The ten usability heuristics published by Jakob Nielsen in the 1990s are still one of the fastest tools for evaluating an interface. They are not rigid rules or a checklist to tick off: they are general principles that help you spot problems before you even involve users. Their value lies in applying them to concrete cases, which is what we will do here.
1. Visibility of system status
Users should always know what is going on. A «Submit» button that does not change appearance or show anything once pressed invites people to press it again, with the risk of duplicate orders. It is enough to disable it, show a progress indicator and then a clear confirmation message.
2. Match between the system and the real world
The interface should speak the language of the people using it, not of those who built it. «Error 500: unhandled exception» means nothing to someone booking an appointment; «We couldn't save your booking, please try again in a few minutes» does. The same goes for the order of information: an address form should follow the order in which people write an address in their own country.
3. User control and freedom
People make mistakes, click by accident, change their minds. They need obvious ways out: undo, go back, close. A good example is the message that appears after deleting an email, offering «Undo» for a few seconds: it is more respectful than a confirmation dialog that interrupts every single time.
4. Consistency and standards
If the primary button is blue on the right on one page and grey on the left on another, users have to relearn the interface on every screen. Consistency applies both inside the product and to external conventions: the logo at the top left leading back to the home page, the cart at the top right, the magnifying glass for search. Reinventing these elements rarely pays off.
5. Error prevention
Preventing an error is better than explaining it well. A date picker that does not let you choose a return date earlier than the departure date removes an entire category of problems. Likewise, asking for confirmation only for destructive and irreversible actions, such as deleting an account, works better than asking all the time.
6. Recognition rather than recall
Short-term memory is limited. An interface that forces you to remember a code seen on the previous screen demands needless effort. Showing recent searches, viewed products and available options in a menu, instead of expecting commands typed from memory, are direct applications of this principle.
7. Flexibility and efficiency of use
A good system serves both beginners and experts. Keyboard shortcuts, saved templates and quick actions are accelerators that daily users appreciate, but they must not get in the way of someone opening the product for the first time. The key is that they stay optional.
8. Aesthetic and minimalist design
Every extra element competes for attention with the important ones. Minimalism does not mean empty pages, but the absence of irrelevant information. A product page with ten promotional banners hides precisely the button the user is looking for.
9. Help users recognize, diagnose and recover from errors
A useful error message says what went wrong, why, and how to fix it, in plain language. «Invalid password» is not much; «Your password must be at least 12 characters long: you entered 9» lets people fix it right away.
10. Help and documentation
Ideally an interface needs no instructions, but some complex tasks do. In that case help should be contextual, short and task-oriented: a hint next to a tricky field is worth more than a fifty-page manual.
How to run a heuristic evaluation
A heuristic evaluation can be done in a few hours. The method I use most often in my work is this:
- Define two or three typical tasks, for example «sign up and complete the first purchase».
- Have two or three evaluators go through the interface separately, noting every problem and the heuristic it violates.
- Assign each problem a severity, from cosmetic to blocking.
- Merge the results, remove duplicates and sort by severity.
Working independently matters: different evaluators tend to notice different problems, and the final comparison produces a more complete picture.
| Severity | Meaning | Example |
|---|---|---|
| 1 | Cosmetic | Slightly misaligned icon |
| 2 | Minor | Ambiguous label that context makes clear |
| 3 | Major | No message after the form is submitted |
| 4 | Blocking | Payment cannot be completed on mobile |
The limits of the method
Heuristic evaluation does not replace testing with real users. Evaluators are experts and do not have the same difficulties or expectations as the people who will use the product. It also tends to flag many small issues and miss those tied to the real context of use. It works best as a first filter: it removes the obvious mistakes, so that tests with people can focus on what only they can reveal.
In short
- The ten heuristics are general principles, useful for finding problems quickly and cheaply.
- Each heuristic becomes effective once translated into concrete examples from your own product.
- An evaluation with several independent evaluators and a severity scale makes results comparable.
- The method is a first filter: it does not replace observing real users.
