Site Structure - Page 14

No matter what navigation design you pick for your site, there is one common theme to all navigation: All it does is visualize the user's current location and alternative movements relative to the structure of the underlying information space. If the structure is a mess, then no navigation design can rescue it. Poor information architecture will always lead to poor usability.

Most sites have a hierarchical structure with progressively more detailed levels of information. Other sites have a tabular structure in which pages are classified relative to a number of attributes or parameters. For example, the 1996 Olympic Games site classified events pages relative to their sport, their data, and their location, so users could, if they wanted, see all pages related to soccer or all pages related to events in a certain city. A linear structure makes sense for Web-enabled applications that are a progression of steps.

The Vice-Presidential Button

A classic sign of a mismanaged website is when the home page has a button for each of the senior vice presidents in the company. Remember, you don't design for your VPs; you design for the users. Therefore, it will be quite common that you can't tell VPs where "their" button is on the home page.

The two most important rules about site structure are to have one and to make it reflect the users' view of the site and its information or services. It may seem obvious to have a site structure, but many sites evolve without any planned structure and end up in total chaos as a collection of random directories without any systematic relations among different parts of the site. A second common mistake is to have the site structure mirror your organizational charts instead of reflecting the user's view. Users should not have to care how your company is organized, so they should not be able to deduce your organizational structure from the structure of your website. Admittedly, it is easiest to distribute responsibility for the site to divisions and departments according to already established chains of command and budget categories, but doing so results in an internally centered site rather than a customer- focused site.

The site structure should be determined by the tasks users want to perform on your site, even if that means having a single page for information from two very different departments. It is often necessary to distribute information from a single department across two or more parts of the site, and many subsites will have to be managed in collaboration between multiple departments.

A linear information structure is often a warning sign for bad usability because the Web is inherently non-linear in nature. Users don't want to have to step through all the site elements one at a time. In this example, it is hard to imagine a user scanning through a list of 753 films in alphabetical order. It would be better to have a set of alternative structuring principles available, including a way to sort by quality (review) rating and a way to filter out certain types of films. Even if an alphabetical listing is the best that can be done, at least abandon the numeric links to jump around the list. It makes sense to jump to films starting with the letter G, but not to jump to film number 451.

In April 1998 The Christian Science Monitor experimented with a site structure and navigation interface that mirrored its printed newspaper. Users would page through reduced pictures of actual pages from the daily paper and could click on the image of an article to get the full text displayed in the right half of the window.

The main problem with the design is that it is extremely slow to navigate. Each of the page miniatures is about 60 KB, taking about 20 seconds to download over a 28.8 modem. We know that 10 seconds is the absolute maximum response time for getting web pages before users rebel; optimal navigation requires even shorter response times.

Navigating between pages is done through a pop-up menu that lists nothing but page numbers. Because users do not know what articles are on what page, they are left guessing and are relegated to jumping to pages at random. It would have been better to provide a menu of section titles or main headlines. For example, it would be more meaningful to go to "Food Section" than "Page 14", assuming that the food articles were on page 14. Page numbers make sense in print, where users flip through a physical product, but on the Web, everything is equally far away (one mouseclick). Thus, a site structure should rarely be linear.

Furthermore, the Monitor design did not allow users to get a quick overview of the current news or focus their attention on parts of the paper that were of particular interest to them. It was a purely linear navigation system that was made almost useless by the slow speed of "turning the page." While viewing a page, it was not possible to read subheads, decks, bylines, or any of the other short but important components of a story design that help people decide what to read.

After a user did manage to find an article of interest, then the design failed again because only half of the window was available to display the actual article, thus necessitating more scrolling than usual and making the text harder to scan. Also, the use of frames made it difficult to bookmark articles of interest or to email URLs of recommended stories to friends and family (otherwise, a great way for a website to grow usage through social interaction).

Ironically, even though I would not recommend this design for any newspaper, it may be particularly ill suited to the Monitor. Most of its pages contain no more than one or two articles. Thus, the page layout gives very few cues as to the relation between articles or particular attributes of stories. Other newspapers have more intricate layouts where the relative placement of stories on the page carries more information.

Navigation Support in Browsers - Page 13
Designing Web Usability
Hierarchical Structure - Page 15