Heuristic Evaluation: How To Run One Using Nielsen’s 10 Usability Heuristics
Lily Roberts
Senior Marketing Strategist
10.10.2026
Short answer: A heuristic evaluation is an expert review in which a small group of evaluators, usually three to five, inspect an interface on their own and log every place it breaks an accepted set of usability principles, most often Jakob Nielsen’s 10 usability heuristics. Each problem gets a severity rating from 0 to 4, the evaluators merge their findings, and the team ships a prioritized fix list. It is fast and cheap, but it finds likely problems, not proven ones, so it works best alongside usability testing.
A heuristic evaluation is one of the quickest ways to find usability problems in a product, website or prototype without recruiting a single user. You give a few trained reviewers a checklist of proven design principles, they walk through the interface, and they write down every place the design breaks one. Done well, it takes days rather than weeks and costs a fraction of a research study.
This guide covers where the method came from, all 10 of Nielsen’s heuristics with a plain example of a violation for each, a step-by-step process, how many evaluators you actually need, a copyable template, how the method compares with usability testing and a full UX audit, and where it falls short.
A heuristic evaluation is a usability inspection method. Instead of watching real users struggle with a product, a small group of evaluators compares the interface against a list of broad rules of thumb, called heuristics, and records each violation they find. The output is a list of usability problems, each tied to the heuristic it breaks and rated for severity.
The method was introduced by Jakob Nielsen and Rolf Molich in 1990. Nielsen then refined the list in 1994 using a factor analysis of 249 usability problems, which produced the ten heuristics most teams still use. Nielsen Norman Group (NN/g), the consultancy Nielsen co-founded, refreshed the names, descriptions and examples in 2020, but it states that the ten principles themselves have stayed the same since 1994. You can read the current wording in NN/g’s 10 Usability Heuristics for User Interface Design.
You will also see the method called a heuristic analysis or a heuristic review. In agency proposals it often appears as one part of a broader UX audit, alongside analytics review, accessibility checks and competitor benchmarking.
Why Teams Still Use It
Speed: a focused evaluation of a few key flows can be finished in a working week.
Cost: no participant recruitment, incentives or lab setup.
Timing: it works on wireframes, Figma prototypes and live products, so you can run it before any code exists.
Shared language: tying each issue to a named heuristic makes findings easier to explain to product managers and engineers.
Nielsen’s 10 Usability Heuristics, With Examples
The Nielsen heuristics are deliberately broad. They are not a design spec, and they do not tell you what a button should look like. They describe qualities a usable interface tends to have, which is why they still apply to mobile apps, SaaS dashboards and AI chat interfaces decades after they were written. The table below lists all 10 usability heuristics, what each one means in plain language and a typical violation.
#
Heuristic
What It Means
Example Of A Violation
1
Visibility of system status
Keep users informed about what is happening, with timely feedback.
A file upload shows no progress bar, so users click Upload again and create duplicates.
2
Match between the system and the real world
Use words, concepts and order that users already know.
A checkout error says “Error 4012: tokenization failed” instead of “Your card was declined.”
3
User control and freedom
Give users a clear way out of mistakes, such as undo and cancel.
A multi-step setup wizard has no Back button, so fixing step 2 means starting over.
4
Consistency and standards
Follow platform conventions and stay consistent inside the product.
“Save” is a blue button on one screen and a grey text link labeled “Apply” on the next.
5
Error prevention
Design so that errors are less likely to happen in the first place.
A date field accepts any text and only rejects “31/02” after the form is submitted.
6
Recognition rather than recall
Make options visible so users do not have to remember them.
A search page forgets the filters you applied as soon as you open a result and come back.
7
Flexibility and efficiency of use
Offer shortcuts for experienced users without confusing new ones.
A tool used daily by admins has no bulk edit, so updating 50 records takes 50 separate edits.
8
Aesthetic and minimalist design
Show only what is relevant; extra content competes with what matters.
A pricing page buries the plan comparison under three promo banners and a video.
9
Help users recognize, diagnose and recover from errors
Error messages should explain the problem in plain words and suggest a fix.
