Skip to Content

That Mess on Your Web Site

Fixing a few common design mistakes would make the Web a far more pleasant and useful place to hang out, says a guru of interactive interfaces.
September 1, 1998

In some ways, the most difficult connection on the Internet may be the very last one: between the computer screen and the human mind. Although finicky servers and overloaded phone lines cause their share of problems, the bits usually get delivered. The message, however, often doesn’t. It remains imprisoned within a confusing, indecipherable Web page. Viewers of the page follow wrong links. They can’t find what they want. They get lost, or they get bored–and they move on.

The hypertext markup language (HTML) used to create Web sites was created less than a decade ago. And as with any language, people learning it must babble for a while before they achieve eloquence. In the meantime, users of the Web struggle to decipher its mangled metaphors and broken syntax.

Jakob Nielsen has made it his business to watch people engaged in that struggle. Nielsen is a distinguished engineer at Sun Microsystems, the computer company known best for its high-powered Internet servers and its creation of the Java programming language. His observations, carried out in Sun’s usability labs in Menlo Park, Calif., are teaching Nielsen and his colleagues the rules of effective communication on the Web (see sidebar: “Jakob Nielsen’s Seven Deadly Sins of Web Design”). Nielsen shares those lessons in a biweekly Web column, “Alertbox,” which is considered a must-read among usability engineers ( His latest book, Designing Excellent Websites: Secrets of an Information Architect, is due out this month.

Nielsen also travels widely, relentlessly advocating the cause of the beleaguered Web user. William Allstetter caught up with Sun’s usability guru recently at the Human Factors & the Web Conference at the AT&T Learning Center in Basking Ridge, N.J.

TR: What are the defining characteristics of an unusable interface?
NIELSEN: It’s difficult to learn. You get low performance even if you are an expert. You hate it. You make lots of errors. And you can’t remember how to use it after you have been on vacation.

TR: And you think a lot of Web sites fit that description?
NIELSEN: Yes. The default on the Web is that people cannot use it. Jared Spool, founder of User Interface Engineering ( and an expert on interactive design, did an interesting study. He showed people a home page and asked them to find a specific piece of information that he knew existed on that Web site. But users succeeded in this task only 42 percent of the time.

TR: Why does the Web present such difficulties?
NIELSEN: I think the reason is that it is designed by people who don’t understand interactive design. Most Web sites have never been tested on real users. Often, the only test is taking somebody into the art director’s office to look at the screen. And that somebody is typically a vice president of marketing, who is thoroughly familiar with the company and so of course can understand the site design perfectly. But this is not a person who represents the general Web population.

TR: What are telltale signs that a site has been designed by somebody who doesn’t understand interaction?
NIELSEN: Big blocks of text or big pictures are two dead giveaways. Another symptom is the absence of links.

TR: Are the wrong people doing Web design?
NIELSEN: That’s part of the problem. Web sites are created by people in a company who are immersed in some project or business activity. They want to use the Web to communicate about what they’re doing, but along the way, they seem to forget that their job is to convey information to people outside of their own circle. The problem is that once a person becomes familiar with a company’s structure and lingo, it’s very difficult to design an interface that will work for someone lacking that familiarity. It’s like those Where’s Waldo books–once you know where Waldo is, you can no longer pretend that you don’t.

TR: How does this kind of inward thinking lead to bad site design?
NIELSEN: The classic mistake is to design a site based on the internal structure of the company, not based on how the user accesses the site. IBM’s site is a good example. Say you own an IBM PC 300PL and you want to buy a compatible printer. If you go to the product page for the computer, you won’t find any links to printers. That’s because printers are made by a different division of IBM, and apparently no one bothered to link the printer pages to the PC pages. So you have to go all the way up to the IBM home page, or use the search button. Clicking on the “Assistant” button brings up a window that is supposed to help you find things. But the menu of products includes only ones from the division of IBM that designed the page you are on, not IBM’s entire product line.

TR: How do you create a user-oriented Web site?
NIELSEN: You start with the users. That may sound like obvious advice, but most people don’t follow it. Instead, they look at the Web as an opportunity to accomplish some goal of their organization. But that will only take you so far. If you are going to create a golf site, for instance, first find out what golfers want.

TR: How do you test usability?
NIELSEN: There are many phases to that. The most basic one is to get customers in our usability lab and have them sit and use the system. Direct observation is the best way of gaining usability knowledge and insights. We pay attention to facial expressions and body language and voice. It’s just so painful to observe people make the same mistake again and again and again. That’s where the observation really hits home.

TR: What commonly surprises Web-site designers when they watch users?
NIELSEN: Designers are taken aback to see users misinterpret interface elements that were supposed to be obvious. Users aren’t stupid, but they are not in your company, and they do not understand your particular way of thinking, your vocabulary. Here at Sun, for example, we have something called Java WorkShop. In the early days, a lot of people would come across a link by that name and think, “Aha, this will be a Java training seminar.” But that’s not true–it’s actually a piece of software. People who were looking for training would click on it, and people looking for programming tools would not click on it. Both types of users were fooled.

