Interview - Jakob Nielsen, Ph.D.

SP: In your book "Homepage Usability" you berate Websites for having a link (image or text) to the homepage, on the homepage. But now this is fast becoming the norm, with no public outcry from accessibility groups. Do you feel that usability guidelines must evolve quickly as Web users' experience and expectations surge ahead? How do you go about creating new recommendations for usability?

In general, usability calls for complying with users' expectations so that things simply work without the need for people to figure out what to do every time they visit a new site.

Still, having the homepage link to itself is always going to be bad because there is no conceivable reason users would want to click on such a link. It wouldn't do anything except waste their time and display the same page once again. And a homepage link on the homepage can easily do harm because it increases navigational confusion in two ways. First, the link will reduce the user's feeling of certainty that the page is indeed the homepage (since there is a link to somewhere that's called "homepage"). Second, if following the link, the user may or may not recognize that the destination is the same page as the departure page, so the user may think there were two different pages. Luckily, the harm is not great, but it's there, and since there is no benefit, this is a design feature we should do without. Much better to use the space to display an indication that the current page is the homepage and not have this be an active link.

New usability guidelines are always derived the same way: we conduct user testing on a range of Websites and if we see the same issue cause difficulties in several different designs then we know that it's a general problem that's likely to occur in new designs and hurt even more users unless we warn against it. Similarly, if something works well in many different cases, then it's likely to be beneficial in even more cases.

The trick lies in abstracting away from the specific instances of observed behavior to discover the underlying principle. That's what separates the beginners from the experts in usability: anybody can test a Website and discover fifty places users stumble. It's so obvious when people are stuck or click the wrong link. But only the most senior usability professionals can take a range of findings and generalize them into guidelines that will apply to new cases.

SP: There's a common perception that usability is largely about making sites usable by the "lowest common denominator" (and thereby meeting the most basic needs of everyone else). What are your thoughts on this?

False. Usability consists of five parameters: learnability, efficiency, memorability, error avoidance, and subjective satisfaction. The balance between the five changes for different kinds of projects. The initial phases of learnability have dominated Web usability because people leave Websites that they can't figure out on the first visit. But if you can make a site sufficiently approachable to get a substantial number of repeat visitors, then memorability would be important as well, and if people used the site frequently, then efficiency for the expert user would become important.

Good usability includes ideas like progressive disclosure where you show a small number of features to the less experienced user to lower the hurdle of getting started and yet have a larger number of features available for the expert to call up.

SP: How important is a consideration of a site's target audience in the identification of specific usability problems or issues? Are there general blanket guidelines that can be thrown over any Website in terms of usability? Or should each site be regarded as unique, as having its own particular audience and usability issues, and recommendations made in accordance with its individual characteristics?

In principle, usability is always relative to the users and their tasks. When you run user tests, you should take care to recruit representative users as test participants and to give them realistic tasks to perform. Many detailed usability problems would certainly depend on the specifics of the domain: the best user interface for a site selling 10 products would not be the best way of dealing with 1,000 products and one would need a third design for a million products. Similarly, the vocabulary used to describe pharmaceuticals would be different, depending on whether the users were physicians or patients.

But most of the principles of interaction design are fairly general and relate to the problem of bridging the gap between humans and machines. Principles like providing feedback, making the main options visible, and minimizing memory load have been the same for fifteen years, and apply to all interactive systems, whether mainframes, PCs, or Websites. Same for the top three guidelines for error messages: human-readable, specific, and constructive, which haven't changed since Ben Shneiderman described them twenty years ago. Even much more narrow guidelines, like the need to use relative font sizes, will hold across a very wide range of designs, unless you are *sure* that all of your users are going to be eagle-eyed.

SP: You sell an excellent PDF with 200+ tips for usability studies, and post Alertbox columns about usability testing, but it's difficult to find a good "step-by-step" tutorial on how to actually conduct usability studies. Can you point out any good resources on this topic?

Good question: this happens to be the topic of my next book!

...which we look forward to reading, Jakob! On behalf of the SitePoint community, thanks for taking the time to chat with us! If you'd like to hear more from Mr. Nielsen, check out our review of LIFT - Nielsen Norman Group edition, which contains more of this interview.