Chapter 17

Search

Many Web sites allow users to search for information contained in the site. Users access the search capability by entering one or more keywords into an entry field—usually termed a ’search box.’ When there are words in the Web site that match the words entered by users, users are shown where in the Web site those words can be found.

Each page of a Web site should allow users to conduct a search. Usually it is adequate to allow simple searches without providing for the use of more advanced features. Users should be able to assume that both upper- and lowercase letters will be considered as equivalent when searching. The site’s search capability should be designed to respond to terms typically entered by users. Users should be notified when multiple search capabilities exist.

Where many users tend to conduct similar searches, sometimes it works best to provide search templates. Users tend to assume that any search they conduct will cover the entire site and not a subsite. The results presented to users as a result of searching should be useful and usable.

Research-Based Web Design & Usability Guidelines

17:1 Ensure Usable Search Results

180

Relative Importance:

Guideline: Ensure that the results of user searches provide the precise information being sought, and in a format that matches users’ expectations.

Comments: Users want to be able to use the results of a search to continue solving their problem. When users are confused by the search results, or do not immediately find what they are searching for, they become frustrated.

Sources: Amento, et al., 1999; Bailey and Koyani, 2004; Dumais, Cutrell and Chen, 2001; Nielsen, 2001a; Nielsen, et al., 2000; Pollock and Hockley, 1996; Rosenfeld and Morville, 2002; Spool, et al., 1997.

Example:

Strength of Evidence:

Search

Returned search results in the main panel contain snippets of the searched page with the user’s search terms highlighted (allowing the user to gain a sense of the context in which the terms are used) and a clustered list of related search terms is contained in the left panel.

These search results are difficult to use. There is no discernable order and no ability to sort results by characteristics (e.g., price, size, etc.)

See page xxii for detailed descriptions of the rating scales

17:2 Design Search Engines to Search the Entire Site

181

Guideline: Design search engines to search the entire site, or clearly communicate which part of the site will be searched.

Comments: Designers may want to allow users to control the range of their searches. However, users tend to believe that a search engine will search the entire Web site. Do not have search engines search only a portion of the site without clearly informing users which parts of the site are being searched.

Keep in mind that what a designer may consider to be the entirety of a site may not be the same as what the user thinks is the ’whole’ site. For example, many large sites have various subsections that are maintained by different designers, so the user may think of a site as something that designers think of as several sites. Make sure it is clear to users what part(s) of the Web site are being searched. Provide a means for users to narrow the scope of searches on large Web sites by providing easy access to specific subsites when searching.

Sources: Bailey and Koyani, 2004; Spool, et al., 1997.

Example:

Relative Importance:

Search

Strength of Evidence:

This design allows users to easily bound their search to a selected subsection of the Web site, or to run an unbounded search by selecting the ‘All of SSA’ menu choice.

17:3 Make Upper- and Lowercase Search Terms Equivalent

Guideline: Treat user-entered upper- and lowercase letters as equivalent when entered as search terms.

Comments: For example, ’STRING,’ ’String,’ and ’string’ should be recognized and accepted equally by the Web site. When searching, users will generally be indifferent to any distinction between upper- and lowercase.

The site should not compel a distinction that users do not care or know about, or that the user may find difficult to make. In situations when case actually is important, allow users to specify case as a selectable option in the string search.

Sources: Smith and Mosier, 1986.

Relative Importance:

Strength of Evidence:

Research-Based Web Design & Usability Guidelines

17:4 Provide a Search Option on Each Page

182

Guideline: Provide a search option on each page

of a content-rich Web site.

Comments: A search option should be provided on all pages where it may be useful–users should not have to return to the homepage to conduct a search. Search engines can be helpful on content-rich Web sites, but do not add value on other types of sites.

Designers should be careful not to rely too heavily on search engines. They are not a substitute for good content organization, and do not always improve users’ search performance. Designers should carefully consider the advantages and disadvantages of including a search engine, and whether their Web site lends itself to automated searches.

Sources: Detweiler and Omanson, 1996; Farkas and Farkas, 2000; Levine, 1996; Nielsen, 1996a; Nielsen, 1997e; Nielsen, 1999d; Spool, et al., 1997.

Example:

Relative Importance:

Search

Strength of Evidence:

As users delve deeper into the site’s content, the search capability remains immediately available.

See page xxii for detailed descriptions of the rating scales

Research-Based Web Design & Usability Guidelines

17:5 Design Search Around Users’ Terms

17:6 Allow Simple Searches

183

Guideline: Construct a Web site’s search engine to respond to users’ terminology.

Comments: Users seem to rely on certain preferred keywords when searching. They will generally conduct one or two searches before trying another site or search engine (or abandoning the search altogether). Therefore, it is important that users succeed on their first try.

