What heuristic evaluation is
Heuristic evaluation is a method in which experienced reviewers inspect an interface against a set of recognised usability principles - "heuristics" - and record where it falls short. It does not require recruiting users, which makes it fast and inexpensive, and it can be done on anything from early wireframes to a live product.
The most widely used set is Jakob Nielsen's 10 usability heuristics, first published in 1994. Nielsen Norman Group notes that the heuristics themselves have remained unchanged since then, even as examples have been updated - a sign of how fundamental they are.
The ten heuristics, in practical terms
| Heuristic | What it asks | Common failure in B2B and SaaS |
|---|---|---|
| Visibility of system status | Does the product show what's happening? | Uploads and long processes with no progress or confirmation |
| Match between system and real world | Does it use the user's language? | Internal jargon and database field names in the UI |
| User control and freedom | Can users undo and exit easily? | No undo for destructive actions; modals with no clear exit |
| Consistency and standards | Do the same things behave the same way? | Different patterns for the same action across modules |
| Error prevention | Does design stop errors happening? | Free-text fields where a picker would prevent mistakes |
| Recognition rather than recall | Is information visible when needed? | Users must remember IDs or specs from another screen |
| Flexibility and efficiency | Can experts work faster? | No shortcuts, bulk actions or saved views for power users |
| Aesthetic and minimalist design | Is only relevant information shown? | Dashboards where everything has equal weight |
| Help users recover from errors | Are errors clear and fixable? | "Something went wrong" with no cause or next step |
| Help and documentation | Is help available where it's needed? | Help hidden in a separate portal, not in context |
Why more than one reviewer helps
Any single evaluator, however experienced, misses issues. Nielsen Norman Group recommends that three to five people evaluate the interface independently and then combine findings. In practice I pair my review with at least one person who knows your domain - a product manager, support lead or experienced user - because domain knowledge catches problems a generalist can miss, especially in healthcare, finance and industrial software.
Strengths and limits
| Heuristic evaluation is good at | It is less good at |
|---|---|
| Finding many issues quickly and cheaply | Telling you how often real users hit each issue |
| Reviewing early designs before build | Uncovering problems specific to users' real context |
| Explaining why something is a problem | Measuring business impact on its own |
| Prioritising fixes across a whole product | Replacing usability testing for critical flows |
That is why I often combine it with funnel data and a small round of usability testing - see UX audits - to confirm which findings matter most.
