cafeyn.co — AI Site Grade

Analysis

Cafeyn.co — AI-Visibility Audit

The entire cafeyn.co domain is a JavaScript single-page application shell that returns zero words of visible text to every AI crawler tested — GPTBot, ClaudeBot, PerplexityBot, Google-Extended, OAI-SearchBot, ChatGPT-User, Bytespider, Applebot-Extended, and anthropic-ai all receive the same empty <div id="app-vue"></div> HTML with no title, no meta description, no headings, no JSON-LD, and no content of any kind. The site is functionally invisible to AI engines.

Crawler Access

All AI bots receive HTTP 200 with identical byte size (3,931 bytes) — the Vue.js bootstrap HTML. No bot is blocked by robots.txt or WAF, but every bot gets a content-empty page. The robots.txt at cafeyn.co/robots.txt contains only two rules: a disallow for meta-externalagent (Facebook's scraper) and a disallow of /article/ for *. No AI bot user-agents are mentioned. The llms.txt path returns the same JS shell HTML (not a valid llms.txt). The .well-known/ai-plugin.json path also returns the JS shell. The site runs on nginx/1.15.5 behind Gandi DNS, with no Cloudflare or other WAF layer — the empty shell is served directly, not a block page.

Content & Schema Posture

Every page on the domain — /, /fr, /en, /fr/magazines, /fr/offers, /fr/search, /en/about, /en/press, /en/legal, /en/faq, /blog — returns the same Vue.js shell with zero JSON-LD schemas, zero <title>, zero <meta name="description">, zero <h1> through <h6>, and zero visible text. The only server-rendered page found is /errors/browser-update.html, which contains 20 words ("Your browser is not compatible with Cafeyn"). The sitemap at /sitemap.xml lists only 4 URLs — three Vue shell pages and one PDF. No FAQ schema, no breadcrumb, no Organization, no WebSite schema exists anywhere on the domain.

Cold-Knowledge Gap

The LLM queried cold knows Cafeyn as "a digital newsstand and subscription service providing access to thousands of magazines and newspapers" — founded in France in 2011 (as LeKiosk), rebranded in 2019, offering over 1,000 titles, with a mobile app and corporate/library offerings. None of this information exists on cafeyn.co. The site's own domain communicates nothing about its product, pricing, publisher partnerships, app availability, or value proposition. The cold model's knowledge derives entirely from off-domain sources (press, Wikipedia, app store listings) — the site itself contributes zero signal.

External Signals

External search results for "cafeyn" across multiple queries returned zero indexed results from DuckDuckGo, suggesting the site has minimal to no search-engine visibility. No Reddit threads, no press mentions, no reviews surfaced in web searches. The DNS TXT records reveal an anthropic-domain-verification token, indicating the brand has engaged with Anthropic's domain verification process — yet ClaudeBot receives the same empty shell as every other crawler. Multiple Google site-verification tokens are present, suggesting past SEO efforts that have not resulted in indexable content.

Findings

1. All pages are empty JavaScript shells with no visible content High

   Every page on cafeyn.co returns a Vue.js bootstrap HTML with zero visible text, no title, no meta description, no headings, and no JSON-LD. AI crawlers receive an empty <div id="app-vue"></div>.

   What to change: Implement server-side rendering (SSR) or pre-rendering for all pages so that crawlers receive HTML with actual content, titles, and metadata.

2. Zero JSON-LD schemas on any page High

   No structured data (JSON-LD) exists anywhere on the domain. Missing Organization, WebSite, FAQ, Breadcrumb, and other schemas that help AI understand the site's content and purpose.

   What to change: Add JSON-LD schemas for Organization, WebSite, BreadcrumbList, and FAQPage on relevant pages.

3. Missing title and meta description on all pages High

   No page includes a <title> tag or <meta name="description"> tag. This prevents search engines and AI from understanding the page topic and reduces click-through rates.

   What to change: Add unique <title> and <meta name="description"> tags to every page, reflecting the page's content.

4. llms.txt returns the JavaScript shell instead of a valid llms.txt file Medium

   The path /llms.txt returns the same empty Vue.js HTML as other pages, not a valid llms.txt file. This prevents AI crawlers from discovering the site's content via the llms.txt protocol.