TR: What behaviors or characteristics do you see as being prevalent among Web users?
NIELSEN: People using the Web are impatient and very goal-driven; they always complain about anything that they perceive is slow. Also, they don’t have a lot of tolerance for what you might call marketing speak or for companies that try to oversell to them.

TR: How about reading on the Web? Do people read differently on the Web than they do the printed page?
NIELSEN: This is one way the impatience shows itself. People don’t read long blocks of text carefully or thoroughly on the Web. Screen resolution is too low, too coarse, so the letters don’t feel smooth to the eye. That slows down the eye when it tries to read the text. So Web users tend to just scan sites, picking out the little snippets that are of interest to them.

TR: How should the writing and design of a site change to accommodate people who don’t read?
NIELSEN: First, think how you would cover a topic in print–then cut the word count in half. People are not going to read every single word. Acknowledge that. Restructure the site to take advantage of hypertext.

TR: What are some specific principles to follow?
NIELSEN: It’s important to specify what the page is about at the very beginning. People allocate very few seconds to deciding whether a site is of interest. If the good stuff is at the bottom of the page, the user may move on without ever seeing it. For the same reason, it’s good to make heavy use of highlighted keywords–you want something that pops out as people scan down the page, something the eye can rest on. Also, you don’t want to present a topic as one long, scrolling page.

TR: Some online publications break up articles into sequential chunks. At the bottom of each page it says, “Click here for the next page.”
NIELSEN: I call that “page turning,” and it’s a truly bad solution to the problem of scrolling. It completely ignores what makes hypertext so powerful. You still have to go through these four pages in sequence. You can’t jump to the one thing that you care about.

TR: What’s a better way to handle long text, then?
NIELSEN: Make a brief introductory page that tells people very, very clearly what it is about. That page can be linked to a lot of sidebars and background information and other detail. If the links clearly tell which sidebars have what information, people can skip two-thirds of them and choose only the ones that are important to them. And linking to material from outside your own site allows you to take advantage of what others on the Web have produced. If you can select the correct links, you can enhance your service thousands of times beyond what you have the capacity to deliver yourself.

TR: The Web is a pretty unruly place, but you often harp on the need for structure. Why is that so important?
NIELSEN: A lot of Web sites don’t have an information architecture–a structure–at all. They are just collections of pages that people lumped together on a Web site. Structure helps users navigate better by making it apparent at all times where they are on the site, where they have been, and where they can go. Structure also provides a clear underlying concept on which to map the diverse elements of a Web site. For example, if you are on a news-articles site with no structure, you say, “Okay, I am now reading article number 2,125.” But how does that help you get to any other place on the site? How does it help you understand the relationship between this article and the previous one?

TR: In that example, what might you suggest?
NIELSEN: Well, you could add a timeline. Or you could divide the site into sections, like a newspaper–business, sports, whatever. You could provide the ability to sort through articles according to bylines. These kind of structural mechanisms make it much easier to deal with a vast pool of articles. Now there is one more point. You have to give the users what they say they want. But then you have to go beyond that and give them something extra. The Internet has to be better than reality.

TR: What sites meet this test?
NIELSEN: I think that, the online bookseller, is better than reality in three ways. One, it is a bookstore that actually has every book. A physical bookstore can’t do that. Second, for every book listed it also offers the titles of three other recommended books. Now, if you go to a regular bookstore, a knowledgeable staff will know good books to recommend. But they will know that only in a limited area of their expertise. At Amazon, they keep track of what people buy. People who bought the book you are looking at have also bought other books. You get a recommendation for the three they bought most frequently. Finally, sends out an e-mail message when the book has been shipped. It lets you know that in two days you will get the book you ordered. In the physical world, it wouldn’t really make sense to send a postcard that says that your package is now in the mail, because the package and the postcard would arrive at about the same time.

TR: What usability improvements can help companies make money on the Web?
NIELSEN: A classic mistake to avoid is that companies make it difficult for people to buy online. They don’t have the complete list of their products online. Also, they don’t have descriptions of all the product variations. The Web allows you to have infinite detail. You can have all your products and all the obscure variations of those products online. So you can expand your sales dramatically just by making it easy for people to spend their money on your site.

Keep Reading

Most Popular

Large language models can do jaw-dropping things. But nobody knows exactly why.

And that's a problem. Figuring it out is one of the biggest scientific puzzles of our time and a crucial step towards controlling more powerful future models.

OpenAI teases an amazing new generative video model called Sora

The firm is sharing Sora with a small group of safety testers but the rest of us will have to wait to learn more.

The problem with plug-in hybrids? Their drivers.

Plug-in hybrids are often sold as a transition to EVs, but new data from Europe shows we’re still underestimating the emissions they produce.

Google DeepMind’s new generative model makes Super Mario–like games from scratch

Genie learns how to control games by watching hours and hours of video. It could help train next-gen robots too.

Stay connected

Illustration by Rose Wong

Get the latest updates from
MIT Technology Review

Discover special offers, top stories, upcoming events, and more.

Thank you for submitting your email!

Explore more newsletters

It looks like something went wrong.

We’re having trouble saving your preferences. Try refreshing this page and updating them one more time. If you continue to get this message, reach out to us at with a list of newsletters you’d like to receive.