UX Audits

Heuristic Evaluation: A Practical Guide with Examples

By the UXreviewer.io team4 min read

A heuristic evaluation is a usability inspection in which evaluators check an interface against a set of established principles, most often Jakob Nielsen's ten usability heuristics, and log every violation with a severity rating. It is fast, needs no participants and reliably uncovers common problems such as missing feedback, confusing language and inconsistent controls.

This guide explains the ten heuristics with concrete examples, shows how to run an evaluation step by step, and covers how to adapt the method for chatbots and AI agents, where some classic principles need a fresh interpretation.

What are the ten usability heuristics, with examples?

The heuristics are broad rules of thumb, not strict checklists. Each one below comes with a typical violation evaluators find in real products.

  • Visibility of system status: an upload runs with no progress indicator, so users click again and create duplicates
  • Match between system and the real world: a banking app labels transfers as "outbound transactions" instead of "send money"
  • User control and freedom: a multi-step form offers no way back without losing entered data
  • Consistency and standards: "Save" is a blue button on one screen and a text link on the next
  • Error prevention: a date field accepts any text and only complains after submission
  • Recognition rather than recall: users must remember an order number from a previous screen to continue
  • Flexibility and efficiency of use: no keyboard shortcuts or saved filters for daily power users
  • Aesthetic and minimalist design: promotional banners push the main task below the fold
  • Help users recognize, diagnose and recover from errors: an error reads "Something went wrong" with no next step
  • Help and documentation: help exists but is not reachable from the screen where people get stuck

How do you run a heuristic evaluation step by step?

The method is simple, but discipline makes the difference between a useful evaluation and a list of personal preferences.

  1. Define scope and users. Choose the flows to evaluate and describe the target users and their goals in a sentence or two.
  2. Pick the heuristic set. The classic ten work for most products; add domain rules for accessibility, mobile or AI where relevant.
  3. Evaluate independently. Each evaluator walks the flows alone, usually twice: once to get a feel, once to log issues.
  4. Document each finding. Note the screen, the heuristic violated, what happens, why it matters and a screenshot.
  5. Merge and de-duplicate. Combine findings from all evaluators into one list.
  6. Rate severity together. Agree on a rating for each issue using a shared scale.
  7. Write recommendations. Turn each issue into a specific, buildable fix.

How should you rate severity?

A widely used scale runs from zero to four: not a problem, cosmetic, minor, major and catastrophic. Severity combines three factors: how often the problem occurs, how hard it is for users to overcome, and whether it keeps bothering them once they know about it.

In practice, add business impact as a tie-breaker. A minor issue on the checkout page can deserve more attention than a major one in a rarely used settings screen. Write down why you chose each rating, so the development team can challenge or accept it quickly.

Want an expert opinion on your product?

Talk to a UX expert or AI Code Engineer — 30 minutes, free, no obligation.

Book a call for free

How many evaluators do you need?

A single evaluator misses a meaningful share of problems, because everyone has blind spots. The traditional recommendation is to use several evaluators, commonly three to five, and merge their findings. Beyond that, returns diminish for most products.

Diversity matters more than headcount. An evaluator with interaction design experience, one with psychology or research background and one who knows the domain will find more distinct issues than three people with the same profile. That mix of perspectives is one reason we combine UX experts, psychology experts and AI Code Engineers in our reviews.

How do heuristics apply to chatbots and AI agents?

Classic heuristics still hold for conversational and agentic interfaces, but they show up differently. Visibility of system status now means telling users what the agent is doing and what it has already done. User control means being able to stop, undo or review an action before it is executed. Error recovery means the assistant admits uncertainty and offers a way forward instead of confidently guessing.

Evaluators should add a few AI-specific checks: is it clear when users talk to an AI, are limits explained, can users reach a human, and do suggestions match the user's actual context. For a full checklist, see our agentic UX audit checklist and the guide to conversational UX reviews.

What are the limits of heuristic evaluation?

Heuristic evaluation finds violations of general principles, but it cannot tell you whether your audience understands your domain terms or values your features. It can also produce false positives: an evaluator may flag something that real users handle fine.

Use it as a fast, cost-efficient first pass and pair it with real user input where decisions are risky. Our comparison of UX review vs. usability testing shows how to sequence the two.

Frequently asked questions

Are Nielsen's heuristics still relevant in 2026?

Yes. They are deliberately general, which is why they still apply to mobile apps, chatbots and AI agents. Most teams extend them with domain-specific checks rather than replacing them.

How long does a heuristic evaluation take?

For a handful of core flows, evaluators often need a day or two each, plus time to merge findings and write recommendations. Large products with many roles take longer.

What is the difference between a heuristic evaluation and a cognitive walkthrough?

A heuristic evaluation checks screens against general principles, wherever they appear. A cognitive walkthrough follows one task step by step and asks at each step whether a new user would know what to do next. Many audits combine both methods.

Can I run a heuristic evaluation myself?

You can, and it is a good habit. Internal teams tend to miss issues they are used to, though, so an outside review adds value before major releases. You can book a call for free to discuss an external review.

Book a call for free

From UX audits and Agentic UX reviews to AI development, agentic coding workshops and publishing your AI-built app on the App Store and Google Play — we help you ship. 24/7 call support in English and German.

Related insights