enterprisecanada.com — AI Site Grade

Analysis

Cold-knowledge hallucination gap: the LLM fabricates a "Public Affairs Barometer," a "Crisis Communications Playbook," a COVID-19 contract controversy, and claims the firm was "founded by former political staffers" — none of which appear anywhere on the actual site.

Crawler Access

All major AI crawlers — GPTBot, ClaudeBot, PerplexityBot, Google-Extended, OAI-SearchBot, ChatGPT-User, anthropic-ai — receive a 200 status with the full 90KB HTML payload, identical to a browser. Only Bytespider is blocked (403). The robots.txt is a bare Yoast default (User-agent: * Disallow:) with no AI-bot rules whatsoever. The llms.txt returns 404. The site runs on Media Temple hosting (ns1/ns2.mediatemple.net) with no visible CDN or WAF — the CSP and HSTS headers are present but no Cloudflare or Akamai edge. The homepage delivers only 75 words of visible text from a plain GET (a JavaScript shell), yet the raw HTML contains the full content for crawlers that parse server-rendered markup. This creates a bifurcated experience: AI crawlers see the full site, but any bot that executes JS (or a headless browser) would see almost nothing.

Content & Schema Posture

The site uses WordPress with Yoast SEO and outputs Organization + WebSite + WebPage + BreadcrumbList JSON-LD on every page. However, the Organization schema is thin: it lists only two sameAs URLs (X/Twitter and LinkedIn), omitting any description, foundingDate, numberOfEmployees, areaServed, or knowsAbout properties. The WebSite schema has an empty description field. No LocalBusiness, ProfessionalService, or FAQPage schema types are used. The site has no FAQ pages, no comparison tables, and no structured data for its case studies — the five case studies (Conceivable Dreams, Casey House, AMHO, 9-8-8, North York General) use only generic Article schema. The /case-studies/ listing page redirects to /what-we-do/, meaning there is no discoverable case study index for crawlers.

Cold-Knowledge Gap

A frontier LLM queried cold about Enterprise Canada produced four fabricated claims: that the firm publishes a "Public Affairs Barometer" quarterly survey and a "Crisis Communications Playbook" (neither exists on the site), that it was "founded by former political staffers" (the site says CEO Barbara Fox has been there 20+ years but does not state founding origin), and that it was involved in a "2023 controversy over a contract with the Ontario government related to COVID-19 messaging" (no evidence of this on the site or in web search results). The LLM correctly identified the firm as a Canadian public affairs and strategic communications firm with offices in Toronto, Ottawa, and Calgary — but the site actually lists eight offices (Vancouver, Calgary, Edmonton, Hamilton, Toronto, Ottawa, Halifax, St. John's). The cold knowledge is stale and partially hallucinated, missing the firm's actual breadth.

External Signals

Web search returns zero indexed results for the domain name, the firm name with "public affairs," or the CEO's name — a striking absence for a 40-year-old firm that claims to work with "Canada's largest companies." The site's only external backlink footprint visible on-page is to Worldcom Public Relations Group (a global partnership) and its sister agency Creative Currency. No press mentions, no Reddit threads, no review sites surfaced. The Twitter/X account @EntCanada is linked but its content was not verified. This near-total external signal vacuum means AI models have almost no third-party corroboration to draw on, making them more likely to hallucinate or rely on stale training data.

Surprising Details

The homepage dateModified is set to 2026-05-22 — a future date from the perspective of the training data — which may confuse temporal reasoning in AI models. The who-we-are page lists 70+ team members with detailed bios, making it one of the most content-rich pages on the site, yet it has no structured Person schema for any individual. The goodspeak newsletter archive contains 27 monthly editions dating back to March 2024, a strong content asset that is entirely absent from the sitemap's primary post feed and has no dedicated schema markup. The careers page states "No listings at this time" — a static dead end.

Findings

1. LLM cold knowledge fabricates products and controversies High

   A frontier LLM queried cold about Enterprise Canada fabricated a 'Public Affairs Barometer' survey, a 'Crisis Communications Playbook', a founding by former political staffers, and a COVID-19 contract controversy. None of these appear on the site or in web search results. The LLM also underestimated the firm's office count (3 vs. 8).

   What to change: Publish an llms.txt file listing accurate firm facts, products, and office locations. Add structured data (ProfessionalService, LocalBusiness) with complete details to ground AI knowledge.

2. Zero indexed search results for the firm High

   Web searches for 'Enterprise Canada public affairs', the domain name, and the CEO's name returned zero results. This absence of third-party corroboration increases reliance on stale or hallucinated training data.

   What to change: Implement a link-building and PR strategy to generate authoritative external mentions. Ensure the site is properly indexed by Google and Bing.

3. Organization schema lacks key properties Medium

   The Organization JSON-LD on every page includes only two sameAs URLs (X/Twitter, LinkedIn) and omits description, foundingDate, numberOfEmployees, areaServed, and knowsAbout. The WebSite schema has an empty description field.

   What to change: Expand the Organization schema to include description, foundingDate, numberOfEmployees, areaServed, and knowsAbout. Populate the WebSite description.

4. Team bios lack Person schema markup Medium

   The who-we-are page lists 70+ team members with detailed bios but uses no structured Person schema for any individual, reducing discoverability in AI knowledge graphs.

   What to change: Add Person schema markup to each team member bio, including name, jobTitle, description, and sameAs links.

5. Case studies listing redirects and lacks schema Medium

   The /case-studies/ page redirects to /what-we-do/, so there is no discoverable case study index. Individual case studies use only generic Article schema instead of more specific types like MedicalWebPage or CreativeWork.

   What to change: Restore a dedicated case studies index page and add appropriate schema types (e.g., MedicalWebPage for health case studies) to each case study.

6. llms.txt file missing Medium

   The site returns a 404 for /llms.txt, missing an opportunity to provide AI crawlers with a curated set of facts and resources.

   What to change: Create an llms.txt file listing key pages, firm description, office locations, and services.

7. Homepage dateModified set to future date Low

   The homepage's dateModified is set to 2026-05-22, a future date that may confuse AI models' temporal reasoning.

   What to change: Update the dateModified to the actual last modification date.

8. Goodspeak newsletter archive missing from sitemap Low

   The goodspeak newsletter archive contains 27 monthly editions but is not included in the primary sitemap post feed, reducing crawl priority.

   What to change: Add goodspeak archive pages to the sitemap.

9. Careers page shows no listings Low

   The careers page displays 'No listings at this time' with no dynamic content or structured data, offering no value to crawlers or job seekers.

   What to change: Remove the careers page or replace it with a redirect to a job board.

What's working

• All major AI crawlers receive full HTML content — GPTBot, ClaudeBot, PerplexityBot, Google-Extended, and others receive a 200 status with the full 90KB HTML payload, identical to a browser. No AI bots are blocked except Bytespider.
• JSON-LD schema present on all pages — Every page includes Organization, WebSite, WebPage, and BreadcrumbList JSON-LD markup via Yoast SEO, providing a baseline structured data foundation.
• 70+ team member bios with rich content — The who-we-are page contains detailed biographies for over 70 team members, offering substantial text content for AI models to extract information.
• 27 monthly newsletter editions available — The goodspeak archive contains 27 monthly editions dating back to March 2024, providing a rich, regularly updated content asset.
• Sitemap index with 80 URLs — The sitemap index contains 80 URLs, covering most site pages and ensuring crawlers can discover content.
• Robots.txt allows all crawlers — The robots.txt file has a single User-agent: * with Disallow: (empty), meaning no restrictions on any crawler.