mrlemonade.co.uk — AI Site Grade

Analysis

Mr. Lemonade — AI-Visibility Audit

The site has a fully operational Universal Commerce Protocol (UCP) endpoint, an llms.txt, and an agents.md — yet a frontier LLM queried cold has zero knowledge of the brand, and no external mentions surface in web search results.

Crawler Access

All 11 tested AI crawlers (GPTBot, ClaudeBot, PerplexityBot, Google-Extended, OAI-SearchBot, ChatGPT-User, anthropic-ai, Bytespider, Applebot-Extended, Perplexity-User) receive 200 OK with full content — identical byte size (~648KB) to a browser baseline. The site runs on Shopify behind Cloudflare with no UA-based blocking. The robots.txt is the standard Shopify template with Allow: / for * and no AI-bot-specific rules. No AI crawler is blocked, rate-limited, or served a thin shell.

llms.txt and Agent Protocol

The site hosts a detailed llms.txt (and an identical agents.md) that goes far beyond the typical convention. It documents the Shopify UCP endpoint (/.well-known/ucp), MCP tool discovery, a full agent shopping flow (search → cart → checkout → fulfill), supported protocol versions (2026-04-08, 2026-01-23), and payment handlers including Google Pay. The UCP endpoint returns a valid JSON merchant profile. This is Shopify's standard agent-commerce template — not custom-authored — but it is present and functional, placing the site in the small minority of ecommerce stores with any agent-facing infrastructure at all.

Cold-Knowledge Gap

A frontier LLM queried on "Mr Lemonade mrlemonade.co.uk" returned: *"I do not have specific, verifiable information... The entity does not appear in my training data as a well-known or widely referenced business."* This is the most striking finding. The site describes itself as a wholesale drinks distributor serving London and the UK since 2016, stocking 80+ brands (Fever-Tree, San Pellegrino, Oatly, Dalston's, Tony's Chocolonely, etc.) across 919 SKUs — yet it has zero external footprint in LLM training data, web search results, or review platforms.

Content and Schema Posture

Product pages carry well-formed Product schema with sku, brand, offers (price, currency, availability), and BreadcrumbList. The homepage has Organization and WebSite schema with SearchAction. However, no LocalBusiness schema is present despite a physical address in Barking, London. All prices are hidden behind a login wall ("Login or Register to view price"), which means AI crawlers see product names and descriptions but no pricing data — a significant limitation for agent-driven commerce. The site has no FAQ schema, no comparison tables, and no blog content.

External Signals

The site references Facebook, Instagram, LinkedIn, and WhatsApp contact channels, but no external reviews, press mentions, Reddit threads, or Google Business profile citations were found across multiple web searches. The Wayback Machine has no snapshot of the domain. The brand operates in a near-complete off-domain vacuum, which explains the LLM's blank prior. For a business that has been operating since 2016 and claims to serve independent businesses across London, this absence of external signals is anomalous and directly limits AI-generated brand awareness.

Findings

1. Zero external footprint in LLM training data and web search High

   A frontier LLM queried cold has no knowledge of Mr. Lemonade. Multiple web searches for the brand, reviews, and social media returned zero results. The Wayback Machine has no snapshot. This absence of external signals directly limits AI-generated brand awareness.

   What to change: Build an external presence: claim a Google Business Profile, encourage reviews on Trustpilot or Google, publish content (blog, press releases), and maintain active social media profiles. Ensure these pages are indexed and linked from the main site.

2. Product prices hidden behind login wall High

   All product prices on the site require login or registration to view. AI crawlers see product names and descriptions but no pricing data, which severely limits the site's utility for agent-driven commerce and price comparison.

   What to change: Expose prices publicly for AI crawlers, either by showing them without login or by embedding pricing data in structured data (e.g., Product schema offers) that is accessible to bots.

3. No LocalBusiness schema despite physical address Medium

   The site has a physical address in Barking, London, but does not include LocalBusiness schema. This reduces visibility in local AI-generated answers and local search results.

   What to change: Add LocalBusiness schema with address, phone, and opening hours to the contact page and homepage.

4. No FAQ or comparison schema on key pages Medium

   The site lacks FAQ schema and comparison tables, which are commonly used by AI to generate rich answers. This limits the site's ability to appear in AI-generated summaries for queries about wholesale drinks.

   What to change: Add FAQ schema to relevant pages (e.g., delivery, about us) and consider adding comparison tables for product categories.

5. No blog or educational content Medium

   The site has no blog, articles, or guides. This reduces the surface area for AI to reference the brand in informational queries and limits organic search visibility.

   What to change: Create a blog with articles about drinks trends, wholesale tips, or brand stories to build topical authority and provide content for AI to cite.

6. Robots.txt lacks AI-bot-specific rules Low

   The robots.txt uses the default Shopify template with no explicit rules for AI crawlers. While all tested bots currently get full access, the absence of explicit allowances could lead to future blocking if defaults change.

   What to change: Add explicit Allow directives for GPTBot, ClaudeBot, and other major AI crawlers to ensure continued access.

What's working

• Universal Commerce Protocol endpoint is functional — The site hosts a valid UCP endpoint at /.well-known/ucp that returns a JSON merchant profile, enabling agent-driven shopping flows.
• llms.txt and agents.md are present and detailed — The site hosts a comprehensive llms.txt (and agents.md) that documents the UCP endpoint, MCP tools, agent shopping flow, and supported protocol versions, placing it ahead of most ecommerce stores in AI readiness.
• All 11 tested AI crawlers receive full content — Every major AI crawler tested receives a 200 OK with full HTML content identical to a browser, with no blocking or rate-limiting.
• Product schema is well-formed on product pages — Product pages include valid Product schema with sku, brand, offers (price, currency, availability), and BreadcrumbList, aiding AI understanding of product data.
• Homepage has Organization and WebSite schema — The homepage includes Organization and WebSite schema with SearchAction, providing basic brand and site information to AI crawlers.
• Sitemap is present with 80 URLs and indexable — The sitemap returns 200 OK with 80 URLs and is set to index, helping crawlers discover all pages.