spoke-london.com — AI Site Grade

Analysis

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

---

AI crawlers get full, equal access — but find a content desert

Every major AI bot (GPTBot, ClaudeBot, PerplexityBot, Google-Extended, ChatGPT-User, OAI-SearchBot, Bytespider, Applebot-Extended, anthropic-ai) receives a 200 with identical 158KB payload as a browser — no blocks, no UA-based throttling, no JS shell. The robots.txt explicitly allows all listed AI agents with Allow: / (minus admin/cart/account paths). The site runs on Shopify + Next.js, served via AWS CloudFront + S3, with no WAF gating AI traffic. This is an unusually open posture. Yet the content those bots consume is structurally thin.

Cold-Knowledge Gap

A frontier LLM queried cold about Spoke London recalls a 2015 founding by Ben Farren and James Cox, a "size finder" algorithm, Trustpilot 4.6 stars, and a focus on men's trousers with 30+ size combinations. The actual site tells a different story: founded 2013 by Ben alone (no James Cox mentioned anywhere), no blog/journal (the /blogs/journal path 404s), no press page, no Trustpilot widget or external review integration visible. The model's prior about "Chino and Wool Trousers lines with Vitale Barberis Canonico fabrics" does not match the site's current product taxonomy — the site sells Heroes, Sharps, Fives, Hackneys, Bulletproofs, Smarts, and Denim sub-brands, none of which the model named. The model also claims a "Perfect Fit Guarantee" with free returns — the site actually offers a 200-day return policy with a £4.95 shipping charge, not free.

Schema Posture

Product pages carry valid Product schema with AggregateRating (e.g., 4.68 from 1,086 reviews on the Sharps page), Offer with price, return policy, and shipping details. However, no collection, homepage, or content page carries any schema — no Organization, WebSite, BreadcrumbList, FAQPage, or Article markup. The homepage has zero JSON-LD. The About page has zero JSON-LD. The Fit Proposition page has zero JSON-LD. For a brand whose entire pitch is "ridiculously well fit," the sizing explainer page (/pages/fit-finder-fit-proposition) — which describes 15 waists, 3 builds, and 10 leg lengths — has no structured data to help AI engines surface that differentiation.

Missing Infrastructure

/llms.txt returns 404. The help center at help.spoke-london.com is a JS-rendered shell — GPTBot receives a 59KB page with a loading spinner and zero visible FAQ content. /pages/faqs also 404s. There is no blog, no editorial content, no press section, no "as seen in" section. The footer links to a /pages/about-us page that tells the founding story well (786 words) but has no schema and no author/publisher markup. External search returns zero indexed Reddit threads, zero Trustpilot results, zero press mentions from the DuckDuckGo crawl — the brand's off-domain footprint is nearly invisible to discovery search, which means AI engines have little third-party citation material to draw on.

The Core Contradiction

Spoke London's site is technically open to AI crawlers and product pages carry good schema, but the informational layer is absent. The model's cold knowledge is outdated (wrong founder story, wrong founding year, wrong product names) and the site offers no blog, no FAQ page, no press page, no llms.txt, and no schema on any non-product page to correct it. AI engines retrieving the site live will see a rich product catalog but no narrative, no brand story, no sizing explainer in structured form — leaving them to fall back on stale, partially incorrect priors.

Findings

1. No /llms.txt file published Medium

   The site returns a 404 for /llms.txt, missing an opportunity to provide AI crawlers with a curated summary of the brand and key pages.

   What to change: Publish an /llms.txt file with a concise brand description, key product categories, and links to important pages like the About page and Fit Proposition.

2. No structured data on homepage, About, or Fit Proposition pages High

   Only product pages carry JSON-LD schema. The homepage, About page, and Fit Proposition page have zero structured data, reducing AI understanding of the brand and its unique sizing proposition.

   What to change: Add Organization, WebSite, and BreadcrumbList schema to the homepage; add Organization or AboutPage schema to the About page; add FAQPage or HowTo schema to the Fit Proposition page.

3. Blog and FAQ pages return 404 errors High

   The /blogs/journal and /pages/faqs URLs return 404 pages, indicating missing editorial and support content that AI crawlers could use for brand narrative and common questions.

   What to change: Restore or create a blog with brand stories and a FAQ page with common sizing and return questions, and mark them up with Article and FAQPage schema.

4. Help center renders as empty JS shell for AI crawlers High

   The help center at help.spoke-london.com delivers a 59KB page with a loading spinner and no visible content when fetched as GPTBot, making FAQ content inaccessible to AI crawlers.

   What to change: Server-side render the help center content or provide a static HTML fallback for AI crawlers.

5. AI model cold knowledge is outdated and incorrect High

   A frontier LLM's prior knowledge about Spoke London includes a wrong founding story (2015, co-founder James Cox), incorrect product names, and a different return policy, indicating the site fails to correct AI models with accurate information.

   What to change: Publish an /llms.txt file and add structured data on key pages to provide authoritative brand information to AI crawlers.

6. No external review or press footprint found Medium

   Web searches for Trustpilot reviews, Reddit discussions, and press mentions returned zero results, meaning AI models have no third-party citations to validate or supplement the brand's claims.

   What to change: Encourage customer reviews on third-party platforms and build a press page with media mentions to create an off-domain citation trail.

7. No press or 'as seen in' section on the site Low

   The site lacks any press mentions or media coverage section, missing an opportunity to build credibility and provide AI crawlers with authoritative external references.

   What to change: Add a press page or 'as seen in' section with logos and links to media coverage, marked up with schema.org/Article or schema.org/CreativeWork.

8. Fit Proposition page lacks structured data for sizing Medium

   The Fit Proposition page describes 15 waists, 3 builds, and 10 leg lengths but has no schema markup, making it harder for AI to surface the brand's key differentiator.

   What to change: Add HowTo or FAQPage schema to the Fit Proposition page to help AI engines understand and present the sizing process.

What's working

• All major AI crawlers allowed with full access — The robots.txt explicitly allows GPTBot, ClaudeBot, PerplexityBot, Google-Extended, and others with Allow: /, and the server returns identical content to bots and browsers.
• Product pages carry valid Product schema with aggregate ratings — Product pages include JSON-LD with Product, AggregateRating, Offer, and return policy, providing rich data to AI crawlers.
• Sitemap.xml is available and contains 80 URLs — The sitemap at /sitemap.xml returns 200 and lists 80 URLs, helping crawlers discover site content.
• About page provides a detailed brand story — The About page contains 936 words of brand narrative, including founding year and mission, which can inform AI models if properly marked up.
• Collection pages contain descriptive product text — The trousers collection page has 878 words of descriptive content, providing context for AI crawlers beyond just product listings.