dowerandhall.com — AI Site Grade

Analysis

Dower & Hall: A Live Site That No AI Crawler Can Read

The entire dowerandhall.com domain is locked behind a Cloudflare JavaScript challenge (Under Attack mode) that returns HTTP 403 to every request — browser, GPTBot, ClaudeBot, PerplexityBot, Google-Extended, and all other AI crawlers tested — without ever serving a single byte of real content.

Crawler Access

Every AI crawler tested (GPTBot, ClaudeBot, PerplexityBot, Google-Extended, OAI-SearchBot, ChatGPT-User, anthropic-ai, Bytespider, Applebot-Extended) receives the same Cloudflare JS challenge wall as a plain browser. No bot gets past it. The robots.txt is also behind the wall (returns the Cloudflare challenge HTML, not a robots file), meaning crawlers cannot even discover what they are allowed to crawl. There is no llms.txt. The sitemap is blocked. The site runs on Shopify (confirmed via DNS TXT records: shopify-verification-code, shops.shopify.com SPF include) behind Cloudflare's managed challenge platform — a configuration that makes the entire domain invisible to non-JS-capable HTTP clients.

Cold-Knowledge Gap

A frontier LLM queried cold describes Dower & Hall as a "British fine jewelry brand" founded in 1990 by John Dower and Sally Hall, known for charm collections, stackable rings, recycled metals, and mentions in Vogue and Harper's Bazaar. This prior knowledge is richer than anything the live site currently exposes to any crawler. The model knows the brand's founding story, its ethical positioning, and its retail presence at Liberty London — none of which can be verified or updated via the live site because no crawler can reach it. The gap is extreme: the model's knowledge is frozen at training time, and the live site provides zero signal to correct, reinforce, or expand it.

Schema and Content Posture

The Wayback snapshot (February 2025) shows a Shopify storefront with Organization and Website schema types present but minimal — the Organization schema includes a street address in Clapham, London, a phone number, and a description, but no logo, sameAs links, or social profiles. The BreadcrumbList schema has a broken label ("Translation missing: en.general.breadcrumb.home"). The homepage contains rich product content (bestsellers, collections, sustainability promises) and FAQ-style signals, but none of this is accessible to crawlers on the live site. The live site serves exactly 10 words of visible text: "Your connection needs to be verified before you can proceed."

External Signals

External search returns zero indexed results for the brand name, the domain, or any combination of brand + review sites. No Trustpilot, no press mentions, no Reddit threads appear in search results. The DNS records confirm Microsoft 365 mail and Apple developer verification, but the web presence is effectively null — the Cloudflare wall has erased the site from search engine visibility entirely.

Findings

1. Cloudflare JS challenge blocks all AI crawlers High

   Every AI crawler tested (GPTBot, ClaudeBot, PerplexityBot, Google-Extended, OAI-SearchBot, ChatGPT-User, anthropic-ai, Bytespider, Applebot-Extended) receives a Cloudflare JavaScript challenge wall and HTTP 403, preventing any content from being accessed.

   What to change: Disable Under Attack mode or configure Cloudflare to allow AI crawlers by whitelisting known bot user agents or using a firewall rule to bypass the JS challenge for verified bots.

2. Robots.txt is inaccessible behind Cloudflare wall High

   The robots.txt file returns the Cloudflare challenge HTML instead of a valid robots file, so crawlers cannot discover crawl rules.

   What to change: Ensure robots.txt is served without JS challenge, or configure Cloudflare to allow access to /robots.txt for all user agents.

3. Sitemap is blocked by Cloudflare High

   The sitemap.xml returns HTTP 403, preventing crawlers from discovering the site's URL structure.

   What to change: Allow access to /sitemap.xml for all user agents by configuring Cloudflare firewall rules.

4. No llms.txt file present Medium

   The site does not provide an llms.txt file, missing an opportunity to guide AI crawlers to key content.

   What to change: Create an llms.txt file listing important pages like /pages/our-story and /collections.

5. Zero indexed pages in search engines High

   Web searches for the brand name, domain, and related terms return zero results, indicating the site has been completely de-indexed due to the Cloudflare wall.

   What to change: Remove the JS challenge for search engine bots to allow re-indexing, and submit the site to Google Search Console.

6. Cold LLM knowledge cannot be verified or updated High

   A frontier LLM has rich prior knowledge about the brand (founding story, ethical positioning, retail presence) but the live site provides zero signal to confirm or update this information, leaving AI systems with potentially outdated or incorrect data.

   What to change: Allow AI crawlers to access key pages so that LLMs can refresh their knowledge from the live site.

7. Schema markup is minimal and contains errors Medium

   The Wayback snapshot shows Organization and Website schema types present but missing logo, sameAs links, and social profiles. The BreadcrumbList schema has a broken label ('Translation missing: en.general.breadcrumb.home').

   What to change: Add logo, sameAs, and social profile URLs to Organization schema. Fix the broken breadcrumb translation.

8. No external backlinks or press mentions indexed Medium

   Searches for brand mentions on review sites, press outlets, and social media return zero results, indicating the site has no external web presence.

   What to change: Build backlinks through PR, partnerships, and listings on reputable directories. Encourage customer reviews on Trustpilot.

What's working

• LLMs have rich prior knowledge of the brand — A frontier LLM correctly describes Dower & Hall as a British fine jewelry brand founded in 1990, known for charm collections, recycled metals, and retail at Liberty London, providing a strong foundation for AI visibility once the site is accessible.
• Wayback snapshot shows rich product and brand content — The February 2025 snapshot contains detailed product descriptions, collection pages, sustainability promises, and FAQ-style signals that would be valuable for AI crawlers if accessible.
• Shopify platform provides built-in schema markup — The site uses Shopify, which automatically generates Organization and Website schema types, providing a baseline for structured data.
• DNS configuration is valid and includes verification records — DNS records include Shopify verification, Microsoft 365 mail, and Apple developer verification, indicating proper domain ownership and email setup.