The agent successfully gathered comprehensive information about VouchedFor by working around a largely inaccessible main website. The homepage (step 1) provided only metadata and a description; actual content lived on separate support and about subdomains (steps 11, 12, 14) and required web search to discover. The agent synthesized pricing, positioning, competitor comparisons, and use cases, but had to rely heavily on prior knowledge and search results rather than the site's own navigation—indicating the main domain is poorly structured for crawlers or direct information retrieval.

• ›The main website (vouchedfor.co.uk) returned only a 200 status with minimal visible content; all standard information architecture paths (/about-us, /pricing, /how-it-works, /for-advisers, /faq) returned 404s, forcing the agent to search for and fetch separate domains (about.vouchedfor.co.uk, support.vouchedfor.co.uk).
• ›The support.vouchedfor.co.uk knowledge base articles (steps 12, 14) contained actual pricing and membership tier details; these were only discoverable via web search, not through links on the main site.
• ›Step 18 (FAQ search result) pointed to about.vouchedfor.co.uk/faqs, which was cited in the final sources but the actual FAQ content is not visible in the trajectory—agent relied on prior knowledge and search results to construct the 'things that were confusing' section.
• ›The agent had to infer competitor positioning (Unbiased vs. VouchedFor comparison) from external articles and prior knowledge; the site itself does not document how it differs from alternatives.
• ›Pricing information for advisers (monthly subscription costs, per-enquiry fees) was fragmented across multiple support articles and not presented as a cohesive table or pricing page on the main domain.
• ›The site's use of JavaScript-heavy front-end rendering meant even the homepage content was not fully exposed in the HTML response, limiting what a crawler could extract without rendering.