A password form says “Invalid input” without saying which rule the password breaks.
10
Help and documentation
Provide help that is easy to search and focused on the task at hand.
The only help link opens a 60-page PDF manual with no search.
Some teams add their own domain heuristics on top of Nielsen’s ten, for example accessibility, trust signals for ecommerce, or explainability for AI features. That is fine, as long as every evaluator works from the same written list and the same definitions.
How To Run A Heuristic Evaluation, Step By Step
NN/g describes the process in three broad phases: preparation, independent evaluation and consolidation. The six steps below follow that structure and add the reporting step that most teams need to turn findings into work. NN/g’s own walkthrough, How to Conduct a Heuristic Evaluation, is worth reading alongside this.
1. Define The Scope
Decide exactly what is being evaluated before anyone opens the product. A vague scope like “the whole app” produces a long, shallow list. A good scope names the user type, the platform and the specific flows, for example “new trial users on desktop: sign up, connect a data source, create the first report.”
List the 3 to 6 tasks or flows that matter most to the business.
Write a one-paragraph persona so evaluators know who they are judging for.
Agree on the heuristic set (Nielsen’s ten, plus any domain additions).
Confirm the build or prototype version so everyone reviews the same thing.
2. Choose And Brief Your Evaluators
Pick evaluators who understand usability principles. UX designers and researchers are the obvious choice, but a product manager or support lead who has been trained on the heuristics can add useful perspective. Evaluators who know both usability and the product’s domain tend to find more problems. Brief everyone with the scope, the persona, the heuristic definitions, the severity scale and the logging template so their notes are comparable later.
3. Review The Interface Independently
This is the core of the method. Each evaluator works alone and does not see anyone else’s notes until their own pass is finished. NN/g suggests roughly one to two hours per evaluator per session. A common approach is two passes: a first pass to get a feel for the flow, then a second, slower pass to inspect each screen and interaction against the heuristics.
For every issue, the evaluator logs where it happens, what happens, which heuristic it breaks and, ideally, a screenshot. One issue can break more than one heuristic; record the main one and note the others.
4. Rate Severity With Nielsen’s 0 To 4 Scale
Severity tells the team what to fix first. Nielsen’s scale, described in NN/g’s article on rating the severity of usability problems, runs from 0 to 4. Severity combines how often the problem occurs (frequency), how hard it is to overcome (impact), whether users keep hitting it (persistence) and its effect on the product’s market success (market impact).
Rating
Meaning
What To Do
0
Not a usability problem
Drop it from the list.
1
Cosmetic problem only
Fix only if extra time is available.
2
Minor usability problem
Low priority; batch with related work.
3
Major usability problem
High priority; schedule a fix soon.
4
Usability catastrophe
Must be fixed before release.
Nielsen recommends that each evaluator rates severity on their own, and notes that ratings from a single evaluator are too unreliable to trust. Averaging the ratings of three evaluators is good enough for most practical purposes. A practical way to do this is to rate after consolidation: share the merged issue list and have each evaluator score every item privately before you compare.
5. Consolidate The Findings
Bring the evaluators together and merge their lists. Expect heavy overlap phrased in different ways, plus a long tail of issues only one person spotted. Those single-evaluator finds are the reason you use more than one reviewer, so do not discard them just because nobody else saw them.
Combine duplicates into one issue and keep the clearest description.
Group issues by flow or screen so related fixes ship together.
Resolve disagreements about whether something is a real problem by going back to the heuristic definition.
Average the independent severity scores and sort the list from highest to lowest.
6. Write The Report
The report should make it easy for a product team to act. Lead with a short summary of the biggest risks, then list each issue with its location, description, heuristic, severity, evidence and a recommended fix. Keep recommendations specific (“show inline validation on the date field as the user types”) rather than generic (“improve error handling”). Positive findings are worth including too, so the team knows what not to break.
How Many Evaluators You Need
NN/g’s current guidance is that three to five people should evaluate the same interface independently, because any one evaluator, however experienced, will miss problems. The figure comes from Nielsen’s earlier research: in “The Theory Behind Heuristic Evaluations,” he reported that single evaluators found only about 35 percent of the usability problems, averaged across six projects, and recommended “about five evaluators, but certainly at least three.”
