carmoola.co.uk — AI Site Grade

Analysis

Carmoola's blog listing page returns zero visible text to plain GET — a JS-rendering shell that all AI crawlers (GPTBot, ClaudeBot, PerplexityBot, etc.) receive as a 200 with ~98KB of HTML but no article content, despite the site having 1,463+ URLs in the sitemap and individual blog posts rendering full text fine.

Crawler Access

All 11 tested AI bot UAs (GPTBot, ClaudeBot, PerplexityBot, Google-Extended, OAI-SearchBot, ChatGPT-User, Bytespider, Applebot-Extended, anthropic-ai, Perplexity-User) receive a 200 with full content on the homepage and core product pages — byte sizes match the browser baseline within 0.003%. The site runs on HubSpot behind Cloudflare (AWS DNS, London IPs). The robots.txt contains zero AI-specific directives: no GPTBot, no ClaudeBot, no PerplexityBot rules. The catch-all User-agent: * disallows /testimonials/ and /car-adviser among a few admin paths. An llms.txt returns 404. The DNS TXT records include an anthropic-domain-verification token, confirming Carmoola has actively engaged with Anthropic for AI crawler verification — yet no corresponding robots.txt rules exist.

Cold-Knowledge Gap

The LLM knows Carmoola as a 2019-founded fintech by ex-Zopa/Wonga executives, offering app-based car loans from £1,000–£35,000 with a "smart credit" virtual card. The actual site tells a different story: Carmoola now offers HP from £2k–£40k and PCP from £2k–£55k (PCP is newly launched, flagged as a "New Feature" on the homepage), plus refinance. The site claims "6,461,674 customers applied" and "4.8 stars on Trustpilot" — the cold LLM only recalled "some customer reviews mention high APR." The LLM referenced a £100M raise in 2023; the site's blog details a £15.5M equity raise in January 2024 plus a prior £95M debt facility and £8.5M Series A, totalling £146M. The LLM also missed the 2025 Car Finance Provider of the Year award prominently displayed in the homepage JSON-LD.

Schema Posture

The homepage carries rich Organization schema with aggregateRating (4.9/5 from 3,270 reviews), award (two named awards), sameAs links to 7 platforms including Wikipedia, and makesOffer for HP, PCP, and refinance. The HP and PCP product pages each embed Service + LoanOrCredit schema with precise minValue/maxValue amounts, annualPercentageRate, and loanTerm. The "How it works" page includes a HowTo schema with 5 named steps and an FAQPage with 7 questions. However, the FAQ page itself (/faqs) has zero JSON-LD — a missed opportunity for the ~20+ FAQ sections rendered in plain HTML. The blog listing page (/blog) returns 0 words of visible text to a plain GET — it is a JS-rendered shell that even AI crawlers receive as empty content, despite individual blog posts rendering full text with BlogPosting schema.

External Signals

The DNS includes anthropic-domain-verification, apple-domain-verification, and a Microsoft verification TXT record. The site references Trustpilot (4.8–4.9 stars), Facebook, Instagram, TikTok, YouTube, LinkedIn, and Wikipedia in its sameAs schema. The investment blog post names QED Investors, VentureFriends, InMotion Ventures (JLR), AlleyCorp, and u.ventures as backers. The blog contains ~1,000 articles across categories like Car Finance, Pop Culture, and Tips & Advice — but the listing page's JS dependency means AI crawlers cannot discover them through the index, only via the sitemap.

Findings

1. Blog listing page renders as empty JS shell to AI crawlers High

   The /blog page returns 0 words of visible text to plain GET requests, delivering ~98KB of HTML but no article content. All AI crawlers (GPTBot, ClaudeBot, PerplexityBot, etc.) receive this empty shell, preventing discovery of 1,000+ blog articles through the listing page.

   What to change: Implement server-side rendering (SSR) or static generation for the blog listing page so that AI crawlers receive full article summaries and links.

2. FAQ page lacks JSON-LD schema Medium

   The /faqs page contains ~20+ FAQ sections in plain HTML but has zero JSON-LD markup, missing an opportunity for rich results in AI-generated answers.

   What to change: Add FAQPage JSON-LD schema to the /faqs page covering all questions and answers.

3. llms.txt file returns 404 Low

   The site does not provide an llms.txt file, which is a recommended standard for guiding AI crawlers to key content.

   What to change: Create an llms.txt file listing important pages like /how-it-works, /faqs, and /blog.

4. LLM cold knowledge is outdated on product range and funding Medium

   The LLM recalled Carmoola as offering loans from £1,000–£35,000, but the site now offers HP (£2k–£40k) and PCP (£2k–£55k). The LLM referenced a £100M raise in 2023, while the site details a £15.5M equity raise in January 2024 plus prior rounds totalling £146M. The 2025 Car Finance Provider of the Year award was also missed.

   What to change: Ensure key updates (product range, funding, awards) are prominently featured in crawlable content and linked from authoritative sources like Wikipedia or Crunchbase.

5. No AI-specific directives in robots.txt despite Anthropic verification Low

   The robots.txt contains zero AI bot rules (no GPTBot, ClaudeBot, PerplexityBot), even though the DNS includes an anthropic-domain-verification token indicating active engagement with Anthropic.

   What to change: Add explicit Allow/Disallow rules for AI crawlers in robots.txt to signal which content should be crawled.

6. Blog posts are not discoverable via listing page by AI crawlers High

   Because the blog listing page is a JS shell, AI crawlers cannot navigate from it to individual posts. Discovery relies solely on the sitemap, which may not be crawled by all AI bots.

   What to change: Ensure the blog listing page renders full HTML with links to all articles, or provide an alternative index page for crawlers.

What's working

• All 11 tested AI bots receive full content on core pages — The homepage and product pages return 200 with full content to all tested AI crawlers, with byte sizes matching browser baseline within 0.003%.
• Homepage has rich Organization schema with ratings and awards — The homepage includes Organization schema with aggregateRating (4.9/5 from 3,270 reviews), two named awards, sameAs links to 7 platforms, and makesOffer for HP, PCP, and refinance.
• HP and PCP product pages have detailed LoanOrCredit schema — The HP and PCP pages embed Service + LoanOrCredit schema with precise min/max amounts, APR, and loan terms.
• How it works page has HowTo and FAQPage schema — The /how-it-works page includes a HowTo schema with 5 steps and an FAQPage with 7 questions.
• Individual blog posts have BlogPosting schema — Individual blog articles like /blog/pcp-vs-hp-car-finance render full text with BlogPosting schema.
• DNS includes anthropic-domain-verification token — The DNS TXT records contain an anthropic-domain-verification token, indicating proactive engagement with Anthropic for AI crawler verification.
• Trustpilot ratings prominently displayed and in schema — The site displays 4.8–4.9 stars from Trustpilot and includes aggregateRating in schema, building credibility for AI-generated summaries.