I expect that quite a number of you have had to endure an Awards Evening. You end up sitting at a table that is far too large to have conversations with anyone else other than the guests on either side of you. This serves to highlight how long the gaps are between courses as the kitchens struggle with hundreds of identical meals all being served at (in theory at least) an identical time. In the distance you can see the tables at which the award judges are seated, usually with a more liberal supply of wine. Then comes the awards ceremony itself, complete with Oscar-emulating speeches in which the recipients thank just about every one in the company, and their families, for making their life so interesting and enjoyable. During this, you are trying to work out the basis on which the awards have been made, and fail.

That had been my experience until last summer, when I was asked to be a judge of the annual awards of the UK Directory Publishers Association. The six judges have to come up with winners in categories for print, CD-ROM, and Internet products, as well as an award for the best marketing campaign. Together with another member of the panel, I was the lead judge in the CD-ROM and Internet categories. The Internet category has of course been growing steadily over the last couple of years, but there are still many CD-ROM products on the market.

The process of judging involved first looking at the submissions in the Internet category over a couple of weeks leading up to the main judging day, and then working through the CD-ROM products on that day. This process highlighted the fact that the content of most of the products submitted was very high, but the usability was variable, and usually quite poor. It is not until you are faced with evaluating over 40 Web-based directories, and nearly 20 CD-ROM directories in a very short space of time that you become aware of the inability of publishers to understand how their products are going to be used. There are some important lessons that need to be learned by Web content publishers, be they in the public Web site arena, or in the intranet arena.

Information Architecture
It all starts with understanding the key elements of good information architecture. "Information architecture" is fast becoming a buzzword, so it is probably worth a definition. To me, information architecture is the process of designing the access to information so that users can rely mainly on their intuition to navigate quickly and productively around the site. When you go through the front door of an office building you expect to find a reception area with company literature and helpful staff, rather than walking into a massive warehouse with products listed in a numerical sequence of code numbers. On the evidence of their Web sites, some publishers have very strange office layouts.

These problems are even more acute in intranets, which have usually not had the benefit of being designed (at least initially) with clear objectives, and have had new content added without any thought about the overall structure and integrity of the site. Far too many intranets are organized on the basis of the departmental structure of the organization because that is how the content publishing operation is managed. This requires users to know implicitly which department has responsibility for certain information, and how that department organizes this information. Even for long-term employees this can be a problem as departments change their names and responsibilities, but for new employees it's too much of a challenge.

Maintain the Brain
The human brain is quite an amazing computer, and like a computer, has both a long-term and a short-term (or working) memory. This working memory has only a limited capacity, a fact first identified by G.A.Miller in 1956 when he was working on the way in which radar systems could be easier for operators to use.

Recently, Kevin Larson and Mary Czerwinski at Microsoft Research looked again at the tradeoff between breadth and depth. Their paper, "Web page design: implications of memory, structure and scent for information retrieval" is published on the Microsoft site at www.research. microsoft.com/users/marycz/chi981.htm. One of the conclusions of their research is that three levels of depth result in significantly more problems during searching than two, regardless of breadth.

There are at least two lessons to draw from this work. The first is that because there is a tradeoff between depth and breadth, just adding new content to what may seem to be a convenient node in the hierarchy without looking at the overall implications for retrieval may be counter productive. The second is that it is important to understand how the brain works when dealing with lists, symbols, and other structural design elements. It is very difficult for designers to place themselves in the situation of someone coming new to an Internet or intranet site. This is why usability testing is so important, a fact that is only gradually becoming obvious in the intranet arena, though still far too many Web sites seem not to have undertaken the most basic of usability tests.

Usability Testing
In my column last issue, I referred to Designing Web Usability by Jakob Nielsen (New Riders Publishing) and Usability for the Web by Tom Brinck, Darren Gergle, and Scott D. Wood (Morgan Kaufmann Publishers). These are currently the two best books on usability issues.

The main point that I want to make under this heading is that usability tests need to be carried out by users, real users. Therefore, this excludes any member of the design team. It is also quite pointless to carry out a usability test by asking a user "What you think about the site?" The information you gain will be virtually no use at all. Usability testing has to be structured around the tasks that a user will undertake on the site. Carrying out usability tests with a group of users is also of limited value unless your site is designed to be used by five people simultaneously.

The tasks users perform will vary according to their role in the organization, and their existing experience and expertise. Developing profiles of user segments is, therefore, very valuable. In an intranet situation, it's important to test the site with both long-term employees and with new hires, and to understand what use will be made of the information once it has been presented on the screen. Many of the directory products I alluded to earlier seem not to have recognized that users will want to print out, or save to a file, the information that they have retrieved, and that record formats have to take this into account.

It's All About RAD
Another mistake that many Web designers fall into is to take a user specification, and then go away and spend the next few months developing a site that exactly meets the specification. The result is usually that the organization has moved on and user requirements have changed, or that the user now realizes that their specification was not appropriate. The only way to develop Web applications that are usable is to use a Rapid Application Development (RAD) approach. In essence, this involves incremental testing against specification, a realization that an 80% solution to a deadline may well be better than a 95% solution several months late, and that a 95% solution to a user requirement is unattainable in the current business environment.

Intranet users are especially critical of poor usability, and rightly so. There are none of the problems associated with undertaking usability tests with current and potential customers. However, in my experience, less attention is paid to usability in the intranet environment on the basis that all users will know just where to find the information they are looking for, and they are not under the sort of time pressures that Web visitors are faced with. Wrong! Internal productivity gains are of great value to the organization.

Humour me. Write down the number of intranet users you have. If you assume that good usability will save ten minutes a day in terms of productivity then you will gain a week of time per employee. So divide the number of users by 50 and that is the net gain in head count. That is why good usability counts.