Adding evaluators beyond five tends to bring diminishing returns, because most new reviewers rediscover problems that are already on the list. The exact number depends on what is at stake:
Small feature or early prototype: three evaluators is usually enough.
Core product flow such as onboarding or checkout: aim for five.
High-risk or regulated product: consider more evaluators, and pair the review with usability testing.
A solo review by one designer is still better than nothing, but call it what it is: an expert review, not a full heuristic evaluation, and expect it to miss a large share of the issues.
Heuristic Evaluation Template And UX Audit Checklist
Copy the table below into a spreadsheet or a Figma or FigJam board and give one copy to each evaluator. The columns are the same ones most agencies use in a UX audit template, so findings from different reviewers merge cleanly.
ID
Flow / Screen
Issue Description
Heuristic Violated
Severity (0 to 4)
Evidence
Recommendation
H-01
Sign up, step 2
No indication of how many steps remain
1. Visibility of system status
2
Screenshot 04
Add a step indicator (“Step 2 of 4”)
H-02
Settings, billing
Delete card has no confirmation or undo
5. Error prevention; 3. User control
3
Screenshot 11
Add a confirmation dialog and a short undo window
H-03
Alongside the log, give evaluators a short UX audit checklist of questions to ask on every screen. It keeps the second pass systematic.
Do users always know where they are, what just happened and what will happen next?
Are labels written in the user’s language rather than internal or technical terms?
Can users go back, cancel or undo without losing work?
Do the same actions look and behave the same way everywhere?
Does the design stop likely mistakes before they happen, such as with constraints, defaults and confirmations?
Are options, filters and previous choices visible instead of remembered?
Are there shortcuts or bulk actions for frequent tasks?
Is every element on the screen earning its place?
Do error messages say what went wrong and how to fix it?
Is help available in context, at the moment it is needed?
Heuristic Evaluation vs Usability Testing vs A Full UX Audit
These three terms get used loosely, and buyers often receive proposals that mix them up. The distinction is simple: a heuristic evaluation uses experts and principles, usability testing uses real users doing real tasks, and a UX audit is a broader package that usually contains a heuristic evaluation plus several other inputs.
Factor
Heuristic Evaluation
Usability Testing
Full UX Audit
Who finds the issues
3 to 5 expert evaluators
Real or representative users, observed by a researcher
A UX team combining several methods
Main input
Usability principles
User behavior on set tasks
Heuristics, analytics, accessibility checks, sometimes user testing
What it tells you
Likely problems and why they break principles
Where real users actually struggle
A prioritized picture of UX health across the product
Typical timeline
Days
1 to 3 weeks including recruitment
2 to 6 weeks depending on scope
Relative cost
Low
Medium
Medium to high
Best stage
Prototype or before a test round
Prototype through live product
Live product with traffic and data
Main weakness
Can flag false problems and miss real ones
Small samples, setup effort
Higher cost; quality varies a lot by provider
The methods work best in sequence. Run a heuristic evaluation first to clear out the obvious problems cheaply, then spend your usability testing budget on the questions only users can answer. For a wider view of where each method fits, see our guide to UX research methods.
Limitations And Common Mistakes
A heuristic evaluation is a strong first filter, not a final verdict. Knowing its limits helps you plan around them.
What The Method Cannot Do
It predicts, it does not prove. Evaluators flag what they believe will cause trouble. Some flagged issues never bother real users, and some real problems never show up in the review.
It misses context. Experts are not your users. They may not know the domain vocabulary, the job users are trying to get done, or the workarounds people rely on.
It does not measure. You get a list of issues, not task success rates, time on task or conversion impact.
It is not an accessibility audit. Nielsen’s heuristics overlap with accessibility but do not replace a review against WCAG.
It depends on the evaluators. Results vary with skill and domain knowledge, which is why independence and multiple reviewers matter.
