April 25, 2011, 4:22 a.m.
posted by mage
J3 Organized Search Results
A visible SEARCH ACTION MODULE (J1) and a STRAIGHTFORWARD SEARCH FORM (J2) address some of the common problems that site visitors have using search engines. However, it might still be difficult for your customers to comprehend search results, especially when there are a large number of them. Structuring and organizing search results can make them much easier to understand. This pattern covers ways of arranging and categorizing search results to make them more valuable to your customers. It applies to both local searches on your site and Web-wide search engines.
It can be difficult for site visitors to understand search results if there are too few or too many results.
A search engine is only as good as the results it presents. It does not matter if it is fast and can simultaneously support thousands of queries per second if people cannot understand the results.
Provide Relevant Summaries with the Search Results
The Zagat restaurant guides are indispensable for many city dwellers, providing an overview of all the major restaurants in a metropolitan area. Zagat provides a high-quality online version of its popular restaurant guidebooks. Figure J3.1 shows how zagat.com provides useful, domain-specific information in its search results, including a short excerpt from the full review, as well as ratings of the food, décor, service, and cost of the restaurant.
Figure J3.1. In its extensive online restaurant guide, Zagat lets its customers sort by name, cuisine, area, food, décor, service, and cost. It also shows a restaurant's address and phone number, as well as a short excerpt from the review.
(www.zagat.com, August 21, 2001)
There are no hard rules here, though. You have to determine what should be shown in search results on a per Web site basis. On e-commerce sites, for example, it makes sense to show the price and availability of products. And search results on community message boards could show the author and date of posting.
Offer Clear Organization of the Search Results
No one likes sifting through hundreds of hits to find the right one. One way of addressing this problem is to group the hits according to a coherent, logical scheme, such as alphabetically or chronologically. For example, Figure J3.2 shows how Amazon.com groups related search results together. This is one way to make sure your search results present customers with BROWSABLE CONTENT (B2).
Figure J3.2. Amazon.com groups search results together by category. On a search for "addison wesley computer," the various hits are grouped by Books and zShops.
(www.amazon.com, May 8, 2002)
Provide Good Hyperlinked Titles for Each Hit
Search engines typically display a Web page's HTML title as the name of the search result. Because these titles are what visitors see, it is crucial to give your Web page a DISTINCTIVE HTML TITLE (D9).
Use Log Files to Tailor Results for the Most Common Search Terms
Search engines by themselves do not always provide the best results. The key to better results for your customers lies in your analysis of the log files, which keep track of the terms your customers use to search.
Find the most common terms, and make sure that search results using those terms point to the right place. This approach solves two problems. First, it compensates for the fact that customers use different words to mean the same thing. Figure J3.3 shows how IBM does this on its Web site. The search terms "notebook," "laptop," and "ThinkPad" all point to the same page, greatly increasing the chance that a customer will find the right page.
Figure J3.3. IBM presents specially tailored search results for common searches, such as "notebook" and "laptop," in addition to the standard search results.
(www.ibm.com, April 2, 2001)
Second, this approach lets you create the best possible hit for search terms. Look at Figure J3.3 again, and you will notice a special area for tailored search results. Whenever a customer searches on "notebook," "laptop," or "ThinkPad," the search page shows a special promotion for selling these computers. However, the regular searches on the site show only ordinary search results. IBM keeps these ordinary search results in case their customers' "best" hits are not what they're looking for.
Compensate for Common Misspellings
Figure J3.4 shows how Amazon.com handles misspelled search terms. At the top of the Web page, it notes that there are no search results for the exact search terms entered, but then it corrects the spelling and searches on the new terms. It is likely that customers will not notice that they misspelled the terms, nor will they see the message at the top of the Web page stating that they misspelled the terms, but Amazon.com does the right thing and shows them the results anyway! This is a customer-savvy, elegant design.
Figure J3.4. Amazon.com automatically corrects certain misspellings. In this case, it corrects the search "crossing cashm" to read "crossing chasm."
(www.amazon.com, April 14, 2001)
Compensating for spelling mistakes means you have to tailor the spell checker's dictionary to your specific domain. An ordinary spell checker dictionary will not be useful for Web sites dealing with legal matters, for example, because it will not have legal terms in it.
Another way to compensate is to go through the log files to find the most commonly misspelled search terms. You will have to manually enter these misspellings into the search engine to make them point to the right terms (see Figure J3.4). This is clearly a tedious proposition, especially if there are a huge number of search terms. Concentrate on finding the most popular terms, perhaps by limiting your corrections to the top 100 or 500.
Provide Support for Common Search Tasks
Common search tasks include things like going to the next page of results, starting a new search, and refining a current search. Figure J3.5 shows a good example of how the Google search engine handles these tasks. The top of the page provides a search entry form, letting visitors quickly create a new search if they did not get the results they wanted. In addition, the new-search form includes the search terms that the customer originally typed in, making it easy for them to remember what they entered and to edit it if necessary.
Figure J3.5. Google has a clean, minimalist design for displaying search results and supporting common tasks.
(www.google.com, April 15, 2001)
The bottom of the page has a second search entry form because the customer may look through all of the search results for a page, reach the bottom, and then decide to start a new search or refine the existing one. This design is more considerate than forcing the customer to scroll back to the top of the page.
The bottom of the page also contains links to the next set of search results. One subtle but very useful thing that designers have done here is to make the "gle" part above the Next button a link to the next search results. The upshot is that there is a fairly large target to click on that will take customers to the next page of search results. Incidentally, designs with large targets are also good for improving SITE ACCESSIBILITY (B9) for customers with poor motor control.
At first glance the controls at the bottom might seem to argue against designing for ABOVE THE FOLD (I2). But visitors tend to read the page from top to bottom. If they have not found what they want by the time they reach the bottom, then they can go to the next page of results.
Provide your customers with relevant summaries in their search results. Offer a clear organization of the search results. Provide hyperlinked titles for each hit on the search result page. Use log files to tailor the search engine for the most common search terms. Compensate for common misspellings. Provide support for your customers' common search tasks.
Figure J3.6. Organize search results, and continue to update the search database with common synonyms.
Consider These Other Patterns
You can also use the card-sorting techniques described in Chapter 3—Knowing Your Customers: Principles and Techniques to find common search terms, but do this before launching the site and before performing a log file analysis. This strategy can help you prevent errors, instead of waiting for them to happen and then fixing the problems after the fact.
Use the SITE ACCESSIBILITY (B9) pattern to ensure that the results you return can be used by all of your potential customers, including those with disabilities.