The agent successfully gathered enough information to explain Controlaccount's core business, target customers, and positioning, but hit a significant wall on pricing transparency. The site's main pages (homepage, solutions, services list, FAQs) contained clear messaging about what they do and who they serve, but intentionally published no pricing models or ranges. The agent assembled a reasonable explanation by combining explicit site content with reasonable inference (commission-based, no-collect-no-fee) but could not verify specific cost structures without contacting the company.

• ›Steps [1] and [6] (homepage fetches) provided the core positioning: '45+ years in business, recovering £80M+ annually, credit management and debt recovery outsourcing.' This was the backbone of the 'what they do' and 'who they are' sections.
• ›Steps [11], [12] (debt recovery and receivables management solution pages) detailed their service offerings and process (pre-legal, legal, tracing), but returned truncated HTML—the agent inferred the breadth of services from fragments and the '50+ services' claim rather than seeing a complete service matrix.
• ›Step [13] (FAQs page) reinforced the 45-year history, employee ownership, and ethical positioning but added little new detail about pricing or differentiation beyond what the homepage stated.
• ›Step [18] (services-list page) was fetched but the response was truncated; the agent could not access the full service catalog despite the page existing.
• ›The site exposed NO pricing page, NO pricing examples, NO commission ranges, and NO cost-benefit comparison. The agent filled this gap by inferring 'no-collect-no-fee' and 'commission-based' from domain expertise and web search context, not from explicit site content.
• ›The agent had to rely on web search (steps [8], [15], [16], [19], [21]) to find LinkedIn, Companies House filings, and competitor context—none of which appeared in direct site navigation. The site's IA (information architecture) does not surface these resources; the agent used prior knowledge of where to search for such data.
• ›The site's Webflow-based structure returned clean HTML for main pages but provided no machine-readable schema (no JSON-LD, no microdata) to help an agent extract structured data about services, pricing, or company info programmatically.
• ›Navigation was poor: the agent tried /services, /pricing, /about (all 404s) before discovering the correct URL structure (/solutions/debt-recovery, /services-list) through web search results, not site navigation.