keplin-group.com — AI Site Grade

Analysis

Cold-Knowledge Catastrophe

The most critical finding is a total identity mismatch in AI training data. When queried cold, a frontier LLM describes Keplin Group as a US-based IT managed-services provider founded in 1999 in New Jersey — a complete hallucination. The actual Keplin Group is a UK-headquartered B2B consumer goods supplier (Blake House, Cowley Business Park, Uxbridge) that owns 9+ in-house brands across home, kitchen, garden, pet, and baby categories, supplying retailers including Morrisons, Aldi, Asda, Iceland, and SuperValu. The model has zero awareness of the real business.

Crawler Access

All major AI crawlers — GPTBot, ClaudeBot, PerplexityBot, Google-Extended, OAI-SearchBot, Bytespider, Applebot-Extended — receive full 200 responses with identical byte-size content from the B2B site at keplin-group.com. No UA-based blocking, no Cloudflare challenge, no JS shell. The robots.txt is permissive (User-agent: * Disallow: /_panel) but contains no AI-bot-specific directives at all. The llms.txt returns a 404 — the site has no AI-friendly content map. The consumer-facing site at keplin-online.com is blocked to all traffic (403 Cloudflare challenge), including every AI crawler and even browser UAs, making it invisible to both search engines and AI retrieval systems.

Schema and Structured Data

Every page examined — homepage, Our Brands, Our Products, FAQs, Stockists, Milestones, Blog — contains zero JSON-LD schema of any type. No Organization, WebSite, FAQPage, BreadcrumbList, Product, or Article markup exists. The FAQs page has rich Q&A content (about the company, import/distribution, brands, supply chain, compliance) but is not marked up as FAQPage, so AI engines cannot extract structured answers. The blog has 30+ press-coverage articles but no Article or NewsArticle schema.

External Signals and Discoverability

The brand has near-zero external web footprint. Multiple search queries for "Keplin Group" combined with retailer names, product categories, and company descriptors returned zero results across DuckDuckGo. No press mentions, no review sites, no Reddit threads, no industry directory listings surfaced. The blog claims coverage in Good Housekeeping, The Sun, The Express, OK Magazine, Closer, ITV This Morning, and others, but none of these external articles appear in search results for the brand name. The DNS TXT record contains an anthropic-domain-verification token, indicating the brand has proactively registered with Anthropic — but the cold-knowledge gap shows this verification has not translated into accurate model recall.

Content and Platform

The site runs on Symphony Commerce (a UK ecommerce platform), hosted on a DigitalOcean VPS behind Apache with Varnish caching. The content security policy is extremely permissive (default-src * blob: data: 'unsafe-inline' 'unsafe-eval'). The homepage has strong B2B positioning copy ("Brands Built for Belief", "Products Built for Life") with specific retailer names and claims of 1,000+ SKUs, 12+ brands, 4,000+ retail stores, and 1M+ positive reviews — but none of this data is surfaced in structured format. The sitemap includes 58 URLs with thin pages like /pages/test, /pages/40, /pages/interns, and /pages/search-roles that dilute crawl budget.

Findings

1. Frontier LLMs hallucinate Keplin Group as a US IT firm High

   When queried cold, a frontier LLM describes Keplin Group as a US-based IT managed-services provider founded in 1999 in New Jersey, which is completely fabricated. The actual company is a UK-headquartered B2B consumer goods supplier with 9+ in-house brands supplying major retailers.

   What to change: Publish accurate company information on the website with Organization schema, and submit verified business profiles to authoritative data sources like Wikipedia, Crunchbase, and industry directories.

2. Near-zero external web presence across search engines High

   Multiple search queries for 'Keplin Group' combined with retailer names, product categories, and company descriptors returned zero results across DuckDuckGo. No press mentions, review sites, Reddit threads, or industry directory listings were found, despite blog claims of coverage in major outlets.

   What to change: Build external backlinks through PR, guest posting, and directory listings. Ensure press coverage articles are published on high-authority domains and link back to the site.

