Home > Articles

  • Print
  • + Share This
This chapter is from the book

Smoothing the Navigation

Web visitors navigate both on a single site and across sites. They navigate within a single site to find specific information or simply to explore. They navigate the World Wide Web to find a site relevant to their goals or to reach a site related to a currently visited site. In either scenario, the visitor starts with a mental model of a navigation map. In some cases, the map is very primitive and incomplete, consisting primarily of a search engine and some keywords. In other instances, because of experience, the user has in mind a quite well-developed and precise map, as in the case of someone looking for the current DOW numbers on the CNNfn home page.

Human Navigational Strategies. Research on site maps and navigational aids can be useful for designing Web navigational strategies. Let's consider what others have said about human navigational strategies. Wickens (1992) summarizes work by Thorndyke (1980) suggesting that human knowledge acquisition of navigational space progresses along three stages: landmark knowledge, route knowledge, and survey knowledge. To help us navigate, we start by using landmarks present in the environment to find our way and to remind us of how to navigate the same path again. Using distinctive visual clues is the critical attribute of this navigational strategy. In a Web space, titles, labels, link names, and icons can represent landmarks. Route knowledge enables navigators to connect landmarks into a distinctive path, helping them revisit a previously traveled route. Using visual history paths in a Web site can facilitate the acquisition of route knowledge for the visitor. With survey knowledge, navigators form a structural mental model of their navigational space. The inclusion and use of a Web site map would accelerate a visitor's survey knowledge of the site navigational space.

There are at least two different practical ways of presenting paths and site routes on a Web site: linear lists and actual schemata or site maps. Is one more effective than the other as a navigational aid? Bartram (1980) compared the two representations for travelers using the London bus system. In this case, people had to construct a path between two locations from either a list or a map presentation of the route system. The findings indicated that map users made decisions more rapidly than list users. On the other hand, when people were asked by Wetherell (1979) to learn a route from either of the two presentation formats and then actually navigate the route between two locations, the map users made more mistakes. In a Web navigational context, designers need to include both formats. For novices, who are trying to construct a path, it is likely that maps are more effective. For experienced people, who are trying to remember a path, route lists are probably more effective.

Well-Defined versus Ill-Defined Task Statements. One useful way of classifying navigational tasks is on the basis of the completeness of information that a person perceives as available to perform the task. The key word is "perceives." We call a task statement ill-defined to a user if the information necessary to perform the task successfully is perceived as incomplete by that person (Reitman, 1970; Badre, 1974). A well-defined task statement is one that, in the perception of the user, contains all the information required to perform the task successfully. In addition, one of the following two conditions is true: (1) the task statement explicitly says what is necessary for successful performance, or (2) the user can reconstruct the necessary elements for successful performance from memory when presented with the task statement. For example, consider the word-processing task of moving a block of text from one place to another. To a novice, the task statement is ill-defined because it does not explain the sequence of steps required to perform the task successfully. To the experienced user, the statement implies a sequence of actions, available for retrieval from memory, that when executed will permit a successful completion of the task. The task is well defined to the experienced user. It is ill defined to the novice. The task statement could be made well defined to the novice by making it explicit: "Place pointer at first word in block; press F8 to mark word; press F6 to extend; press F8 again."

More precisely, a task may be defined in terms of the triple {I,O,G}, where I stands for initial task state, O denotes the set of admissible operators, and G stands for the goal state. An ill-defined task, one likely to be encountered by a novice user, is a task with a statement that, in the perception of the user, is incomplete in at least one of three specifications: (1) a complete description of the initial state; (2) a complete description of goal-state properties; and (3) the set of operators or rules that permit transformation from initial to goal states. When a user is faced with a task where at least one of these three sets is not completely specified in the task statement—and at the same time, the statement does not trigger a set of memory reconstruction rules—then the task is considered ill defined. At that point the user must move to reformulate the task statement by either (a) completing the specification of the incomplete set or (b) identifying the set of rules required to reconstruct the necessary elements for successful task completion.

Navigating in Real-World Environments. When we navigate in real-world environments, we do so either to explore—often for the thrill of discovery—or because we are searching for a specific object. When we explore, we start with very little information about the task and the environment and with no specific object as the goal of our exploration. We are in an ill-defined task environment. As we navigate, we continuously redefine the environment to make it more familiar by keeping a record, even if a mental one, of our path. We like to be able to retrace our steps, visualize a context for our present position, and wonder what we will find if we go around the corner or over the next hill. When we find something that interests us—an antique store or a restaurant—we stop to investigate. As we explore and gain more information, our mental statement of the task environment becomes less ill defined. We are at the least gaining information about paths and landmarks that we can use to retrace our path to go back to where we started.

