UX Heuristic Evaluation: How to Evaluate a Website or App

A heuristic evaluation is an expert review of an interface against recognized usability principles, most often Jakob Nielsen's 10 usability heuristics. To run one, define the scope and key tasks, then have three to five evaluators review the product independently. Each evaluator makes one pass to learn the product and a second pass to record violations with their location and evidence. Afterwards, merge the findings, rate each problem's severity on a shared scale and report them in priority order. It's fast and inexpensive, but it predicts problems rather than proving them, so confirm the important ones with usability testing.

Heuristic evaluation is a usability inspection method developed by Jakob Nielsen and Rolf Molich. Evaluators compare an interface with a short list of broad rules of thumb, the heuristics, and note where the design breaks them. No users are involved, which is why it's quick to run on wireframes, prototypes or live products.

It fits into a larger evaluation toolkit. It's one of the main methods inside a UX audit, and it's often used to decide what to test with users next.

The two methods answer different questions and work best together.

A cognitive walkthrough is another inspection method. Instead of checking against general principles, evaluators step through one task as a first-time user and ask at each step: will the user know what to do, will they see how to do it, and will they understand the feedback? It's especially useful for onboarding and other flows where learnability matters most. Many teams use a walkthrough for new-user tasks and a heuristic evaluation for the product as a whole.

The most widely used set is Nielsen's 10 usability heuristics. The examples below are typical violations to look for.

Nielsen's list is general by design. For specialist products, supplement it rather than replace it: platform guidelines such as Apple's Human Interface Guidelines and Google's Material Design for apps, Baymard Institute's ecommerce guidelines for online stores, and your own design system rules for consistency. Accessibility should be reviewed against WCAG as a separate pass; it's a conformance standard, not a set of heuristics, and deserves its own checklist.

Whatever you choose, give every evaluator the same list and make sure they understand it before starting. Nielsen Norman Group suggests a short practice round on a simple design to calibrate the team.

Each evaluator works alone. Nielsen Norman Group's process has evaluators go through the interface at least twice. The first pass is for learning the product's flow and scope without judging it. The second pass is for inspecting each screen and interaction against the heuristics and recording every violation. Sessions of one to two hours per evaluator are typical for a focused scope.

Evaluators should record the problem, where it occurs, which heuristic it violates and a screenshot, without discussing findings with others until everyone has finished. Independence is what makes several evaluators more valuable than one.

A finding should let someone who wasn't in the session understand the problem, reproduce it and judge its seriousness.

Once everyone has finished, combine the individual lists. Cluster duplicates, keep the clearest description with all the evidence, and note how many evaluators found each issue. Discuss disagreements: sometimes one evaluator has spotted something real, and sometimes it's a personal preference that should be dropped. The result is one consolidated list of distinct problems.

Send the consolidated list to each evaluator and ask them to rate every problem independently, then average or agree the scores. Nielsen's severity ratings combine frequency, impact and persistence, and also consider market impact.

The following is a hypothetical illustration of how findings from a checkout review might be recorded, not results from a real study.

The same heuristics apply to apps, with extra attention on platform conventions, gestures users can't discover, back-navigation behaviour, touch targets (WCAG 2.2 sets a minimum of 24 by 24 CSS pixels at level AA, and platform guidelines recommend larger), keyboard types for each input, permission requests, and what happens when a session is interrupted or the connection drops. Review on real devices, not only in a desktop simulator. For broader app guidance, see mobile app UX design.

Lead the report with the highest-severity problems, grouped by task so readers see which journeys are most affected. Include the heuristic, evidence and a recommended fix for each. Then prioritize with engineering, weighing severity against reach and effort; the prioritization buckets in how to conduct a UX audit work well here.

Use a heuristic evaluation to clean up obvious problems cheaply, then test the most important tasks with users. Testing a design that still has obvious violations wastes sessions on problems experts could have found. Combine them when a finding is severe but disputed, when a fix is expensive, or when the product serves specialists whose work evaluators don't fully understand. See usability testing for how to run those sessions.

Heuristic evaluation is one of the fastest ways to find usability problems: independent evaluators, a shared set of principles, two passes, merged findings and severity ratings. Use it early and often, and pair it with usability testing for the issues that matter most. For how it fits into a complete review, see the UX audit guide.