Designing Web Usability: The Practice of Simplicity
by Jakob Nielsen. Indianapolis: New Riders, 2000. 419p. $45 (ISBN 1-562-05810-X).
by Tom Zillner, ITAL Editor
I came to Jakob Nielsen's book, Designing Web Usability, with a certain amount of trepidation. I have been a sometime visitor to Nielsen's Web site on usability, http://useit.com. It is very spare and sparse in terms of visual feel. There are no graphics whatsoever, with all features that would normally require graphical elements provided through variations in font size and use of color. I half-expected that his book would include some sort of diatribe against overuse of graphics. In point of fact, when you do a quick page-through of the book you will find actual Web pages, with plenty of graphics, sprinkled on almost every other page. This provides your first hint of a book prepared with a great deal of care and attention to detail.
Designers on the Web face a virtual tabula rosa, the ultimate in blank slates. There is a lot of flexibility, and with that flexibility comes an almost infinite number of ways to screw up. Nielsen has probably seen and identified them all. He masterfully proceeds through page design, content design, and site design, reinforcing his suggestions for doing better with examples, both good and bad. Particularly interesting are examples where a site has gone through a number of generations of redesign with consequent improvement. Seeing this sort of evolution is much better than viewing dummy mock-ups illustrating the same points.
You may think that I have gotten sidetracked from the get-go by starting out discussing the examples and illustrations that Nielsen provides, but they are extremely important in a book of this sort. The Web is an inherently visual medium, and all of the principles of design that apply in other visual media, and more, must be brought to bear. Therefore, it is essential to provide numerous screen shots to illustrate the good, the bad, and the ugly in the Web world. Nielsen does this admirably.
Nielsen has all of the requisite credentials to discuss usability. For many years he worked for Sun Microsystems, dealing with user interface usability issues on a number of hardware and software platforms. His role was expanded to include Web usability upon the wildfire spread of Web content. He now partners with another usability guru, Donald Norman (author of The Design of Everyday Things, Things That Make Us Smart, and The Invisible Computer), in a private consulting firm addressing usability issues. Nielsen definitely has the chops, and he comes through with an extremely well-written and well-produced volume.
As I mentioned above, the book proceeds from page design to content design to site design. There are other ancillary sections, but the real meat and substance of the book are in these three sections. Because page design is "the most immediately visible part of Web design" Nielsen discusses it first. He writes at great length upon the subject of "screen real estate," in terms of maximizing utilization of the screen without wasting space with gratuitous or unimportant elements, and more importantly in terms of the ability to display information across a wide range of platforms, both in terms of browsers and of screen sizes and resolutions.
One of the big problems in dealing with a wide variety of possible Web configurations is that many designers simply don't deal with the problem at all. This sometimes results in pages that, for example, display well on Macintoshes (a frequent choice of Web designers) but look bizarre on Wintel computers. These problems are, in many or most cases, browser rather than computer problems. To partly overcome this problem Nielsen recommends testing on a wide variety of computers with multiple versions of browsers. He also believes that the pace of upgrades of browsers is slowing, so it is important to support back-level browsers for some time. Further more, because of this kind of user inertia it is also important to not introduce "bleeding-edge" technology to Web pages until a sufficient number of people are likely to have the browser-end plug-in on their machine. His recommended waiting period before introducing such features is one year.
A particularly nice illustration of the differences inherent in Web viewing is the rendering of a set of graphics using Netscape 3.01, Netscape 4.01, and Internet Explorer 3.01. As you might expect, there are subtle (and not-so-subtle) variations among the browsers, and this exacerbates the problems a designer faces in making pages usable.
After discussing problems with variations among computers, Nielsen very logically proceeds to discuss the move to separate content from rendering or presentation. If there are so many variations among the devices, browsers, and resolutions available to people, it makes sense to accommodate these variations as much as possible without changing the underlying content; thus the separation. Nielsen follows up this observation with a subsection discussing style sheets, although he notes that at the present time, the major incentive to use style sheets is not the ability to display the same content across a wide range of platforms. Because information about platform is not, by and large, provided to the server, it is impossible to render content to reflect the browser and hardware used by the viewer. This will come in the future, but for now the main motivation for style sheets is that they can force uniform and consistent design across a site. This is very important, because consistency aids the user in traversing and exploiting a site's content. As usual, Nielsen passes on a great deal of information about style sheets and how to best use them.
Similarly, he is very clear and detailed in discussing the use of links within pages, passing from the broadest picture of link placement and description right down to abstruse but useful technical details, for example, that links to content that can be identified as, for example, http:// foobar.com/info, should always carry a trailing slash: http://foobar.com/ info/. This facilitates slightly more rapid retrieval. It is this exhaustive coverage, with the most important broad generalities first followed by detailed nitty-gritty, that typifies Nielsen's presentation throughout the book.
If content is king then Nielsen's section on content treats the monarch with the appropriate respect. Something that he emphasizes almost immediately is the importance of professional writing and editing by Web-savvy people. As a sometime writer and editor myself it is usually very clear which sites were put together with the assistance of professional wordsmiths versus those cobbled together by technicians or others without much in the way of written communication skills. Much of Nielsen's advice with regard to content are things that might seem obvious to an experienced writer or editor: keep text minimal and tightly edited, split or "chunk" material to screen-size bites, use plain language and pertinent titles. But for amateur designers or those who are not intimately conversant with both the Web and writing, these are truisms that need emphasis.