We also navigate in the real world to search for a specific object. We search in both ill- and well-defined task environments. Searching for a specific object in an ill-stated task environment is similar to exploring because we are learning to construct paths that we can retrace and to notice landmarks that we can use until we find the object of our search. Searching for specific objects in a well-defined environment requires recognizing relevant cues, remembering landmarks, and retracing previously traveled paths.

Navigating in Hyperspace. Navigating in Web hyperspace is not unlike navigating in the real world. Exploring on the Internet requires selecting an environment to explore, an environment that is defined conceptually rather than physically, as in the real world. For example, if we wish to meander into the area of "usability," we can use a search engine to visit that topic. By doing so, we can be said to be exploring the "usability environment" on the Internet. The constituents of the environment are ill defined, and the goal is simply to explore. On the other hand, if we wanted to know the meaning of the term discount usability, then we are specifying a target goal in an exploratory environment. Once we find the meaning of discount usability in a document on Jakob Nielsen's site (http://www.useit.com/papers/web_discount_usability.html), or we have specific instructions on where to go to find the definition, then we are performing an Internet search in a well-defined environment for a specific object.

Navigational Aids. If users visit a site and find themselves unable to navigate easily and efficiently, they are likely to exit the site and look further for one that meets their navigational goals. How can a designer make a site easier to navigate? We can use several navigational tools to make both exploration and a well-defined search more efficient.

  • Links

  • Buttons and controls

  • Site maps, content lists, and indexes

  • Landmarks and history trails

  • Keywords and site search engine


Well-designed links show visitors how to navigate most effectively the content of a Web site. Accordingly, what a link says and where it is located relative to other links in a site and on a page are paramount to its effectiveness as a navigation tool. Haine (1998) suggests that to be effective and unambiguous, a link's label should indicate both a unique reason for selecting the link and the expected results of selection. In their study comparing the results of the Edmunds and Disney sites, Spool, et al. (1997, URL) report that one reason visitors did better on the Edmunds site is the superior design of the links. The Edmunds links contained more text describing the results of selecting the links. The Disney links were cryptic and terse. Figure 7.6 shows the Edmunds site with its textual links.

It is important that same-site links be labeled in a way that indicates definite distinctions. Visitors should be able to figure out where links will lead from the labeling. Spool, et al. report that link labeling relates to the time and accuracy of selecting a link. They also report that the greater the number of links on a page, the less efficient the search. Larson and Czerwinski (1998) report in their study of experienced users that site organizations with few links in a broad structure were more effective than sites with deep link structures. On the other hand, as we increase the number of links, the efficiency of access to content decreases. Designers must ensure that link destinations are accurate. The link label, or words used in a link, should be the same as the title of the destination page. In addition, some experts have found that confusion can occur when links are used for destinations within the same long page (Spool, et al., 1997; Heath, 1998). My recommendation is that same-page links should be labeled as such or put in a category with a title such as "links on this page."

Buttons and Controls

Navigation buttons and controls, such as "back" and "forward," are included in the navigation bar on the browser, and sometimes they are available on the Web site. As designers, should we prefer that navigation controls be available only on the browser or also on Web pages? It is quite important to keep in mind that the browser interface remains constant irrespective of which Web site the visitor navigates. Because most visitors consistently use the same browser, we would expect visitors to become very familiar with the browser's controls and functionality and to use them frequently. This assumption is all the more likely because not all Web sites include navigational controls on their pages. See Figure 7.7 for the Netscape browser buttons.

Figure 7.7Figure 7.7: Netscape browser buttons (© 2000NetscapeCommunicationsCorporation. screen shot usedwith permission.)

Evidently, however, the results of this observation are mixed. Various studies (Tauscher and Greenberg, 1997; Cateledge and Pitkow, 1995) have reported that the browser back button is used frequently, in 30 to 40 percent of all navigation. On the other hand, the "forward" and "home" buttons are seldom used. This inconsistency in the use of browser controls suggests that visitors may be confused about the functionality on the browser and the functionality within the Web site. Therefore, it is preferable that navigation aids, particularly those related to content (Bachiochi, et al., 1997), be placed within the Web site.

