Origin and positioning: from a 1990 paper to the 10 rules of thumb fixed in 1994
Heuristic evaluation was proposed by Jakob Nielsen and Rolf Molich in a 1990 CHI paper: have evaluators inspect an interface against a small set of general usability principles, item by item, and flag every violation.
In 1994, Nielsen ran a factor analysis of 249 real usability problems and refined the principles into the 10 with the greatest explanatory power — today's 'Nielsen's 10 usability heuristics.' NN/g is explicit: these 10 have remained unchanged since 1994.
The ten at a glance: official names + one-line definitions
Let's walk through them, starting with the first 6 — the ones most often used to 'diagnose' interfaces:
- 1 Visibility of system status: 'The design should always keep users informed about what is going on, through appropriate and timely feedback.' Examples: loading progress bars, success messages after submitting, upload percentages. Missing feedback is the most common usability flaw in novice designs.
- 2 Match between the system and the real world: 'The design should speak the users' language — words, phrases, and concepts familiar to the user, rather than internal jargon.' Example: a trash-can icon for deleted items rather than 'logically deleted entries.'
- 3 User control and freedom: users often perform actions by mistake and need a clearly marked 'emergency exit' to leave the unwanted state — undo/redo, a cancel button, dismissible dialogs.
- 4 Consistency and standards: 'Users should not have to wonder whether different words, situations, or actions mean the same thing.' Two layers: internal consistency (the same meaning gets the same wording/control within one product) and external consistency (follow platform and industry conventions). Design systems are consistency engineered at scale.
- 5 Error prevention: 'Good error messages are important, but the best designs carefully prevent problems from occurring in the first place.' Prevent 'slips' with constraints and good defaults; prevent 'mistakes' with clear information.
- 6 Recognition rather than recall: 'Minimize the user's memory load by making elements, actions, and options visible. Users should not have to remember information from one part of the interface to use in another.' Recognition is far easier than recall — dropdown menus, recently-used lists, autocomplete, carrying key info from the previous step into the next.
The remaining four in brief: 7 Flexibility and efficiency of use — provide accelerators, shortcuts hidden from novices, so the design serves both inexperienced and experienced users; 8 Aesthetic and minimalist design — interfaces should not contain irrelevant or rarely needed information; every extra unit of irrelevant information competes with the useful information for attention; 9 Help users recognize, diagnose, and recover from errors — error messages in plain language (no error codes), precisely indicating the problem and constructively suggesting a solution; 10 Help and documentation — best if the system needs no extra explanation, but when needed, provide task-focused, searchable documentation with concrete steps.
How to use them: 3-5 evaluators work independently, then merge — don't count on a single expert
NN/g's standard heuristic evaluation process has three steps:
1. Prepare — recruit 3-5 evaluators, train them on the 10 heuristics, and scope the evaluation narrowly;
2. Evaluate independently — each evaluator spends 1-2 hours alone: first getting familiar with the product, then a second pass specifically hunting for violations of the 10 heuristics;
3. Merge — pool everyone's findings, cluster similar issues with an affinity diagram, and prioritize.
When diagnosing a scenario, cite the official heuristic name in your report so the team shares a common language: an upload with no progress feedback → violates 'Visibility of system status'; an error that only throws a code → violates 'Help users recognize, diagnose, and recover from errors'; a mis-tap with no undo or cancel → violates 'User control and freedom'; one action with three names on three pages → violates 'Consistency and standards.' That's exactly how the following exercises work.
自测 · 学完检查一下
想真正动手做题、记进度、攒连胜？到互动课里练。
Which statement about the origin of Nielsen's 10 usability heuristics is correct?
答案：Heuristic evaluation was proposed by Nielsen and Molich in a 1990 CHI paper; in 1994 Nielsen refined the principles into the 10 with the greatest explanatory power, based on a factor analysis of 249 real usability problems, and they have remained unchanged since
Heuristic evaluation was proposed by Jakob Nielsen and Rolf Molich in a 1990 CHI paper: evaluators inspect an interface against a small set of general usability principles and flag violations. In 1994, Nielsen ran a factor analysis of 249 real usability problems and refined the principles into the 10 with the greatest explanatory power; NN/g states these 10 have remained unchanged since 1994. They are not an annually updated vote-based list, were not decreed by Norman (though they cross-validate with his principles), and are not a mandatory W3C standard. (Sources: NN/g, '10 Usability Heuristics for User Interface Design'; Nielsen & Molich, CHI '90; Nielsen, CHI '94)
True or False: Nielsen's 10 heuristics are a concrete design specification — they prescribe which controls, what spacing, and what colors an interface should use, and copying them guarantees usability.
答案：False
False. NN/g's key positioning of the 10 is that they are 'rules of thumb / broad rules,' not concrete design specifications — they point directions, not pixels. They give you a lens for 'where to look and by what standard to find faults,' not hard rules on controls, spacing, or colors. Treating them as a mandatory spec to copy line by line misunderstands the very word 'heuristic.' (Source: NN/g, '10 Usability Heuristics for User Interface Design')
After clicking 'Upload file,' the page shows nothing at all: no progress bar, no message — the user has no idea whether the upload is running or finished. Which heuristic does this design most directly violate?
