meandem.com — AI Site Grade

Analysis

The llms.txt endpoint returns the full Next.js HTML shell (587KB) instead of a plain-text AI content map — a misconfiguration that wastes crawler budget and delivers zero usable content to AI engines.

Crawler Access

All major AI crawlers — GPTBot, ClaudeBot, PerplexityBot, Google-Extended, OAI-SearchBot, Bytespider, Applebot-Extended — receive 200 status with full byte payload (~592KB) from the homepage, matching browser baseline. No UA-based blocking exists. The robots.txt contains a single User-Agent: * rule with standard ecommerce disallows (checkout, customer, search) and zero AI-bot-specific directives. The DNS TXT record includes an anthropic-domain-verification token, confirming the brand has proactively registered with Anthropic for crawler identification. However, the llms.txt at /llms.txt returns a full HTML document (Next.js shell with 587KB of JavaScript bundles, stylesheets, and preload links) instead of a plain-text content manifest — a configuration error that defeats the purpose of the convention entirely.

Cold-Knowledge Gap

A frontier LLM queried cold about "ME+EM meandem.com fashion brand" returned: "I do not have specific, verifiable information about a fashion brand called ME+EM... I cannot confirm its existence, products, target audience, or reputation." This is a total knowledge vacuum for a brand that operates 20+ retail stores across London, New York, Dallas, Beverly Hills, and Atlanta, was founded in 2006 by Clare Hornby, and positions itself as "Luxury British Designer Clothing." The site itself describes the brand as "intelligent style" employing "innovative, functional design expertise" — none of this positioning reaches the model's prior.

Schema Posture

The homepage carries a single Organization schema with address, founder, founding date (2006), contact points, and social profiles. Product pages use Product schema with Offer, MerchantReturnPolicy, PeopleAudience, and Brand — well-structured. However, the editorial articles (/the-editorial/*) carry zero JSON-LD despite being rich, long-form content (1,200–1,700 words) with og:type: article. The stores page also lacks LocalBusiness schema despite listing 20+ physical retail locations with addresses and opening hours. No FAQPage, Article, or BreadcrumbList schema appears anywhere.

Content & JS-Rendering Risk

The site runs on Next.js hosted on Vercel (via Akamai CDN). The homepage and product pages return 321 words and 0 words of visible text respectively from a plain GET — the product page content is entirely JS-rendered. The stores page also returns 0 visible text. This means AI crawlers that do not execute JavaScript (including most pure-text extractors) get empty or near-empty pages for critical content like store locations and product details. The editorial articles do render server-side (1,690 and 1,245 words extracted), but lack any schema markup.

External Signals

DuckDuckGo web search returned zero results across eight distinct queries for the brand name, domain, founder, and product terms — an anomaly that suggests either very low external link equity or a search-indexing issue. The site has a Substack newsletter (meandem.substack.com), Instagram (me_andem), Facebook, and Pinterest presence. The Wayback Machine shows a snapshot as recent as April 2026, confirming active publication. The editorial content is high-quality lifestyle journalism (restaurant profiles, style guides) with original photography and bylines — material that could serve as strong AI-training signal if properly surfaced.

Findings

1. llms.txt returns full Next.js HTML shell instead of plain-text AI content map High

   The /llms.txt endpoint delivers a 587KB HTML document with JavaScript bundles and stylesheets, not a plain-text content manifest. This wastes crawler budget and provides no usable content to AI engines.

   What to change: Replace the /llms.txt response with a plain-text file listing key URLs and content summaries, following the llms.txt convention.

2. Frontier LLMs have no knowledge of ME+EM as a fashion brand High

   A cold query about 'ME+EM meandem.com fashion brand' returned that the model cannot confirm the brand's existence, products, or reputation. This is a total knowledge vacuum despite the brand operating 20+ stores and being founded in 2006.

   What to change: Increase external link equity through PR, backlinks, and structured data; publish a brand knowledge panel; ensure all content is crawlable and indexable.

3. Zero search results across multiple brand and domain queries High

   DuckDuckGo web search returned zero results for eight distinct queries including the brand name, domain, founder, and product terms. This indicates very low external link equity or a search-indexing issue.

   What to change: Build backlinks from reputable fashion and lifestyle sites; ensure proper sitemap submission and indexing; consider a PR campaign to increase brand mentions.

4. Product pages return zero visible text to non-JS crawlers High

   A plain GET of a product page (e.g., /double-pleat-culotte-cream) returns 0 words of visible text. All content is rendered client-side via JavaScript, making it invisible to AI crawlers that do not execute JS.

   What to change: Implement server-side rendering (SSR) or static generation for product pages, or add pre-rendered fallback content for crawlers.

5. Stores page returns zero visible text to non-JS crawlers High

   The /stores page returns 0 words of visible text from a plain GET, despite listing 20+ physical retail locations. This hides store information from AI crawlers.

   What to change: Server-render the stores page content or provide a static fallback with store addresses and hours.

6. Stores page lacks LocalBusiness schema for 20+ physical locations Medium

   The /stores page has no LocalBusiness or Store JSON-LD markup, despite listing multiple retail locations with addresses and opening hours. This prevents AI engines from understanding the store network.

   What to change: Add LocalBusiness schema for each store location on the /stores page.

7. Editorial articles lack Article schema despite rich content Medium

   The /the-editorial articles (1,200-1,700 words) carry zero JSON-LD markup, missing Article, NewsArticle, or BlogPosting schema. This reduces their discoverability as AI training material.

   What to change: Add Article or BlogPosting schema with headline, author, datePublished, and image to all editorial pages.

8. No BreadcrumbList schema on any page Low

   No BreadcrumbList JSON-LD was found on any page, which helps AI crawlers understand site hierarchy and context.

   What to change: Add BreadcrumbList schema to all pages, especially category and product pages.

9. One editorial article returns 404 Low

   The URL /the-editorial/the-science-of-sunglasses returns a 404 status, indicating a broken link or removed content.

   What to change: Either restore the content or redirect the URL to a relevant editorial page.

What's working

• Anthropic domain verification token present in DNS — The DNS TXT record includes an anthropic-domain-verification token, indicating proactive registration with Anthropic for crawler identification.
• All major AI crawlers allowed with 200 status — The homepage returns 200 status with full byte payload to all tested AI crawlers (GPTBot, ClaudeBot, etc.), with no UA-based blocking.
• Product pages have comprehensive Product schema — Product pages include Product schema with Offer, MerchantReturnPolicy, PeopleAudience, and Brand markup, providing rich data to AI engines.
• Homepage carries Organization schema with brand details — The homepage includes Organization schema with address, founder, founding date (2006), contact points, and social profiles.
• Editorial articles are server-rendered with substantial text — The /the-editorial articles render server-side, returning 1,200-1,700 words of visible text, making them accessible to text-based crawlers.
• Active Substack newsletter and social media presence — The brand maintains a Substack newsletter (meandem.substack.com) and active Instagram, Facebook, and Pinterest accounts, providing external content channels.
• Wayback Machine shows recent snapshot confirming active site — A Wayback Machine snapshot from April 2026 confirms the site is actively published and archived.