How CogniPrep built a scalable content cluster by keeping copy out of components

CogniPrep, a candidate prep platform, had two product pages — /interview and /assessment-centre-practice — with no supporting content cluster, internal links, or search visibility. To fix this, the team built a structured content architecture with a clear rule: shared components may own markup, mobile behaviour, and structured data, but never the copy a visitor reads. All user-facing text is written at the page level, preventing the common failure where components accumulate variant props and default headings until no one can trace where a paragraph originates. A central metadata registry drives hub cards, sitemaps, and sibling navigation, while a single utility function generates consistent Next.js metadata across every guide in the cluster. The result is a system where adding a new guide requires only one registry entry and one folder, with build-time errors catching omissions before they reach production.