答案：Visibility of system status
This violates heuristic 1, 'Visibility of system status': 'the design should always keep users informed about what is going on, through appropriate and timely feedback' — upload percentages, loading progress bars, and success messages are its classic implementations. A system must externalize its state so users can tell whether their action worked; missing feedback is the most common usability flaw in novice designs, and it maps directly onto Norman's principles of feedback and visibility. (Sources: NN/g, '10 Usability Heuristics,' Heuristic 1; Norman, The Design of Everyday Things)
When an operation fails, the interface shows only 'Error 0x80070057' — no explanation of what went wrong and no suggested fix. Which heuristic does this most directly violate?
答案：Help users recognize, diagnose, and recover from errors
This violates heuristic 9, 'Help users recognize, diagnose, and recover from errors': error messages should be in plain language (no error codes), precisely indicate the problem, and constructively suggest a solution — 'Error 0x80070057' fails all three. It is not a shortcut problem (heuristic 7), not a memory-load problem (heuristic 6), and not a missing 'emergency exit' (heuristic 3). (Source: NN/g, '10 Usability Heuristics,' Heuristic 9)
A user accidentally taps 'Archive all emails.' There is no undo and no way to cancel while it runs — they can only watch it finish and then restore the emails one by one. Which heuristic does this most directly violate?
答案：User control and freedom
This violates heuristic 3, 'User control and freedom': 'users often perform actions by mistake and need a clearly marked "emergency exit" to leave the unwanted state without going through an extended process' — the classic implementations are exactly undo/redo, cancel buttons, and dismissible dialogs. It corresponds to Norman's reversible actions / forgiveness: users dare to explore when they know mistakes are easy to back out of. (Sources: NN/g, '10 Usability Heuristics,' Heuristic 3; Norman, The Design of Everyday Things)
Within one product, the same 'save and continue' action is labeled 'Save' on page A, 'Submit' on page B, and 'Confirm' on page C — users start wondering whether these are even the same thing. Which heuristic does this most directly violate?
答案：Consistency and standards
This violates heuristic 4, 'Consistency and standards': 'users should not have to wonder whether different words, situations, or actions mean the same thing.' What breaks here is internal consistency — within one product, the same meaning should get the same wording/control; the other layer is external consistency, following platform and industry conventions (back arrows, hamburger menus). Design systems (such as Ant Design or TDesign) are consistency engineered at scale: unified components, spacing, and copy rules. (Source: NN/g, '10 Usability Heuristics,' Heuristic 4)
True or False: Making users memorize a system-generated 8-digit pickup code in one step and type it from memory in the next step complies with 'Recognition rather than recall' — after all, it also trains the user's memory.
答案：False
False. Heuristic 6, 'Recognition rather than recall,' demands exactly the opposite: 'minimize the user's memory load; users should not have to remember information from one part of the interface to use in another' — because recognition is far easier than recall. The right approach is to carry the key information from the previous step into the next, or let the interface remember for the user via autocomplete and recently-used lists, corresponding to Norman's 'knowledge in the world.' 'Training the user's memory' was never an interface's job. (Sources: NN/g, '10 Usability Heuristics,' Heuristic 6; Norman, The Design of Everyday Things)
According to NN/g's standard process, how should a heuristic evaluation be organized?
答案：Recruit 3-5 evaluators and train them on the 10 heuristics; each spends 1-2 hours evaluating independently (get familiar with the product first, then a second pass hunting for violations); finally pool the findings, cluster similar issues with an affinity diagram, and prioritize
NN/g's standard process has three steps: (1) Prepare — recruit 3-5 evaluators, train them on the 10 heuristics, scope narrowly; (2) Evaluate independently — each spends 1-2 hours alone, first getting familiar with the product, then a second pass specifically hunting for violations of the 10 heuristics; (3) Merge — pool the findings, cluster similar issues with an affinity diagram, and prioritize. Group walkthroughs make findings converge, a single expert misses problems, and it is not a large-sample user-voting method. (Source: NN/g, 'How to Conduct a Heuristic Evaluation')
True or False: As long as the evaluator is senior enough, one person doing a heuristic evaluation alone can find all the usability problems in a product — no need to involve anyone else.
答案：False
False. The very reason NN/g requires 3-5 evaluators to evaluate independently and then merge is that any single evaluator — however senior — will miss some of the problems when working alone; only multiple evaluators working in parallel and then pooling their findings gives fuller coverage. 'Just have one guru review it' is one of the most common execution mistakes in heuristic evaluation. (Sources: NN/g, 'How to Conduct a Heuristic Evaluation'; Nielsen J. (1994). Heuristic Evaluation. In Usability Inspection Methods)
True or False: Nielsen's 10 heuristics cross-validate with the fundamental principles in Don Norman's The Design of Everyday Things — for example, 'Visibility of system status' corresponds to Norman's feedback and visibility, 'Error prevention' to constraints and forcing functions (poka-yoke), and 'Recognition rather than recall' to 'knowledge in the world.'
答案：True
True. Almost every one of the 10 maps onto a fundamental principle of Norman's: visibility of system status ↔ feedback and visibility; match between the system and the real world ↔ natural mapping; user control and freedom ↔ reversible actions / forgiveness; error prevention ↔ constraints and forcing functions (poka-yoke); recognition rather than recall ↔ 'knowledge in the world.' Two frameworks confirming each other shows they rest on the same underlying cognitive regularities. (Sources: NN/g, '10 Usability Heuristics'; Norman D. (2013). The Design of Everyday Things, Revised Edition)