Most content is still heavily text-oriented, so it only makes sense to use experts in the world of text to help with content. Increasingly, there are other formats in use, including multimedia, images, animation, video, and audio, and Nielsen has sound advice for integrating all of these formats into Web content. Of course, one of his recommendations for these diverse media is to keep things lean, particularly given the narrow bandwidth of most users' links to the Internet. Video in particular is a huge bandwidth consumer, so it is particularly important to warn the user of the size and time demands of downloads, and to provide alternatives where feasible. For example, Nielsen suggests telling the same story presented by the video using a series of still images with a story line, possibly complemented by an audio clip. This allows users with high bandwidth or high patience levels to download the video, while those who want to move on quickly can get the gist of the video experience through a zippier alternative. Again, this is the kind of prescient, nuts-and-bolts, in-the-trenches advice that is invaluable.
Nielsen notes that although page design gets the most attention, site design is usually more important, because if the design of a site is unintelligible to users they will never see your well-designed pages. This is a sentiment with which I wholeheartedly agree. It is particularly important to have every page reflect its position to the whole, since users will probably burrow deep into your site rather than start at your home page. This is because such "deep linking" is encouraged by the manner in which other sites link to your own, but more importantly, it is a symptom of the fact that most users arrive at your site as a result of use of a search engine.
Even users who start at the entry point of your site need constant awareness of where a page lies within the overall structure. The key questions always to be answered are: Where am I? Where have I been? Where can I go? Although browsers should help answer the question of where I am and where I've been, Nielsen believes they are not particularly helpful. This is one area where I disagree slightly. If the titles of pages are well-chosen, it should be possible to look at a browser history and reconstruct your path through a site. On the other hand, the site itself should assist the user in navigation as much as possible, and the options for further exploration should be clearly displayed.
One of the most important indicators of current location is the name of the site. Usually a site is identified by a name or logo in the upper left hand corner, but it's surprising how often this basic information is omitted or placed somewhere else. Additionally, a miniature sitemap, which can be much more subtle than its name suggests, should be part of any page. Simply displaying a hierarchical set of titles is useful and need not be intrusive. For example, Nielsen illustrates this with an AT&T Web page with the headings "AT&T Easy-commerce Services" and "AT&T Global Alliance Marketing." The Web user is always aware of where on the Web site the current page resides. Again, these indicators are particularly important given how often users will arrive from other sites via deep linking.
Note that if someone has traversed the Web site from the home page down to the current location within its structure, the information marking current location also provides a coarse-grain history of where the user has been. As I suggested above, providing good page titles can help you to see where you've been, because it is difficult for the Web site to provide this information. One aid, of course, is the change in color of links when they have been previously visited. This will aid in either avoiding or revisiting pages depending on the value of their content.
Perhaps the most important user question is "Where can I go?" After all, once you have captured a user's attention you want her to find whatever she needs on your site and not leave for more informative virtual climes. Nielsen offers a great many examples of linking strategies, reflecting the importance of clearly letting the user know what logical choices can be made in moving to another Web page. As discussed below, many if not most users will punt to search features if browsing doesn't work, but others will doggedly move around your site until they either find what they want or leave. So, providing reasonable, clear, and concise links to other pages is a very important task. Nielsen claims, correctly, that your site must have an underlying structure, and that structure should be clear to the site user. It is the links that are found on each page that will best abet the task of revealing site structure. If the structure makes sense to the user he will find it much easier and more intuitive to ferret out the information he needs. For example, if a library site has divisions that include databases and the catalog, and if the database section is further divided alphabetically, it makes it much easier for the user to find the link to Lexis/Nexis from some other part of the site.
One of the tendencies that Nielsen decries is the all-too-frequent presentation of information by a company or institution along divisional and reporting lines rather than in a structure that makes sense to the customer or other user. I have seen this tendency frequently in attempting to find information on hardware and software at vendor sites. Who cares if one printer is made by the office products division and another by the computer division? Not the customer. So why should the customer have to shuttle from one area of a site to another in order to compare pricing and features? It's not quite so easy to make this mistake on a library Web site, although it is easy to wonder why the library is usually an isolated site within a college or university, rather than having deep links from academic departments and other campus units. This seems to be a direct analog to the Balkanization of business sites along operating units rather than function.
Because there are many people who employ searching to find what they want, and because search is a fallback when users don't find information through browsing, it is critical to design your site to include good searching capabilities and, probably more importantly, a good user interface to the search features. Unfortunately, the importance of search within a site is far too often overlooked, which in the worst case can render your site unusable by half its potential viewers. People have very little patience with Web sites that don't directly match their needs and information-seeking strategies, so site builders must take into account the centrality of search to many users' Web behavior. Again, Nielsen is rigorous in his treatment of this aspect of Web-site design, including a discussion of how wide a chunk of your site to search (scoping); the advice that you ought never to provide the user with an opportunity to search the whole Web rather than just your site; advice against use of confusing Boolean search terms at the basic search level; how to lay out the search results page; and, how to use META tagging to describe the content of your pages. This is just the beginning of the detail and rigor with which Nielsen deals with searching on your site.
Nielsen deals with page, content, and site design in the most detail, but he also does a good job in his discussion of intranet design, accessibility, and internationalization. I found only one thing missing, and that is at least a brief treatment of usability testing and its importance. Although Nielsen makes some allusions to the importance of testing, particularly in his sections on accessibility and international use, he really doesn't provide any clues to its importance with respect to Web content. He also fails to offer any information on testing techniques. Although I would not expect any in-depth coverage of testing in a book of this sort, I was disappointed to have almost no mention of the value and practice of testing. I think this is a mistake.
Designing Web Usability is an outstanding book. It belongs on the bookshelf of any Web designer who has any interest in quality Web sites. Nielsen provides quality, cogent advice that should be of use to neophyte and experienced practitioner alike. Buy this book, recommend it for your library's collection. It is well worth the investment.