As a designer, you want to make a work of art. I know I do. Our personal goal is to make the best looking site we’ve ever made, with that one (or more) nifty feature that nobody else has. If you had your way, the only usability test performed would be your own usability.

Sadly, works of art are normally less usable than their bland, “don’t deviate from the template” counterparts that all designers have in their repertoire that they loathe using. In the interest of customers and clients, if you aren’t using a template, your art should be tested to see where its shortcomings fall. Actually, even if you’re using a template you should test its usability.

As you may have already guessed, usability testing is a series of tests to see how regular, typical users who may be using your site. It’s a way to see the shortcomings of your design, and to spot any confusion users might have.

No, you cannot perform usability tests yourself, and neither can any member of your team that’s working on it. You are testing the first experience and anyone who has used it before by definition isn’t a candidate. Candidate selection should be randomly selected, which is sometimes called “Hallway Testing” because you’d pick 5 random people who pass you in a hallway.

Usability testing is important because it shows you the typical behavior of a person visiting a site for the first time. Confused people who don’t see a way to accomplish their goal won’t be purchasing anything from your site, or finding the information they wanted. Unhappy, frustrated users will not return unless your website has no competition; a rare feat.

You should begin testing before any design is made. Run a handful of tests, 4 or 5 random people, on the current site to find its strengths and weaknesses. 4 or 5 more people should also be called on to examine your site’s biggest competitors to see their strengths and weaknesses too. Only then should design begin with this information in your right hand and a pen in your left. Wait – probably vice versa.

Testing continues after you’ve implemented your design. See where you’ve gone right and, more importantly, where you’ve gone wrong. Head back to the drawing board and fix your errors and highlight your strengths even further, implement that and retest. Continue until you, or your client, is 100% happy with the test results.

Testing for usability isn’t as complicated as you would think it is. All you need is a computer with access to your test site located in a quiet, distraction free environment, a volunteer, a pad of paper and a pen. For more in depth analysis you can add cameras on the user’s face and hands, and a screen capturing software but these are not necessary for all but the most extensive testing.

You’ll need a goal for your volunteer. A single instruction with no clues how to accomplish this goal. Let them fire up the site and do not interact with them in anyway. Your job, or the job of your tester, is to observe and record. Observe where the user gets frustrated or lost; observe where they spend more time reading or when a user hits the back button.

Ask the volunteer to read everything out loud, and ask them to voice whatever he or she is thinking. Even if they don’t, do not interfere. Any interaction with the volunteer can ruin the test, because if you answer a question about the site, or give a hint – they may avoid a problem area.

When using a more sophisticated lab, you’re also going over the footage of their face, looking for expressions of frustration, surprise… or anything other than satisfaction. You pair the camera on their face with the screenshot so you know exactly what the volunteer was doing when his expression occurred.

Finally, the last stage of the test is the exit interview, where you ask the user three questions. One: How satisfied they were with the sites design. Two: Would you like another task? And Three: Do you have anything you’d like to add. The odd one out is question number two; it is there only to see if they dread returning to your site.

In total, you should run 4 or 5 tests before looking at your data. A single person’s reaction may not indicate a problem, but more than one negative reaction requires attention.

Obviously the first thing you do is review the information you gathered. Any identical issues should be resolved without hesitation, by revisiting the sites design. After attending to the design, retest! You may have screwed something up, or misunderstood the issue.

Redesign, retest and repeat until you are satisfied with the results of your usability test. Only then should your design go live.