UI design gives an interaction its visible and operable form. Typography, spacing, colour, imagery, wording and behaviour work together. A polished screen is one part of the experience; the task, context and implementation determine what needs to be evaluated.
Sources checked 18 October 2026 · 5 resources
What to examine
Hierarchy and typography: can people identify the purpose, scan the content and distinguish headings, explanations and actions? Check reading order and enlarged text, not just the default screenshot.
Layout and responsive behaviour: preserve relationships as the viewport changes. Review wrapping, content length and spacing on both narrow and wide views.
Colour, imagery and meaning: consider contrast, emphasis and cultural interpretation. Do not use colour as the only way to identify a state or action.
Interaction states: design focus, selection, loading, empty results, errors, success and recovery—not just the ideal populated screen.
Components and systems: document what repeats and when a component should be used. Consistency should support understanding rather than conceal differences in consequences.
Words and feedback: make labels and messages part of the design. Check whether users can predict an action and understand what happened afterwards.
These are editorial review prompts. A visual inspection alone cannot establish usability, accessibility or the quality of the complete experience.
Research and practical references
Two background journal papers and three official guidance resources. The aesthetics papers are included as a pair to encourage scrutiny rather than repeat “beautiful is usable” as a slogan.
Publisher-indexed abstract reviewed; full text not reviewed
Tractinsky, Katz and Ikar examine relationships between perceptions of an interface’s aesthetics and usability. A starting point for asking how visual presentation influences judgement.
Limitations: Perceived usability is not the same as successful task performance. The title should not be treated as a universal design rule; this review is restricted to the abstract.
Publisher-indexed abstract excerpt reviewed; full text not reviewed
Examines the relationship between aesthetics, usability and affect under experimental manipulations. The abstract challenges a simple, one-direction account of beauty determining perceived usability.
Limitations: The reported relationship depends on the study conditions. Full methods and results were not reviewed here; do not infer a general effect size or causal rule for your interface.
Examples of layout, typography, colour, components and task patterns developed for government services. Useful for examining consistency and documented component behaviour.
Limitations: Its context and conventions are those of GOV.UK services. Adopting components does not remove the need to investigate your users, test the implementation or assess accessibility.
Practice: revise one interface, explain every change
Editorial learning activity · use your own interface or a clearly attributed public example
Define the task and context. For example, finding an exhibition time and understanding access information on a phone. List what a visitor needs to know before choosing.
Revise the hierarchy and states. Produce narrow and wide versions, plus a loading, empty or error state. Annotate changes to headings, type, spacing, labels and controls. Include keyboard focus and a plan for checking enlarged text.
Separate appearance from performance. Plan how to assess visual impressions and task success separately. After any required ethics approval, evaluate the design in use; otherwise present an inspection and test plan, not invented user findings.
Portfolio output: before-and-after screens, an annotated component or state specification, reasons grounded in sources, and an evaluation plan. Explain where aesthetics support the experience and where further evidence is needed.
For experienced practitioners: compare the proposed change with your existing design system. Document the implementation and maintenance consequences, including any exception you recommend.