Determining the keywords users are using may require considerable data collection. Designers should make use of search engine logs, surveys, and other techniques to determine the preferred search words for their site, and make information relevant to those terms easy to find through the site’s search engine. Keep in mind that designers’ preferred keywords may not match users’ preferred keywords, and content writers may overestimate the specialized vocabulary of their audience. For the most common searches, provide a ’best bets’ set of results. Ensure that the ’best bets’ do not appear as advertising or paid links.

In addition to responding to users’ keywords, try to design the site’s search engine to accommodate common misspellings, extra spaces, alternative punctuation, misused plurals, and other common user search errors.

Sources: Bailey and Koyani, 2004; Dumais, Cutrell and Chen, 2001; Egan, et al., 1989; Evans, 1998; Hooke, DeLeo and Slaughter, 1979; Koyani and Nall, 1999; Schiano, Stone and Bectarte, 2001; Spyridakis, 2000.

Example:

Relative Importance:

Search

Strength of Evidence:

A search for “tongue cancer” also returns results on Oral Cancer, Head and Neck Cancer, and Lip and Oral Cavity Cancer.

Research-Based Web Design & Usability Guidelines

17:6 Allow Simple Searches

184

Relative Importance:

Guideline: Structure the search engine to accommodate users who enter a small number of words.

Comments: The search function should be easy to use and allow for users to be successful when searching. Most users tend to employ simple search strategies. They rarely use advanced search features (such as Boolean operators), so it is important not to rely on those to improve the effectiveness of the site’s search engine. If most of the site’s users are inexperienced Web searchers, provide simple instructions and examples to help guide users’ searching and use of the search results.

Provide a box (entry field) for entering search terms that is at least 35 to 40 characters wide. Users will self-detect more errors when they see what they have entered.

Sources: Bailey and Koyani, 2004; Bayles and Bernard, 1999; Koyani and Nall, 1999; Nielsen, 2001a; Nielsen, et al., 2000; Pollock and Hockley, 1996; Spink, Bateman and Jansen, 1999; Spool, Schroeder and Ojakaar, 2001.

Example:

Strength of Evidence:

Search

Simple search engines will accommodate most users’ search strategies.

This search page is far too complex for the average user. Such advanced search capabilities are best presented on a page dedicated to advanced searches.

Research-Based Web Design & Usability Guidelines

17:7 Notify Users when Multiple Search Options Exist

185

Guideline: If more than one type of search option is provided, ensure that users are aware of all the different types of search options and how each is best used.

Comments: Most users assume that a Web site has only one type of search. In one study, when there were multiple search types available, users tended to miss some of the search capabilities.

Sources: Bailey, Koyani and Nall, 2000; Levy, et al., 1996.

Example:

Relative Importance:

Search

Strength of Evidence:

These sites all offer multiple ways of searching.

See page xxii for detailed descriptions of the rating scales

Research-Based Web Design & Usability Guidelines

17:8 Include Hints to Improve Search Performance

186

Guideline: Include specific hints to improve search performance.

Comments: A major tradeoff that must be considered in the design of a search input interface is related to the need to provide sufficient instructions for users to take advantage of the power of the search engine, while keeping in mind the reluctance of users to read instructions.

One study found a direct link between the content of search hints and task effectiveness. When syntactic information was included in the search hint, participants achieved significantly greater syntactic performance. When semantic information was included in the search hint, participants achieved significantly greater semantic performance. In addition, participants’ confidence that their queries would retrieve the correct answer was reliably enhanced by the presence of semantic search hints (but not syntactic hints). The presence of examples improved semantic performance, but had no effect on syntactic performance. When hints contained more than one type of information (syntactic, semantic, or examples), performance was generally lower than when only one hint type was presented. Also, participants were able to complete the search tasks faster when only one hint was presented.

Sources: Bandos and Resnick, (2004).

Example:

Relative Importance:

Search

Strength of Evidence:

This site provides search hints to assist the user.

See page xxii for detailed descriptions of the rating scales

Research-Based Web Design & Usability Guidelines

17:9 Provide Search Templates

187

Relative Importance:

Guideline: Provide templates to facilitate the use of search engines.

Comments: Search templates assist users in formulating better search queries. A template consists of predefined keywords that help users select their search terms. The keywords can be used directly, or can help users formulate their own queries. Each template should be organized as a hierarchy of predefined keywords that could help to restrict the users’ initial search sets, and improve the relevance of the returned ’hits.’ One study reported that people using templates find seventy percent more target Web sites than those not using templates.

Sources: Fang and Salvendy, 1999.

Example:

Strength of Evidence:

Search

Some ‘search template’ examples include:

To find information on ’human error’ use

errors fault miscalculation

slips blunder slip-up

mistakes inaccuracy

To find information on ’usability testing’ use

user interface testing cognitive walkthroughs

performance testing automatic tests

heuristics evaluations remote testing

Research-Based Web Design & Usability Guidelines