UI Design Newsletter – January, 2003

Introduction

Sanjay Koyani at the National Cancer Institute and I did a quick survey of the available literature on linking and searching.

We organized our findings into a series of observations and guidelines that may be helpful to designers dealing with similar issues.

Major Observations

Preference for using links versus searching appears to depend on the Web site and the tasks being performed (Spool and Ojakaar, 2001).

When looking for something specific or unique, users prefer to use the search function (Spool, 2002).

Linking tends to be more effective than searching to find content (Ojakaar and Spool, 2001).

Linking through to a target may require fewer clicks than searching (Spool, 2002).

Linking usually results in users exploring more related content (Spool, Schroeder and Ojakaar, 2001a).

Users become less successful with each search attempt (Nielsen, 2001; Spool, Schroeder and Ojakaar, 2001b; Spool and Schroeder, 2001):

About half of unsuccessful searches were the result of users making input errors, which included misspellings, using inappropriate case, spaces and punctuation, misusing plurals, and typing mistakes (Spool, 2002).

Designers and users do not tend to use the same descriptive words. Furnas, et.al. (1984) reported that agreement depended on how the terms were initially determined:

• Without conducting any studies: Up to 20% agreement,
• By conducting a survey of potential users: 20% to 40% agreement, and
• By conducting a survey and allowing up to five synonyms for each term: 40% to 60% agreement.

Advanced search features do not seem to help users. A couple of studies (Nielsen, 2001; Spink, et.al., 1999) have shown that most users tended to employ very simple search strategies, usually by entering one or two words into the search box.

Guidelines

Include both links and a search capability on all key pages of a Web site. Users do not seem to be predisposed to using either linking or searching behaviors; the behavior used appears to depend primarily on the site's content and associated tasks.

When attempting to find information or a product that is not specific and unique, linking is used more frequently and tends to be more successful than searching. Linking requires less effort from users, and frequently allows them to find desired content faster. Also, linking allows users to view substantially more related site content. A good way to help encourage more linking is to provide more links on pages.

For specific, unique items, like a book or CD title, encourage users to use the search capability. This might be best accomplished by how prominently the search capability is positioned throughout the site.

When searching is used:

• Make every effort to ensure that users get relevant results on their first search attempt,
• Be aware of, and make provision for, the terms (keywords) that users typically will use for searching,
• Provide search templates (collections of pre-selected keywords) to help users better formulate their search queries,
• Accommodate misspellings, the use of inappropriate case, spaces and punctuation, misused plurals, and typing errors, and
• Do not rely on advanced search features to improve the effectiveness of a search capability.

References

Furnas, G., Landauer, T., Gomez, L. and Dumais, S. (1984), Statistical semantics: Analysis of the potential performance of keyword information systems, in Human Factors in Computer Systems, Edited by John Thomas and Michael Schneider, 187-242.

Nielsen, J. (2001), Search: Visible and simple, Alertbox, May 13.

Ojakaar, E. and Spool, J.M. (2001), Getting them to what they want, User Interface Engineering Report.

Spink, A., Bateman, J. and Jansen, B.J. (1999), Searching the Web: A survey of Excite users, Internet Research: Electronic Networking Applications and Policy, 9(2), 117-128.

Spool, J. (2002), In search of the perfect search: Building the perfect on-site search, CHI 2002 Tutorial.

Spool, J., Schroeder, W. and Ojakaar, E. (2001), Users continue after category links, User Interface Engineering Newsletter, December 4.

Spool, J. and Schroeder, W. (2001), People search once, maybe twice, User Interface Engineering Newsletter, November 20.

Spool, J. and Ojakaar, E. (2001), Are there users who always search? User Interface Engineering Newsletter, May 13.

So there are some tasks that lead you to make search the primary method of navigation. For example, in book-selling sites, over 80% of users know the title or author. It is easy to put this into search. But in most cases search is a backup. For most sites, frequent use of search is an indication of a broken navigational structure. If the user does not know exactly what they want, search is often time consuming and painful.