   What to change: Serve a proper llms.txt file at /llms.txt that lists the site's key pages and content summaries.

5. ai-plugin.json returns the JavaScript shell instead of a valid plugin manifest Medium

   The path /.well-known/ai-plugin.json returns the same empty Vue.js HTML, not a valid AI plugin manifest. This prevents AI agents from discovering the site as a plugin.

   What to change: Serve a valid ai-plugin.json file at /.well-known/ai-plugin.json describing the site's capabilities.

6. Robots.txt does not mention any AI bot user-agents Medium

   The robots.txt file only disallows Facebook's scraper and blocks /article/ for all bots. No AI bot user-agents (GPTBot, ClaudeBot, etc.) are explicitly allowed or disallowed, leaving them to default access but receiving empty pages.

   What to change: Add explicit allow rules for AI bots in robots.txt, or ensure they receive rendered content.

7. Sitemap contains only 4 URLs Medium

   The sitemap at /sitemap.xml lists only 4 URLs, which is insufficient for a site with many pages. This limits search engine discovery of the site's content.

   What to change: Expand the sitemap to include all important pages, especially magazine listings, offers, and blog posts.

8. No external search results found for the domain or brand High

   Multiple web searches for 'cafeyn.co' and 'Cafeyn' returned zero results on DuckDuckGo, indicating minimal to no search engine visibility. No Reddit threads, press mentions, or reviews were found.

   What to change: Improve on-page SEO with server-rendered content, build backlinks, and ensure the site is indexed by search engines.

9. Cold LLM knowledge about Cafeyn is not reflected on the site High

   The LLM knows Cafeyn as a digital newsstand with over 1,000 magazines, but the site itself communicates none of this information. The site contributes zero signal to AI knowledge.

   What to change: Add descriptive content about the service, including number of magazines, pricing, app availability, and publisher partnerships, using server-rendered HTML.

10. CDN image returns 403 Forbidden Low

    The Open Graph image at cdn.cafeyn.co returns a 403 error, preventing social media and AI crawlers from accessing the image. This can affect link previews and AI understanding.

    What to change: Ensure the CDN allows access to the image, or host it on the main domain with proper permissions.

11. Help subdomain fails DNS resolution Medium

    The help.cafeyn.co subdomain does not resolve, returning a DNS failure. This prevents users and crawlers from accessing help content.

    What to change: Fix the DNS record for help.cafeyn.co or redirect it to a working help page on the main domain.

12. No hreflang tags or localized content visible to crawlers Medium

    Although the site has /fr and /en paths, all return the same empty shell. No hreflang tags or localized metadata are present, hindering international SEO.

    What to change: Implement hreflang tags and serve localized content via SSR for each language version.

What's working

• All AI bots are allowed by robots.txt and return HTTP 200 — No AI bot is blocked by robots.txt or WAF. All tested bots receive HTTP 200 responses, indicating the site does not intentionally block crawlers.
• Anthropic domain verification token present in DNS — The DNS TXT records include an anthropic-domain-verification token, indicating the brand has engaged with Anthropic's verification process, which is a positive signal for AI visibility.
• Multiple Google site verification tokens present — DNS TXT records contain multiple google-site-verification tokens, suggesting past SEO efforts and ownership verification with Google Search Console.
• Stable nginx server with no WAF blocking — The site runs on nginx/1.15.5 with no Cloudflare or other WAF layer, meaning crawlers are not blocked by a firewall. The empty shell is served directly.
• Historical snapshot available on Wayback Machine — A snapshot from January 2024 exists on the Wayback Machine, showing the site had some content in the past. This provides a baseline for recovery.
• Browser update page has server-rendered content — The /errors/browser-update.html page is server-rendered with 20 words of text, demonstrating that the site is capable of serving static HTML when needed.
• CSS and JS assets are accessible to crawlers — The main CSS and JS files are served successfully, indicating that the site's static assets are not blocked and could be used for rendering if SSR were implemented.
• Favicon is accessible — The favicon.ico is served successfully, a minor but positive signal that basic assets are in place.