Nielsen Norman Group: usability and website design

Evidence class: practitioner guidance, with cited observational research and a problem-discovery model. Not a universal controlled experiment. Read: 2026-09-13. Coverage: the specific articles below, not the complete NN/g corpus.

Usability and mental models

Nielsen's heuristics cover visible state, familiar concepts, control, consistency, preventing and recovering from mistakes, recognition, efficient use, focused presentation, and usable help. They are inspection prompts rather than a certification. Source: Ten usability heuristics, read 2026-09-13.

Mental models are users' beliefs about how a system works. A mismatch can require changing the structure or explaining the distinction better. Our application: ask a participant to predict the scope and result before acting, especially around sharing, search, and browser Back. Source: Mental models, read 2026-09-13. The page identifies its current text as an update to Nielsen's earlier article, not an unchanged historical original.

Navigation and complexity

Progressive disclosure defers less common capabilities while leaving frequent ones visible. Its secondary entry must remain findable. Staged disclosure is different: everyone progresses through task stages. Interdependent choices may need simultaneous visibility. Source: Progressive disclosure, read 2026-09-13.

Navigation should be assessed through meaningful labels and wayfinding, not a universal click maximum. Both excessive breadth and depth can obstruct finding. Source: The three-click rule, read 2026-09-13. This article is a first-party expert review of the claim; this pass did not independently reproduce the studies it discusses.

Visual expression and implementation

Scale, hierarchy, balance, contrast, and grouping are tools for directing attention. Our application: define the main object, needed context, and next action before choosing decoration. Source: Principles of visual design, read 2026-09-13.

A design system includes reusable patterns and components plus the rules for using them, with maintenance costs. Our application: stabilize components that real journeys reuse, including their interaction states, rather than build a speculative catalog. Source: Design systems 101, read 2026-09-13.

Testing and limits

The small-sample recommendation is about iterative qualitative discovery with comparable users and tasks. The source explicitly calls for coverage of distinct user groups. It does not guarantee finding every problem. Source: Testing with five users, read 2026-09-13.

Quantitative study size depends on the measurement and desired precision; do not transfer qualitative discovery advice into an uplift estimate. Source: Quantitative study sizes, read 2026-09-13.

Tension to preserve: reducing the initial display can improve focus while making hidden capabilities harder to find. Our default is task-based grouping with discoverable secondary controls. Test common and uncommon essential tasks rather than treat minimalism as an end in itself.