The user's language replaces the system's taxonomy
The user expresses the need in their own words and expects the system to understand it: growing friction with every interface that asks to translate the need into a navigation path or into the “correct” term. Navigation structures and labels must be validated on the natural language actually used, not on the internal vocabulary.
The decision this observation informs
The decision is whether labels follow the user’s vocabulary or the company’s. It falls under the INTERPRET position. Without an explicit choice, the company’s wins, because taxonomies are inherited from internal systems — catalogue, ERP, org chart — and nobody holds the mandate to rewrite them: users’ real vocabulary, moving towards the spoken register, keeps diverging without the divergence being measured.
Adopting the user’s language means giving up the terminological precision that in some sectors is a value or an obligation — pharmaceutical, financial, technical, legal — and misaligning the interface from the upstream systems, at a permanent translation cost. Keeping the corporate language means the user does not find things, and that AI acts as the translator: the product’s vocabulary gets normalised by an intermediary, not by the organisation. The choice belongs to the Information Architect, but only with an explicit mandate to change naming that belongs to other functions; without that mandate it cannot be made and stays a research observation.
Effects and role responsibilities
- Information Architect — Menus, labels, and taxonomies must be validated on the real vocabulary of users, which shifts toward the spoken register.
- Product Manager — Feature discovery and onboarding must presuppose that the user does not read instructions and expects to understand by asking.
- UX Writer / Copywriter — In a conversation words are the interface: tone, word choice, and response structure determine the quality of the experience.