The agent successfully assembled a comprehensive overview of Monzo by combining direct fetches from the site's key pages (/current-account/plans, /current-account/extra, /current-account/perks) with web search results that filled gaps the site's own pages did not adequately expose. The site required a hybrid discovery approach: pricing tiers and product differentiation were fetchable from dedicated landing pages, but competitive positioning and detailed feature comparisons had to be sourced externally because Monzo's own pages did not render complete content or provide comparative data.

• ›Steps [12], [13], [14] returned the core pricing and tier information (Free, Extra £3/mo, Perks £9/mo, Max £19+/mo, and business plans). These pages were directly discoverable via web search [9] but were not linked prominently from the homepage or /about page, requiring the agent to search rather than navigate the site's own IA.
• ›The agent's early attempts to guess natural URL patterns (/pricing, /accounts, /features) all returned 404s, indicating Monzo does not use standard naming conventions—a friction point that forced reliance on search and prior knowledge.
• ›Steps [16] and [17] (external comparisons with Revolut and Starling) were necessary because Monzo's site itself does not publish comparative positioning; the agent had to go off-domain to answer 'how it differs from alternatives,' suggesting the site treats comparison as out-of-scope.
• ›Pots, the signature differentiator, was mentioned on the site pages but lacked implementation details (how to create them, what interest rates apply). The agent filled this from off-site sources, not from Monzo's own content.
• ›The site's JavaScript-heavy rendering and truncated responses in fetches [1], [3], [7], [12-14] mean structured data about features, rates, and comparisons was not fully accessible in the raw HTML, reducing machine-readability.