chs.ca — AI Site Grade

Analysis

The robots.txt explicitly blocks GPTBot, ChatGPT-User, ClaudeBot, anthropic-ai, and CCBot from the entire site — yet every single one of those bots receives a full 200 response with 133KB of content, identical to a browser, because the nginx server does not enforce the directives.

Crawler Access Contradiction

The robots.txt at chs.ca contains five AI-specific Disallow: / rules targeting GPTBot, ChatGPT-User, ClaudeBot, anthropic-ai, and CCBot. However, compare_bot_access on the homepage shows all five bots return HTTP 200 with 133,632 bytes — the same payload a browser receives. The nginx/1.31.0 server does not enforce UA-based blocking at the edge. The site also blocks Google-Extended, PerplexityBot, OAI-SearchBot, Applebot-Extended, and Bytespider nowhere in robots.txt, yet those too get full content. The llms.txt returns 403 Forbidden — a missed opportunity for AI-friendly content mapping. The sitemap contains 1,509 URLs and is accessible to all crawlers.

Cold-Knowledge Gap

A frontier LLM queried cold about Canadian Hearing Services recalled the organization's founding year (1940), its Ontario focus, and its interpreting services — but also surfaced a 2022 class-action lawsuit alleging systemic discrimination and harassment against Deaf staff. The site itself contains zero mention of this lawsuit anywhere on the homepage, about page, or news releases page. The most recent news release mentioning labour relations is "Canadian Hearing Services and CUPE Local 2073 Reach a New Collective Agreement" (January 2023), which frames workforce relations positively. The gap between the model's prior (which includes reputational risk) and the site's silence on the matter is the single largest AI-visibility vulnerability.

Schema and Structured Data Posture

Across the homepage, about page, blog, hearing clinics page, hearing aid services page, and FAQ page, zero JSON-LD schemas were detected. No Organization, LocalBusiness, FAQPage, Product, Service, or WebSite schema exists. The site runs Drupal 10 (x-generator: Drupal 10), which supports schema modules, but none are active. The FAQ page (/HearingAidFAQ) contains 10 expandable Q&A items with rich answer content — a prime candidate for FAQPage markup — yet renders as plain HTML with no structured data. The product pages redirect to 403 "Access denied" for browser UAs, suggesting a cart/checkout gating mechanism that also blocks AI crawlers from indexing product inventory.

External Signals and Reputation

The LLM's cold recall of a 2022 class-action lawsuit is the dominant external signal. Web searches for "Canadian Hearing Services class action lawsuit discrimination Deaf employees" returned zero results in the current index, suggesting the story has faded from surface-web visibility but persists in model training data. The site's news releases emphasize scholarships, research grants, and accreditation — all positive signals — but the absence of any acknowledgment of past controversy creates a reputation blind spot for AI engines that may surface the lawsuit in generative answers without the organization's counter-narrative.

Findings

1. Robots.txt blocks AI bots but nginx serves full content to all High

   The robots.txt disallows GPTBot, ChatGPT-User, ClaudeBot, anthropic-ai, and CCBot from the entire site, yet compare_bot_access shows all five receive HTTP 200 with full 133KB content. The nginx server does not enforce the directives, creating a contradiction that may confuse crawlers and waste crawl budget.

   What to change: Either remove the AI bot disallow rules from robots.txt or configure nginx to enforce them by returning 403 or 503 for those user-agents.

2. llms.txt returns 403 Forbidden Medium

   The llms.txt endpoint returns a 403 Forbidden, preventing AI crawlers from discovering a curated list of AI-friendly resources. This is a missed opportunity to guide LLMs to high-quality content.

   What to change: Create and serve an llms.txt file at the root with links to key pages like the about page, FAQ, and news releases.

3. Site omits any mention of 2022 class-action lawsuit recalled by LLMs High

   A frontier LLM queried cold about Canadian Hearing Services recalled a 2022 class-action lawsuit alleging systemic discrimination and harassment against Deaf staff. The site's homepage, about page, and news releases contain zero mention of this lawsuit, creating a reputational blind spot for AI-generated answers.

   What to change: Publish a dedicated page or statement addressing the lawsuit and the organization's response, and link it from relevant sections of the site.

4. Zero JSON-LD schemas across all key pages High

   No structured data (JSON-LD) was detected on the homepage, about page, blog, hearing clinics page, hearing aid services page, or FAQ page. This includes missing Organization, LocalBusiness, FAQPage, Product, Service, and WebSite schemas, which limits rich snippet eligibility and AI comprehension.

   What to change: Implement JSON-LD schemas for Organization, LocalBusiness, FAQPage (on /HearingAidFAQ), and WebSite using Drupal's schema modules or custom code.

5. FAQ page with 10 Q&A items lacks FAQPage schema Medium

   The /HearingAidFAQ page contains 10 expandable Q&A items with rich answer content, but no FAQPage structured data is present. This prevents AI crawlers from extracting the Q&A pairs for use in rich results or generative answers.

   What to change: Add FAQPage JSON-LD markup to the FAQ page, encoding each question and answer.

6. Product pages return 403 Access denied to all user-agents Medium

   The product page for a hearing aid amplifier returns HTTP 403 with 'Access denied' for browser UAs, indicating a gated checkout or cart system. This blocks AI crawlers from indexing product inventory and pricing, reducing visibility for product-related queries.

   What to change: Allow AI crawlers to access product pages with full content, or provide a separate crawlable product listing page.

7. Web searches for CHS reviews and lawsuit return zero results Medium

   Searches for 'Canadian Hearing Services review' and 'Canadian Hearing Services class action lawsuit discrimination Deaf employees' returned zero results in the current web index. This suggests the lawsuit story has faded from surface-web visibility, but it persists in LLM training data, creating a gap between online reputation and AI recall.

   What to change: Proactively publish content addressing the lawsuit and current workplace practices to influence AI training data and search results.

8. No Wayback Machine snapshot for homepage from 2023 Low

   A Wayback Machine lookup for the homepage in January 2023 returned no snapshot, indicating the site may have been inaccessible or not archived. This limits historical context for AI crawlers.

   What to change: Ensure the site is accessible to the Wayback Machine by not blocking archive.org bots in robots.txt.

What's working

• Sitemap with 1,509 URLs is accessible to all crawlers — The sitemap contains 1,509 URLs and is not blocked by robots.txt, allowing AI crawlers to discover the full site structure.
• Homepage provides 390 words of descriptive content — The homepage contains 390 words describing the organization's mission and services, giving AI crawlers substantive text to index.
• FAQ page has 833 words of detailed Q&A content — The /HearingAidFAQ page contains 833 words across 10 Q&A items with thorough answers, providing valuable information for AI crawlers despite lacking schema markup.
• News releases page with 527 words of recent updates — The /newsreleases page contains 527 words of recent news, including positive stories about scholarships and collective agreements, providing fresh content for AI crawlers.
• Hearing clinics page with 532 words of service details — The /hearing-clinics page provides 532 words describing clinic locations and services, offering useful content for local search and AI comprehension.
• Hearing aid services page with 1,192 words of detailed information — The /hearing-aid-services-and-products page contains 1,192 words covering services and products, providing substantial content for AI indexing.
• Site runs on Drupal 10 with schema module potential — The site uses Drupal 10, which has robust support for structured data modules, making it easier to implement JSON-LD schemas in the future.
• Google-Extended and other major AI bots are not blocked — The robots.txt does not block Google-Extended, PerplexityBot, OAI-SearchBot, Applebot-Extended, or Bytespider, allowing these crawlers full access to the site.