capital.com — AI Site Grade

Analysis

Capital.com: AI crawlers get full access to a site that structurally cannot be understood by them

Every AI crawler tested — GPTBot, ClaudeBot, PerplexityBot, Google-Extended, OAI-SearchBot, Bytespider, Applebot-Extended — receives a 200 with identical byte payload (523 KB) as a browser. No UA-based blocking exists. Yet the site is functionally opaque to machine extraction: zero JSON-LD schemas appear on any page tested (homepage, about, fees, learn, markets, analysis), the robots.txt contains no AI-bot directives whatsoever, and /llms.txt silently redirects to the homepage (no file exists). The homepage is a JS-heavy single-page application shell that, while rendering ~1,200 words of visible text to a plain GET, delivers no structured entity data — no Organization, WebSite, FinancialService, or FAQPage schema.

Crawler Access & Infrastructure

Hosted on Cloudflare (NS: jeff.ns.cloudflare.com, liz.ns.cloudflare.com; IPs: 45.60.76.121, 45.60.85.121). The robots.txt uses a single User-agent: * block that disallows /*? (all query-string URLs except ?page=) and /*/news/ but allows /*/news$. No AI-specific rules exist — GPTBot, ClaudeBot, and all others fall under the catch-all. The compare_bot_access test confirms all 11 UAs get identical 200 responses with full content. The DNS TXT records reveal OpenAI domain verification (openai-domain-verification=dv-cq6K6t6MBLMBJXrMpIelw66m), Anthropic verification (anthropic-domain-verification-bkk5dc=...), and Perplexity verification (perplexity-ai-domain-verification-8hpzsd=...), indicating the brand has proactively registered with all three major AI providers — yet has not created an llms.txt to guide what those crawlers should consume.

Content & Schema Posture

The homepage carries a meta description claiming "OMA'S Best Overall Trading Platform 2024" and "857,000+ global traders," but no structured data backs any of these claims. The about page describes a thoughtful brand philosophy ("built for judgement, not speed") with 1,041 words of narrative, yet has zero schema markup. The fees page contains a detailed table of overnight funding rates, spreads, and GSL premiums — ideal content for AI answer extraction — but it is rendered as plain HTML with no Table or FinancialProduct schema. The learn section offers structured educational content (trading strategies, technical analysis, psychology) that could power AI-generated answers, but again no FAQPage or HowTo schema. The analysis section publishes timely market commentary (articles dated May 2026) that would be valuable for AI retrieval, but none of it is surfaced through structured data.

Cold-Knowledge Gap

The LLM prior knows Capital.com as "a retail-focused online trading platform offering CFDs on forex, indices, commodities, shares, and cryptocurrencies" regulated by FCA and CySEC, founded in 2016. It mentions the "Investmate" educational app and notes "no major scandals." The gap: the LLM knows nothing about the brand's core differentiator — the "decision environment" philosophy, the 1X unleveraged CFD product, the OMA award, the 857K account figure, or the specific 5,500+ instrument count. The site's most distinctive positioning ("built for judgement, not speed") is entirely absent from the model's prior. The model also references cryptocurrency trading, which the site's footer explicitly restricts ("Crypto Derivatives are not available to Retail clients registered with Capital Com (UK) Ltd").

Broken Internal Architecture

Multiple pages linked from the global navigation return 404: /en-gb/about-us/press-centre, /en-gb/about-us/is-capital-com-safe, and /en-gb/about-us/company. These are not obscure pages — they appear in the footer and navigation of every page on the site. The sitemap is a sprawling index of 15+ locale-specific sub-sitemaps (ar, de, el, en, es, fr, hu, mn, nl, pl, ro, ru, vi, zh) each with content and instrument sub-sitemaps, suggesting a massive multi-region content operation. Yet the English UK site has broken trust signals and no press page. The homepage also links to /en-gb/about-us/whitepaper and /en-gb/about-us/investor-relations — neither was tested, but the pattern of broken about-subpages raises questions about content maintenance.

Findings

1. Zero JSON-LD schema on any tested page High

   No structured data (Organization, WebSite, FinancialService, FAQPage, etc.) appears on the homepage, about, fees, learn, markets, or analysis pages. This prevents AI crawlers from extracting entity relationships, claims, and financial product details.

   What to change: Add JSON-LD structured data for Organization, WebSite, FinancialService, and FAQPage schemas to all key pages, including award claims, account numbers, and fee tables.

2. Missing /llms.txt file High

   The /llms.txt endpoint silently redirects to the homepage instead of serving a file. This is a missed opportunity to guide AI crawlers to the most important content, especially given the brand has already verified with OpenAI, Anthropic, and Perplexity.

   What to change: Create an /llms.txt file that lists key pages (about, fees, learn, analysis) and provides a brief summary of the site for AI consumption.