Site Maps, Content Lists, and Indexes

Site maps, content lists, and indexes are content-specific navigational aids that are designed and placed intentionally within a Web site. As mentioned earlier, there is evidence that navigation efficiency improves when content navigation aids are placed consistently near the top of a page. In comparing the effectiveness of maps with content lists, McDonald and Stevenson (1998) determined that visitor performance using maps was superior to that of content lists. The use of either navigational aid yielded superior results over a hypertext condition with no navigation aids. See Figure 7.8 for an example of a site map.

Figure 7.8Figure 7.8: Site map (© 2000 PCChairs, Ltd. Reprinted with permission.)

Furthermore, novice participants who were not knowledgeable in the domain relied more extensively on aids than did knowledgeable participants. It was clear, however, that both novice and knowledgeable participants benefited from the use of a map as a navigational aid. The implication for Web design is that designers should include a site map as a minimum navigational aid in their Web site. The general recommendation is that any site with 20 or more pages should have a site map link on every page where the visitor is likely to enter the site. Some investigators (Allison and Hammond, 1989) have suggested that maps are most useful when visitors are first attempting to gain familiarity with content. Finally, it is important to ensure that links in the site map are correctly named and lead to the intended destination.

Landmarks and History Trails

As discussed earlier, we use landmarks for navigating new environments. Accordingly, designers should supply functional placeholders to help the novice traverse the links and navigational buttons of a Web site. Site functions should be marked by unique icons and labels that visitors can remember easily when they return to a site. This means that landmarks or icons should have distinctive features, such as complex and unique shapes, and should be free standing (Vinson, 1999). Objects such as icons should be concrete, depicting real items (Ruddle, et al., 1997). Figure 7.9 (see also Plate 16) shows some uniquely shaped icons.

Figure 7.9Figure 7.9: Icons (see also Plate 16) (Reprinted with permission of salesforcetools.com.)

Like links, icons should be unique and distinctive relative to other icons and landmarks in the same space. When icons look the same, there is a higher likelihood that navigators will select the wrong one, usually the icon adjacent to the one they want. Darken and Banker (1998) report evidence supporting this navigation error, which they call the "parallel error." The design should also make available a history of pages visited within a site and between sites. Pages should be designed with distinctive and memorable titles. Distinctive titles will help users when surveying a history trail or a list of bookmarks. Avoid the use of frames, which make bookmarking cumbersome. See Figure 7.10 for an example of a history trail.

Figure 7.10Figure 7.10: Web page with history trail (Reprinted with permission.)

Keywords and Site Search Engines

For large sites, 20 or more pages, designers should provide for keyword-based search engines. Spool, et al. (1997) report that when a large site did not have a search engine, visitors gave up on the site. Search engines are most likely to be used by visitors who know what they are looking for but are unfamiliar with the search environment. In such cases, visitors are searching for object(s) or information in an ill-defined environment because the statement of the search problem does not include a specific path or paths to retrace. Figure 7.11 shows the search engine on the upper left of the home page for the Graphic Visualization and Usability Center (GVU) site. Designers should make sure that search engines on large sites appear on the most frequent site-entry pages and that there are multiple search choices. For example, www.metacrawler.com in Figure 7.12 gives "any," "all," and "phrase" searches.

Figure 7.11Figure 7.11: GVU site with search engine (Reprinted with permission.)

Figure 7.12Figure 7.12: Metacrawler site with multiple search choices (© 2001 Infospace, Inc. All rights reserved. Infospace and their designs and related marks are the intellectual property of Infospace, Inc.)

Assuring Reasonable Confidence in the Site's Privacy and Security

Privacy. If site owners track site visitors and collect personal behavior data, then site design etiquette requires that visitors be told about this from the beginning. Site owners should tell potential visitors about the method that they use to collect site behavior data and how the collected information will be used. For example, if "cookies" or temporary ID numbers are used to personalize, the designer should inform the site visitor by providing a site personalization link prominently on the home page and any other frequently visited pages. (Cookies are files that are created and remain attached to a user's browser.) In some cases, full disclosure by site owners is not just a matter of etiquette but a question of legality. For example, the use of cookies is outlawed in many countries, as we discuss in Chapter 11.

Security. Security is a paramount concern for e-business Web site design. Visitors have more confidence in the security of professional-looking sites. Site performance also reveals the level of security (Bouch, et al., 2000). Security technologies, such as SSL, are crucial, and the availability of a link to a "security practices" page helps to reassure visitors. For transaction pages, designers should provide confirmation pages that are easily printable and should remind users to print pages as receipts. Contact information including address, e-mail, and phone number should be prominently displayed. When designing the information needed to complete a transaction, it is important not to ask for more information than is absolutely necessary.

  • + Share This
  • 🔖 Save To Your Account