, address, areaServed) to correct LLM misconceptions.

4. Organization schema lacks key properties Medium

   The Organization schema on every page is minimal, missing logo, sameAs, address, telephone, foundingDate, numberOfEmployees, and areaServed. This limits AI engines' ability to extract firm details.

   What to change: Expand the Organization schema to include logo, sameAs (LinkedIn, etc.), address, telephone, foundingDate, numberOfEmployees, and areaServed.

5. Staff profile pages lack Person schema High

   Over 41 staff profile pages (e.g., gillian-crandles) use WebPage schema but not Person schema, so AI engines cannot extract structured data about individual lawyers' roles, specialisms, or contact details.

   What to change: Add Person schema to each staff profile page with properties for name, jobTitle, description, telephone, email, and knowsAbout (practice areas).

6. No LocalBusiness or LegalService schema used Medium

   The site uses only WebPage, BreadcrumbList, and Organization schema. No LocalBusiness or LegalService schema is present, which would help AI engines classify the firm as a legal service provider.

   What to change: Add LocalBusiness or LegalService schema (with @type LegalService) to the homepage and contact page, including address, telephone, and areaServed.

7. Homepage has no H1 heading Medium

   The homepage lacks an H1 tag, using only H2 and H3 headings. This weakens semantic structure for search engines and AI crawlers.

   What to change: Add a descriptive H1 heading to the homepage, such as 'Turcan Connell | Scottish Private Client Law Firm'.

8. No FAQ or structured answer-format content Low

   The site lacks FAQ pages, comparison tables, or other structured answer formats that AI engines use for featured snippets and direct answers.

   What to change: Create FAQ pages for common legal questions (e.g., 'What is estate planning?') with FAQPage schema.

9. humans.txt file is mostly empty Low

   The humans.txt file exists but contains no creator name or URL, providing no useful attribution or context.

   What to change: Populate humans.txt with the site creator/team name and relevant URLs.

10. No AI-specific directives in robots.txt Medium

    The robots.txt file has no rules for AI crawlers like GPTBot, ClaudeBot, or Google-Extended, leaving their access entirely to server-level rules.

    What to change: Add explicit allow/disallow rules for major AI crawlers in robots.txt to align with server behavior.

What's working

• Server-rendered HTML accessible to most AI crawlers — The site serves full server-rendered HTML to GPTBot, Google-Extended, and other major AI crawlers, ensuring content is indexable without JavaScript rendering issues.
• Regular news articles published — The site publishes news articles (most recent April 2026), providing fresh content that AI crawlers can index for updates.
• BreadcrumbList schema on every page — Every page includes BreadcrumbList schema, helping AI engines understand site hierarchy and navigation paths.
• SEOmatic plugin provides structured data foundation — The site uses Craft CMS's SEOmatic plugin, which automatically generates WebPage and Organization schema, providing a baseline for structured data.
• Sitemap with 41 URLs indexed — A sitemap is available and contains 41 URLs, helping crawlers discover all pages efficiently.