AI Site Grade
mycozmo.com — AI Site Grade
Cloudflare JS challenge blocks every AI crawler, making mycozmo.com completely invisible to LLMs and search engines.
Mycozmo.com, a UK sofa brand with a differentiated product, is entirely blocked from AI crawlers by a Cloudflare JS challenge, resulting in zero LLM knowledge and no search engine discoverability.
- Findings
- 10
- Evidence checks
- 31
- Completed
- 30 May 2026
Analysis
Cloudflare JS Challenge Blocks Every AI Crawler — Site Is Invisible to All Bots
Every AI crawler tested — GPTBot, ClaudeBot, PerplexityBot, Google-Extended, OAI-SearchBot, ChatGPT-User, Bytespider, Applebot-Extended, and anthropic-ai — receives a 403 with a Cloudflare JS challenge page when attempting to access mycozmo.com. The browser baseline also returns 403. No bot or tool can retrieve a single byte of real content from the live site.
Crawler Access
The robots.txt (retrieved via Wayback) is the default Shopify template with no AI-bot-specific directives — no Disallow for GPTBot, ClaudeBot, or any AI crawler. This is irrelevant in practice because Cloudflare's JS challenge (cf-mitigated header present) blocks all non-JS-executing clients at the edge before any content is served. The llms.txt file returns the same Cloudflare 403 wall. The sitemap.xml is also blocked. The site runs on Shopify (A record: 23.227.38.32, Shopify's standard IP) behind Cloudflare's WAF/CDN with a managed challenge rule.
Cold-Knowledge Gap
A frontier LLM queried cold about "mycozmo" reports zero knowledge — cannot confirm what the company does, what products it sells, or any brand facts. The live site is a UK-based sofa brand called Cozmo (branded as "Cozmo" on-page, domain mycozmo.com) selling design-led sofas with interchangeable "sofa jackets" at £1,000–£3,500. The site has ~1,200+ product variants across ranges named Hug, Cozy, Float, Pillow, and Rest. The gap between the brand's actual offering (a differentiated modular sofa system with changeable covers, German Click-Fix assembly, London showroom) and the model's complete ignorance is total.
Schema Posture
The homepage (via Wayback) carries four JSON-LD schema types: FurnitureStore, Organization, WebSite, and BreadcrumbList. The FurnitureStore schema includes address (4 Garden Walk, London EC2A 3EQ), telephone, opening hours, payment methods, and social links (Facebook, Instagram, Pinterest). A SearchAction potentialAction is present on the WebSite schema. Product listing pages only carry BreadcrumbList — no Product schema on collection pages. The blog and about pages have no structured data at all.
External Signals
The site references press quotes from what appear to be real publications ("Is this the world's most comfortable sofa", "Sofas look set to shake things up at the more affordable end of the market") but the source publications are not named on the homepage. A Trustpilot widget is embedded. Social links point to facebook.com/CozmoSofa and instagram.com/cozmosofa. DuckDuckGo searches for "cozmo sofa", "mycozmo.com", and "Cozmo sofa interchangeable jackets" returned zero results — the brand has essentially no discoverable external footprint in search engines, which is consistent with the Cloudflare wall preventing indexing.
Stale Content Signals
The blog sitemap shows the last blog post was from 2023 — the blog has not been updated in over two years. Several page-level sitemap entries (/pages/news, /pages/our-story, /pages/assembly-1) have lastmod dates from 2022. The /pages/faqs page was last updated December 2025, which is the most recent content signal. The product sitemap shows daily changefreq with lastmod dates in December 2025, indicating active product inventory management despite the Cloudflare wall.
Findings
Cloudflare JS challenge blocks every AI crawler High
Every AI crawler tested receives a 403 with a Cloudflare JS challenge page. No bot can retrieve any real content from the live site.
What to change: Remove or relax the Cloudflare JS challenge for known AI crawler user agents, or serve a static HTML version to bots.
llms.txt file returns 403 High
The llms.txt file is blocked by Cloudflare, preventing AI crawlers from discovering content guidelines.
What to change: Allow access to llms.txt by excluding it from the JS challenge rule.
robots.txt lacks AI crawler directives Medium
The robots.txt is the default Shopify template with no Disallow rules for GPTBot, ClaudeBot, or other AI crawlers. While irrelevant due to Cloudflare blocking, it misses an opportunity to guide well-behaved bots.
What to change: Add explicit Allow rules for AI crawlers to specific paths, and disallow irrelevant sections.
Sitemap blocked by Cloudflare High
The sitemap.xml returns 403, preventing search engines and AI crawlers from discovering the site's URL structure.
What to change: Allow access to sitemap.xml by excluding it from the JS challenge rule.
Zero LLM knowledge of the brand High
A frontier LLM queried cold about 'mycozmo' reports zero knowledge, unable to confirm the company's products or brand facts.
What to change: Allow AI crawlers to index the site so LLMs can learn about the brand.
No search engine discoverability High
DuckDuckGo searches for 'cozmo sofa', 'mycozmo.com', and 'Cozmo sofa interchangeable jackets' returned zero results. The Cloudflare wall prevents indexing.
What to change: Remove the Cloudflare JS challenge for search engine bots to allow indexing.
Missing Product schema on collection pages Medium
Product listing pages only carry BreadcrumbList schema; no Product schema is present, reducing rich result eligibility.
What to change: Add Product schema markup to collection pages for each product listed.
No structured data on blog and about pages Medium
The blog and about pages have no structured data at all, missing opportunities for rich results.
What to change: Add appropriate schema types (e.g., Article, AboutPage) to blog and about pages.
Blog not updated since 2023 Low
The blog sitemap shows the last blog post was from 2023, indicating no fresh content in over two years.
What to change: Publish new blog posts regularly to signal site freshness.
Press quotes without source attribution Low
The homepage includes press quotes but does not name the source publications, reducing credibility.
What to change: Add source publication names and links to press quotes.
What's working
- JSON-LD schema on homepage — The homepage includes FurnitureStore, Organization, WebSite, and BreadcrumbList schema with address, phone, hours, payment methods, and social links.
- SearchAction schema on homepage — The WebSite schema includes a SearchAction potentialAction, enabling search box rich results.
- Active product inventory management — Product sitemap shows daily changefreq with recent lastmod dates, indicating active inventory updates.
- Trustpilot widget embedded — A Trustpilot widget is embedded on the site, providing social proof.
- Social media links present — The site links to Facebook and Instagram accounts (CozmoSofa), providing external signals.
Track mycozmo.com across AI search
This is one snapshot. Open the interactive report to inspect evidence, or grade another site free.