In short
A heuristic evaluation is a method in which a small number of evaluators independently inspect an interface against a set of recognised usability principles, most commonly Jakob Nielsen's ten heuristics, record every violation with its location and severity, and combine their findings. It is fast, cheap, and needs no participants, which makes it the right first pass on any product; three to five evaluators find most of the problems a single one misses. It does not reveal whether real users can complete real tasks, which only usability testing shows.
What it is
An expert review with structure. Instead of a designer looking at a product and saying what they dislike, evaluators work through it screen by screen, task by task, checking it against a fixed list of usability principles and recording each place a principle is broken. Each finding gets a location, the heuristic violated, a description, and a severity rating. The findings from several evaluators are merged into one prioritised list.
The method was formalised by Jakob Nielsen and Rolf Molich in 1990 and it remains the standard inspection technique because it is cheap, quick, and reliably productive.
Nielsen's ten heuristics
- Visibility of system status. The interface keeps users informed about what is happening: loading states, progress, confirmation that an action worked.
- Match between system and the real world. Language and concepts users recognise, in a natural order, rather than internal terms.
- User control and freedom. Undo, cancel, back, an exit from any state entered by mistake.
- Consistency and standards. The same thing looks and behaves the same way everywhere, and follows platform conventions.
- Error prevention. Design that stops mistakes before they happen: constraints, confirmations for destructive actions, sensible defaults.
- Recognition rather than recall. Options, actions and information visible or easily retrieved, so users do not have to remember them.
- Flexibility and efficiency of use. Shortcuts and accelerators for experienced users that do not get in the way of novices.
- Aesthetic and minimalist design. No irrelevant or rarely needed information competing with what matters.
- Help users recognise, diagnose and recover from errors. Plain-language error messages that say what went wrong and how to fix it.
- Help and documentation. Available, searchable, task-focused, and rarely needed if the rest is right.
Teams often add domain-specific heuristics, such as accessibility criteria or platform guidelines, and that is legitimate provided the list is fixed before the evaluation begins.
How to run one
- Define the scope: which parts of the product, which tasks a typical user would perform. Give evaluators the tasks so they inspect the interface in use, not as a gallery of screens.
- Choose the heuristics and share them with examples.
- Recruit three to five evaluators. Nielsen's research found a single evaluator catches roughly a third of the problems, and that three to five together catch most of them, with diminishing returns beyond. Evaluators should have usability expertise; domain expertise helps but is secondary.
- Evaluate independently. Each evaluator works through the tasks alone, at least twice: once to get a feel, once to inspect in detail. Each records findings in a shared template: screen, element, heuristic, description, severity.
- Rate severity on a consistent scale, typically 0 (not a problem) to 4 (catastrophic, must fix before release), considering frequency, impact and persistence.
- Merge and deduplicate the findings in a debrief, agree severities, and produce a prioritised list with recommendations.
- Fix and re-evaluate the most severe items.
A product of moderate size takes each evaluator half a day to a day, plus a debrief. It is one of the cheapest ways to find usability problems that exists.
Strengths
No participants to recruit; it can run in a week. It finds the obvious violations that usability tests then confirm, freeing the test to look at harder questions. It gives findings a shared vocabulary, so "it feels clunky" becomes "violates visibility of system status on the upload screen: no progress indicator." And it works on prototypes, mockups and competitors' products, not just finished builds.
Blind spots
It does not involve users, so it cannot tell you whether people can complete tasks, what they expect, or which problems actually matter to them. Evaluators find violations; they cannot always predict which violations users will notice or be stopped by, and they tend to report more false positives than tests do. It is also blind to problems outside the interface: whether the product solves the right problem, whether the content is right, whether the flow matches how people work.
The standard pairing is a heuristic evaluation first, to clear the obvious problems cheaply, then a usability test with five people on the tasks that remain. Doing the test first wastes participants on problems an expert would have found in an hour; doing only the evaluation ships a product nobody has watched anyone use.
Related methods
A cognitive walkthrough inspects a specific task step by step, asking at each step whether a new user would know what to do and see that it worked; it is narrower and more predictive of learnability. A design audit is broader, covering visual consistency and brand as well as usability. An accessibility audit applies WCAG criteria instead of usability heuristics and is often run alongside.
If you would like a heuristic evaluation of a product or a prototype, with three evaluators and a prioritised list within a week, book a call.
Common questions
What are Nielsen's 10 usability heuristics?
Visibility of system status; match between system and the real world; user control and freedom; consistency and standards; error prevention; recognition rather than recall; flexibility and efficiency of use; aesthetic and minimalist design; help users recognise, diagnose and recover from errors; and help and documentation. They are the most widely used checklist for heuristic evaluations.
How many evaluators do you need for a heuristic evaluation?
Three to five. Nielsen's research found a single evaluator identifies roughly a third of the usability problems, while three to five working independently find most of them, with little gain beyond five. Evaluators should inspect separately and then merge their findings to avoid influencing each other.
What is the difference between a heuristic evaluation and usability testing?
A heuristic evaluation is an inspection by experts against usability principles, with no users involved; it is fast and cheap and finds obvious violations. Usability testing observes real users attempting real tasks and reveals whether they succeed and where they struggle. The evaluation is the first pass; testing confirms what matters and finds what experts cannot predict.
How do you rate severity in a heuristic evaluation?
On a fixed scale, commonly 0 to 4: not a problem, cosmetic, minor, major, and catastrophic. Consider how often the problem occurs, how much it impacts the user when it does, and whether users can work around it once learned. Evaluators rate independently and agree final severities in the debrief.
Can a heuristic evaluation be done on a prototype?
Yes, and it is one of the method's advantages: it works on clickable prototypes, mockups and even competitors' products, since it requires only an interface to inspect and a set of principles. Running it on a prototype finds problems before they are built, when they are cheapest to fix.