3. No AI-specific directives in robots.txt Medium

   The robots.txt contains only a catch-all User-agent: * block with no rules for GPTBot, ClaudeBot, PerplexityBot, or other AI crawlers. While this allows full access, it also means no guidance or rate-limiting is provided for AI bots.

   What to change: Add explicit directives for AI crawlers (e.g., GPTBot, ClaudeBot) to allow or disallow specific paths, and consider rate-limiting to manage crawl load.

4. LLM prior lacks key brand differentiators High

   The LLM prior knows Capital.com as a generic CFD broker but is unaware of its core philosophy ('built for judgement, not speed'), the 1X unleveraged CFD product, the OMA award, the 857K account figure, and the 5,500+ instrument count. These differentiators are absent from the site's structured data and meta content.

   What to change: Embed these differentiators in JSON-LD schema (e.g., award, numberOfEmployees, description) and in meta tags to improve LLM knowledge extraction.

5. Multiple about subpages return 404 Medium

   Pages linked from the global navigation and footer, including /en-gb/about-us/press-centre, /en-gb/about-us/is-capital-com-safe, and /en-gb/about-us/company, return 404 errors. This undermines trust and wastes crawl budget.

   What to change: Restore these pages with meaningful content or implement proper 301 redirects to relevant existing pages.

6. Educational content lacks FAQPage schema Medium

   The learn section contains structured educational articles (trading strategies, technical analysis) that could power AI-generated answers, but none use FAQPage or HowTo schema. This reduces the likelihood of content being featured in AI answer boxes.

   What to change: Add FAQPage and HowTo schema to educational articles to increase chances of appearing in AI answer snippets.

7. Fees page lacks FinancialProduct schema Medium

   The fees page contains detailed tables of overnight funding rates, spreads, and GSL premiums, but no FinancialProduct or Table schema. This data is ideal for AI extraction but is not machine-readable.

   What to change: Add FinancialProduct schema to fee tables and consider using Table schema for structured data extraction.

8. About page lacks Organization schema Medium

   The about page describes the brand philosophy and history but has no Organization schema to provide entity details like founding date, number of employees, or social media links.

   What to change: Add Organization schema with name, description, founding date, number of employees, and award information.

9. Homepage lacks WebSite schema Low

   The homepage has no WebSite schema to define the site name, URL, and search action. This is a basic structured data element that helps search engines and AI understand the site identity.

   What to change: Add WebSite schema with name, url, and potentialAction for search.

10. LLM prior incorrectly mentions cryptocurrency trading Medium

    The LLM prior states Capital.com offers cryptocurrency trading, but the site's footer explicitly restricts crypto derivatives for UK retail clients. This discrepancy could lead to misinformation in AI-generated answers.

    What to change: Update the site's meta descriptions and structured data to clarify the availability of crypto products per region, and consider adding a disclaimer in schema.

11. Press centre page returns 404 Low

    The press centre page, linked from the footer, returns a 404 error. This is a trust signal issue and may affect media and AI crawler perception.

    What to change: Create a press centre page with press releases and media mentions, or redirect to a relevant page.

12. Award claims not backed by web search results Low

    Web searches for 'capital.com OMA best overall trading platform 2024 award' and similar queries returned zero results. The award claim on the homepage lacks external verification, which may reduce credibility for AI systems.

    What to change: Ensure award information is published on reputable third-party sites and linked from the homepage to provide external validation.

What's working

• All AI crawlers receive full content access — Every tested AI crawler (GPTBot, ClaudeBot, PerplexityBot, etc.) receives a 200 response with the same content as a browser. No UA-based blocking or cloaking is present.
• Proactive verification with OpenAI, Anthropic, and Perplexity — DNS TXT records show domain verification for OpenAI, Anthropic, and Perplexity, indicating the brand has proactively registered with major AI providers.
• Comprehensive multi-locale sitemap index — The sitemap index contains 15+ locale-specific sub-sitemaps (ar, de, el, en, es, etc.) with content and instrument sub-sitemaps, indicating a well-organized content operation.
• Rich educational content in learn section — The learn section offers structured articles on trading strategies, technical analysis, and psychology, which are valuable for AI retrieval and answer generation.
• Detailed fees and charges page — The fees page contains comprehensive tables of overnight funding rates, spreads, and GSL premiums, providing transparent pricing information that is ideal for AI extraction once schema is added.
• Timely market analysis content — The analysis section publishes current market commentary (articles dated May 2026), which is valuable for AI retrieval and can position the site as an authoritative source.
• Clear brand philosophy articulated on about page — The about page clearly communicates the brand's philosophy ('built for judgement, not speed') and differentiators, providing strong narrative content for AI to reference once structured data is added.
• Robots.txt disallows query-string URLs to prevent crawl waste — The robots.txt disallows all query-string URLs (except ?page=), which helps prevent crawl budget waste on duplicate or parameterized pages.