3. Zero JSON-LD structured data on any page High

   Every page examined — homepage, Our Brands, Our Products, FAQs, Stockists, Milestones, Blog — contains no JSON-LD schema of any type. No Organization, WebSite, FAQPage, BreadcrumbList, Product, or Article markup exists.

   What to change: Add Organization schema to the homepage, FAQPage schema to the FAQs page, Article schema to blog posts, and Product schema to product pages.

4. Consumer-facing site keplin-online.com blocked to all traffic High

   The consumer-facing site at keplin-online.com returns a 403 Cloudflare challenge for all user agents, including all AI crawlers and even browser UAs, making it completely invisible to search engines and AI retrieval systems.

   What to change: Allow AI crawlers and search engine bots through the Cloudflare challenge, or move consumer content to the main domain.

5. llms.txt file returns 404 Medium

   The site does not provide an llms.txt file, which is a recommended AI-friendly content map that helps LLMs discover and understand site content.

   What to change: Create an llms.txt file listing key pages and a brief description of the site's content for AI crawlers.

6. Robots.txt lacks AI-bot-specific directives Low

   The robots.txt file is permissive (User-agent: * Disallow: /_panel) but contains no directives for AI crawlers like GPTBot, ClaudeBot, or Google-Extended, missing an opportunity to guide AI crawlers to valuable content.

   What to change: Add specific directives for AI crawlers to allow access to key pages and disallow thin or irrelevant pages.

7. Sitemap includes thin or irrelevant pages Low

   The sitemap contains 58 URLs, including thin pages like /pages/test, /pages/40, /pages/interns, and /pages/search-roles, which dilute crawl budget and may confuse crawlers.

   What to change: Remove thin or irrelevant pages from the sitemap and ensure only high-quality, canonical pages are included.

8. FAQs page not marked up as FAQPage schema Medium

   The FAQs page contains rich Q&A content about the company, import/distribution, brands, supply chain, and compliance, but is not marked up as FAQPage, so AI engines cannot extract structured answers.

   What to change: Add FAQPage schema to the FAQs page to enable rich results and AI-friendly structured answers.

9. Blog posts lack Article or NewsArticle schema Medium

   The blog has 30+ press-coverage articles but no Article or NewsArticle schema, reducing their visibility in AI-driven search and knowledge panels.

   What to change: Add Article or NewsArticle schema to all blog posts, including headline, datePublished, author, and publisher information.

10. Anthropic domain verification token does not improve model recall Medium

    The DNS TXT record contains an anthropic-domain-verification token, indicating proactive registration with Anthropic, but the cold-knowledge gap shows this verification has not translated into accurate model recall.

    What to change: Ensure the website content clearly and accurately describes the company, and consider submitting a business profile to Anthropic's knowledge base or using other AI content distribution channels.

What's working

• All major AI crawlers receive full 200 responses from B2B site — All tested AI crawlers (GPTBot, ClaudeBot, PerplexityBot, etc.) receive full 200 responses with identical byte-size content from keplin-group.com, with no UA-based blocking or Cloudflare challenges.
• Robots.txt allows all crawlers except /_panel — The robots.txt file is permissive with only a single disallow for /_panel, ensuring no accidental blocking of crawlers.
• Anthropic domain verification token present in DNS — The DNS TXT record includes an anthropic-domain-verification token, showing proactive registration with Anthropic to establish domain ownership.
• Homepage contains strong B2B positioning with specific claims — The homepage includes specific claims of 1,000+ SKUs, 12+ brands, 4,000+ retail stores, and 1M+ positive reviews, providing rich content for AI crawlers to index.
• FAQs page provides detailed Q&A content — The FAQs page contains 1640 words of detailed Q&A covering company background, import/distribution, brands, supply chain, and compliance, which is valuable for AI extraction if marked up.
• Blog features 30+ press-coverage articles — The blog contains over 30 articles covering press mentions in Good Housekeeping, The Sun, The Express, and other outlets, providing authoritative external signals if properly linked.
• Sitemap available with 58 URLs — The site has a sitemap at /sitemap.xml listing 58 URLs, helping crawlers discover content.