My question is tangentially related to this one, but I'd like to take it a step further. Let's assume I'm an outside consultant and I'm working with a product or system that I have very little knowledge about. How much information do you need about a site's users, goals, domain, and features, in order to reliably evaluate it?

I ask because I've sometimes had people ask me to look at a friend's website or something on the fly and, though I'm doing my due diligence, I often give them the thumbs up only to notice a deeper issue a month later. As a UX professional, I find it's already hard to get people to take my advice seriously (as in, actually do what I recommend), and getting egg on my face like that just makes it harder.

A great answer will give me ammunition to give to a client if they ask me to do a quick HE. That is, I should be able to say "I'd love to do a heuristic evaluation for you, and in order to do that properly I should first do x, y, and z..."

2 Answers 2


The goal (of heuristic evaluation) is the identification of any usability issues so that they can be addressed as part of an iterative design process.

I would say you do not need to know too much about the site to perform heuristic evaluation. As long as you know what services/information the site (is intended) to provide you should be good enough.

You the create a list of tasks the evaluator should be performing which are related to what the site is meant to do and use heuristics to measure the success/failure of the evaluator to perform it.

If you want to use NN/g 10 usability heuristic model then this link provides a quite detailed view of how to do it: Heuristic evaluation, a step by step guide.

  • Plan your evaluation - prepare tasks, goals of system, etc.
  • Choose evaluators - experienced and novice
  • Review Heuristics - the nn/g heuristics
  • Conduct evaluation - individual or group
  • Analyze results

And remember There’s no such thing as a “user error”!

  • Isn't it sometimes useful to put yourself into the head of a persona or two?
    – Zelbinian
    Apr 12, 2013 at 20:24
  • Personas are more of a designing phase utility. Helping you communicate ideas better with your team and stay on the same page. uie.com/brainsparks/2006/12/26/when-should-you-use-personas
    – rk.
    Apr 12, 2013 at 20:31
  • 1
    I do realize that, but say I'm evaluating a site with several distinct audiences, aren't I doing my client a disservice if I don't think about how the site effects those different audiences? Like, stack exchange has several user groups: askers, answerers, and administrators. I don't know what the admins on this site actually have to do, so how would I competently do a HE from their perspective without user research data?
    – Zelbinian
    Apr 12, 2013 at 20:37
  • Exactly, you are supposed to think of the audience in the design and in the evaluation, assign the tasks which this audience is supposed to be doing. Your design should accomodate your audiences and your evaluation is a check if that design worked. To accomodate different users, you design their respective tasks and present them to the evaluators.
    – rk.
    Apr 12, 2013 at 22:50

Most of Nielsen's list of Heuristics can be used without knowing much about audience or typical usage. But #2 says "The system should speak the users' language." So we need to know that language.

The minimum amount of information I'd ask for is a definition of the audiences (yes, the various multiple audiences) that includes lingo and language common within their organization.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.