Mistakes That Weaken The Results
Reviewing as a group from the start, so the loudest voice shapes everyone’s findings.
Skipping severity ratings, which leaves the team with an unsorted list it cannot act on.
Logging personal taste (“I don’t like this color”) as a heuristic violation.
Scoping too broadly, which produces dozens of shallow notes instead of deep findings on the flows that matter.
Writing vague recommendations that engineers cannot estimate or build.
When To Bring In A UX Audit Company
Many teams can run a useful heuristic evaluation in house, especially if they have two or three designers and a researcher. Outside help makes sense in a few specific situations:
You do not have enough independent evaluators. A single in-house designer reviewing their own work is the weakest version of the method.
The team is too close to the product. People who built a flow struggle to see it fresh, and internal politics can soften severity ratings.
You need more than heuristics. If the question is “why are trials not converting,” you probably need analytics, session recordings, accessibility review and user testing combined into one audit.
The stakes are high. Before a redesign, a funding round, a migration or an enterprise sales push, an outside audit gives a defensible baseline.
If you are comparing providers, our list of the top UX audit companies covers firms that run heuristic reviews as part of a broader audit. If you mainly need evidence from real users, the top UX research and usability testing agencies list is the better starting point. And if the audit is likely to turn into a redesign, it can help to talk early with one of the top product design agencies as well.
When you brief a provider, ask them to show a sample report, to name how many evaluators will review independently, which heuristic set they use, how they rate severity and whether recommendations come with enough detail to estimate. A good partner will be happy to answer all five.
Frequently Asked Questions
What are Nielsen’s heuristics?
Nielsen’s heuristics are ten broad usability principles published by Jakob Nielsen in 1994, building on work he did with Rolf Molich in 1990. They cover system status, real-world language, user control, consistency, error prevention, recognition over recall, efficiency, minimalist design, error recovery and help. NN/g updated their wording and examples in 2020 without changing the principles.
How long does a heuristic evaluation take?
The independent review itself often takes each evaluator one to two hours per session, following NN/g’s guidance. Add time for scoping, consolidation, severity rating and writing the report, and a focused evaluation of a few flows usually fits within a working week. Larger products with many flows take longer.
Is a heuristic evaluation the same as a UX audit?
No. A heuristic evaluation is one method, while a UX audit is a broader review that usually includes a heuristic evaluation plus analytics, accessibility checks, content review and sometimes usability testing. If a proposal calls itself a UX audit but only contains heuristics, ask what else is included.
What should a UX audit checklist include?
A practical UX audit checklist turns each of the 10 usability heuristics into a question you ask on every screen, such as “Do users always know what just happened?” A full audit checklist also adds accessibility checks, analytics findings, page speed and content clarity. The checklist in this guide is a good starting point for the heuristic part.
Is there a free heuristic evaluation template?
Yes. The template table in this guide covers the columns you need: ID, flow or screen, issue description, heuristic violated, severity, evidence and recommendation. Copy it into a spreadsheet and give each evaluator their own copy so their findings stay independent until consolidation. The same structure works as a basic UX audit template.
Can one person do a heuristic analysis?
One person can do a heuristic analysis, but the results will be incomplete. Nielsen found that single evaluators caught only about 35 percent of usability problems on average, which is why NN/g recommends three to five independent evaluators. A solo review is useful as a quick check, not as a full evaluation.
Should you do a heuristic evaluation before usability testing?
Usually, yes. Running a heuristic evaluation first removes obvious problems cheaply, so your usability test sessions can focus on deeper questions that only real users can answer. It also stops participants from getting stuck on basic issues that experts could have caught in a day.
The Short Version
A heuristic evaluation is a fast, low-cost way to find usability problems by having three to five evaluators independently check an interface against Nielsen’s 10 usability heuristics. Scope it tightly, keep the reviews independent, rate every issue on the 0 to 4 severity scale, merge the findings and deliver a report with specific fixes.
Treat the results as strong hypotheses rather than proof. Pair the evaluation with usability testing for the questions only users can answer, and bring in a UX audit company when you lack independent evaluators, need a broader review or face a high-stakes release.
