gandysinternational.com — AI Site Grade

Analysis

I have enough data to write a sharp audit. Let me compile it.

The site is actively closing down — but no AI model knows that yet

The homepage screams "CLOSING DOWN SALE — UP TO 75% OFF EVERYTHING — NOTHING OVER £100 — ALL PRODUCTS ARE OUT OF STOCK." The cold LLM knowledge from a frontier model describes Gandys as a thriving social enterprise selling flip-flops with a "Buy One, Give One" model, founded by the Forkan brothers after surviving the 2004 tsunami and appearing on *Dragons' Den*. That prior knowledge is entirely obsolete — the brand has pivoted to full apparel and is now liquidating.

Crawler Access

All AI bots (GPTBot, ClaudeBot, PerplexityBot, Google-Extended, OAI-SearchBot, anthropic-ai, Bytespider, Applebot-Extended, ChatGPT-User) receive 200 with full content on the homepage — identical byte size to browser baseline (1.4 MB). The robots.txt has no AI-specific rules; the wildcard User-agent: * allows / broadly. However, every deeper page (product pages, /collections/all, /pages/our-story, /pages/about-us, /pages/impact-report, /agents.md) returns 403 Cloudflare challenge to all bots and browsers alike. The sitemap blog sub-index also 403s. This means AI crawlers can only see the homepage — the entire product catalog, brand story, impact reporting, and blog are invisible to them.

Cold-Knowledge Gap

The model knows Gandys as a flip-flop and sandal brand with an "Orphans for Orphans" model, founded by Rob and Paul Forkan after losing their parents in the 2004 tsunami, with a *Dragons' Den* investment from Peter Jones and Duncan Bannatyne. The actual site sells none of that — it sells fleece gilets, waxed cotton backpacks, waterproof jackets, dungarees, and shirts. The "Buy One, Give One" flip-flop model is gone, replaced by "10% of your purchase funds children's education projects" (per the Wayback snapshot from May 2024). The site claims Certified B Corp status and references an "Impact report" page that 403s. The copyright reads 2026, which is either a placeholder or a forward-dated error.

Schema Posture

The homepage has Organization, WebSite, and BreadcrumbList schema — but no ClothingStore or Product schema on the homepage despite being a storefront. The Wayback snapshot from May 2024 had a full ClothingStore schema with address, telephone, geo coordinates, and price range. That richer schema has been removed from the current live site. No FAQ schema, no comparison markup, no review aggregate schema despite products showing star ratings (e.g., "4.9 Rated 4.9 out of 5 stars 24 Reviews").

External Signals

DuckDuckGo returns zero search results for gandysinternational.com, the brand name, or any combination of "Gandys clothing." No Reddit threads, no Trustpilot results, no press articles surfaced. The brand has virtually no discoverable off-domain footprint in current search. DNS shows Shopify hosting (23.227.38.65), Google Workspace mail, and Klaviyo for email marketing.

Surprising Findings

The llms.txt and robots.txt contain sophisticated agent commerce instructions referencing the Universal Commerce Protocol (UCP), MCP endpoints, and the Shop.app skill — Shopify's bleeding-edge agent-commerce infrastructure. This is unusually forward-looking for a site that is simultaneously running a "closing down sale" with "all products are out of stock." The llms.txt references UCP versions dated 2026-04-08 (future-dated), matching the © Copyright 2026 in the footer. The /agents.md file referenced in robots.txt returns a 403 Cloudflare challenge, defeating its purpose. The blog at /blogs returns a 404 — the blog has been removed entirely. The /pages/our-purpose page also 404s. The brand story, impact metrics, and B Corp credentials — the core social-enterprise narrative that differentiates Gandys — are all behind Cloudflare walls or deleted.

Findings

1. All deeper pages return 403 Cloudflare challenges to AI crawlers High

   Every page beyond the homepage — product pages, collections, brand story, impact report, agents.md — returns a 403 Cloudflare challenge to all 11 tested AI bots, making the entire product catalog and brand narrative invisible to crawlers.

   What to change: Remove Cloudflare challenge rules that block AI bot user-agents on deeper pages, or configure a separate allowlist for verified AI crawlers.

2. Cold LLM knowledge describes an obsolete flip-flop brand High

   Frontier models describe Gandys as a flip-flop social enterprise with a 'Buy One, Give One' model and Dragons' Den investment, but the site now sells apparel and is liquidating. This mismatch undermines AI-generated brand summaries.

   What to change: Publish updated brand information on accessible pages and ensure AI crawlers can index the current product catalog and mission statement.

3. Homepage lacks ClothingStore schema despite being a storefront Medium

   The homepage has Organization, WebSite, and BreadcrumbList schema but no ClothingStore or Product schema. The Wayback snapshot from May 2024 included a full ClothingStore schema with address, telephone, and price range, which has been removed.

   What to change: Restore ClothingStore schema markup on the homepage, including address, telephone, geo coordinates, and price range.

4. Zero external search results for the brand or domain Medium

   DuckDuckGo returns no results for the domain, brand name, or related queries. No Reddit threads, Trustpilot reviews, or press articles are discoverable, indicating virtually no off-domain footprint.

   What to change: Build external signals through PR, social media, and review platforms to improve discoverability.

5. Blog and key narrative pages return 404 or 403 Medium

   The blog at /blogs returns 404, and pages like /pages/our-purpose also 404. The brand story, impact metrics, and B Corp credentials are either deleted or behind Cloudflare walls, erasing the social-enterprise narrative.

   What to change: Restore or redirect key narrative pages so AI crawlers can access the brand story and impact information.

6. Agents.md file referenced in robots.txt is blocked by Cloudflare Medium

   The /agents.md file, intended for AI agent instructions, returns a 403 Cloudflare challenge, defeating its purpose and contradicting the forward-looking agent commerce setup.

   What to change: Ensure /agents.md is publicly accessible without Cloudflare challenges.

7. Copyright year reads 2026, likely a placeholder error Low

   The footer displays '© Copyright 2026', which is either a placeholder or a forward-dated error, potentially confusing crawlers and users about the site's currency.

   What to change: Update the copyright year to the current year or remove the year entirely.

8. Sitemap blog sub-index returns 403 Medium

   The sitemap blog sub-index at /sitemap_blogs_1.xml returns a 403, preventing crawlers from discovering any blog content.

   What to change: Ensure the sitemap blog sub-index is accessible to crawlers.

What's working

• Homepage returns 200 with full content to all AI bots — The homepage is fully accessible to all 11 tested AI bots with identical content as a browser, ensuring the closing-down sale message is indexed.
• Robots.txt has no AI-specific blocks — The robots.txt file allows all user agents to access the root, with no disallow rules targeting AI bots.
• llms.txt file published with agent commerce instructions — The site publishes an llms.txt file containing sophisticated Universal Commerce Protocol and MCP endpoint references, indicating forward-thinking agent-commerce infrastructure.
• Homepage includes Organization, WebSite, and BreadcrumbList schema — The homepage has basic structured data markup that helps search engines understand the site's identity and structure.
• Sitemap index is accessible and contains 80 URLs — The sitemap index at /sitemap.xml returns 200 and lists 80 URLs, providing a roadmap for crawlers.