robic.ca — AI Site Grade

Analysis

Cloudflare silently blocks GPTBot, ClaudeBot, and Google-Extended with HTTP 520 while serving full content to Perplexity and ChatGPT — and the site has no robots.txt rules, no llms.txt, and no FAQ schema on a page that explicitly answers 20+ patent questions.

Crawler Access

The site runs on Cloudflare (DNS via jasper.ns.cloudflare.com / michelle.ns.cloudflare.com, A records at 104.26.4.x). A compare_bot_access test on the homepage reveals a stark split: GPTBot, ClaudeBot, Google-Extended, Applebot-Extended, Bytespider, and anthropic-ai all receive HTTP 520 (Cloudflare's generic "unknown error") with a 7 KB error page. Meanwhile, ChatGPT-User, OAI-SearchBot, PerplexityBot, and Perplexity-User get a full 465 KB 200 response identical to the browser baseline. The /robots.txt file returns 200 with an empty body — no directives at all, not even a Sitemap declaration. The /llms.txt URL redirects to the homepage (effectively a 404 for the AI content-map convention). The sitemap index exists and is well-structured (8 sub-sitemaps, 1,000+ URLs), but the empty robots.txt means crawlers are not being pointed to it.

Cold-Knowledge Gap

A frontier LLM queried cold describes ROBIC as a "full-service IP law firm" with offices in Montreal, Quebec City, and Toronto, and mentions an "ROBIC IP Search" platform. The actual site lists only Montreal and Quebec City offices — no Toronto office exists on the site. The "ROBIC IP Search" platform is not findable on the domain; the client portal at portailclient.robic.ca is the only login surface. The LLM also describes the firm as serving "individual inventors to multinationals," which matches the site's positioning, but the cold knowledge omits that ROBIC is part of the IPH Limited group (an Australian publicly traded IP services company) — a fact the site discloses in its footer. The site's actual tagline ("La reference en propriete intellectuelle depuis 1892") and its claim of being "the 4th largest trademark filer in Canada" are absent from the model's prior.

Schema Posture

The homepage carries a well-formed Organization schema with logo, LinkedIn sameAs, and SearchAction. However, the FAQ page (/foire-aux-questions/) — which contains 20+ detailed Q&A pairs about patent law — has no FAQPage schema. It uses plain WebPage schema only. This is a significant missed signal: the page is a dense, authoritative FAQ that AI crawlers (those not blocked) parse as plain text rather than structured Q&A. Professional profile pages (e.g., /professionnel/samuel-ross/) also lack Person schema — they use only WebPage. The news listing page correctly uses CollectionPage.

Content Freshness and Depth

The site is actively maintained: the news section shows articles from April 2026 (e.g., Benchmark Litigation Canada 2026 recognition), and the Perspectives blog covers timely topics like AI-generated content copyright, Cineplex pricing appeals, and mRNA vaccine patent trends. The sitemap reveals an archive stretching back to 2013, with bilingual French/English versions of every article. The homepage word count is thin (~473 words) — mostly navigation and taglines — but the FAQ page is substantial (~2,690 words of readable patent guidance). The site uses a WordPress stack (evident from wp-content paths and Gravity Forms scripts).

Findings

1. Cloudflare blocks GPTBot, ClaudeBot, Google-Extended with HTTP 520 High

   Major AI crawlers (GPTBot, ClaudeBot, Google-Extended, Applebot-Extended, Bytespider, anthropic-ai) receive HTTP 520 errors from Cloudflare, while ChatGPT-User, OAI-SearchBot, PerplexityBot, and Perplexity-User get full 200 responses. This selective blocking prevents key AI services from indexing the site.

   What to change: Configure Cloudflare WAF to allow GPTBot, ClaudeBot, and Google-Extended by user-agent or IP range, or disable the blocking rules that trigger HTTP 520 for these bots.

2. Empty robots.txt provides no crawl directives High

   The robots.txt file at /robots.txt returns HTTP 200 with an empty body, offering no directives for any crawler. This means crawlers are not pointed to the sitemap and have no guidance on allowed or disallowed paths.

   What to change: Create a robots.txt file that declares the sitemap location and sets appropriate crawl rules for all bots.

3. Missing llms.txt file for AI content discovery Medium

   The /llms.txt URL redirects to the homepage, effectively providing no AI-specific content map. This convention helps AI crawlers discover key pages and context, but is absent here.

   What to change: Publish an llms.txt file listing important pages (FAQ, news, perspectives, professional profiles) to guide AI crawlers.

4. FAQ page lacks FAQPage schema markup High

   The FAQ page (/foire-aux-questions/) contains over 20 detailed Q&A pairs about patent law but uses only WebPage schema. Adding FAQPage schema would enable rich results in search and structured consumption by AI crawlers.

   What to change: Add FAQPage structured data to the FAQ page, marking each question and answer pair.

5. Professional profile pages lack Person schema Medium

   Individual lawyer profile pages (e.g., /professionnel/samuel-ross/) use only WebPage schema and omit Person schema, which would help AI systems understand attorney credentials and expertise.

   What to change: Add Person schema to each professional profile page with name, job title, description, and image.

6. Cold LLM knowledge includes incorrect office locations and missing IPH Group affiliation Medium

   A frontier LLM's prior knowledge about ROBIC lists a Toronto office that does not exist on the site, and omits the firm's affiliation with IPH Limited. The firm's tagline and trademark filing rank are also absent from the model's knowledge.

   What to change: Ensure accurate and prominent placement of office locations, IPH Group affiliation, and key differentiators on the site to improve LLM grounding.

7. Homepage content is thin at 473 words Low

   The homepage contains only 473 words, mostly navigation and taglines, providing limited substantive content for AI crawlers to index and understand the firm's value proposition.

   What to change: Expand the homepage with more descriptive content about services, expertise, and history.

8. Web searches for ROBIC return zero results Medium

   Multiple web searches for ROBIC-related queries returned zero results, indicating low external visibility and citation signals that could affect AI knowledge retrieval.

   What to change: Investigate and improve SEO and external citation profile to increase web presence.

What's working

• Sitemap index with 8 sub-sitemaps and 1,000+ URLs — The sitemap index is well-structured, containing 8 sub-sitemaps and over 1,000 URLs, covering the site's content comprehensively.
• News and blog content updated through April 2026 — The news section and Perspectives blog are actively maintained with recent articles, demonstrating content freshness that AI crawlers value.
• Organization schema with logo, LinkedIn, and SearchAction on homepage — The homepage includes a well-formed Organization schema with logo, LinkedIn sameAs, and SearchAction, providing structured identity data to crawlers.
• Bilingual French/English content archive since 2013 — The site maintains a bilingual archive of articles dating back to 2013, offering rich historical content for AI training and retrieval.
• FAQ page with 2,690 words of detailed patent guidance — The FAQ page contains substantial, readable patent guidance (2,690 words) that is valuable for AI knowledge extraction, despite missing schema.
• CollectionPage schema on news listing page — The news listing page correctly uses CollectionPage schema, aiding crawlers in understanding the page structure.