Guest post by Seth Earley and Jeannine Bartlett

Search has evolved dramatically over the past couple of years. Many people still think of search as the white box in the corner of the page where users type in a term or phrase, with a Google-like page then appearing. However, we need to think of a broader concept of findability--ways that we can facilitate the user navigating and searching for the information they need and iterating through the process.

The old advanced search vs. facets

There was the "advanced search" button that brought users to another page where various parameters could be added--such as preferences for a size range, brand or style. There were several problems with this. One was that users did not like to click to go somewhere else. Another was that the interface was complex, could allow for nonsensical choices (like searching for wine that is both "white" and "Cabernet") and had a combination of choices that would frequently lead to zero results, taking the user could go down lots of blind alleys.

In the retail online environment, this has evolved into faceted search or guided navigation. The name of this approach reveals its elegance--that is, tricking the user into thinking that they are navigating when they are actually conducting a search. A couple of usability tricks makes this happen--the first is that the user does not have to go anywhere--the search terms are presented as navigational links. Another is that if a combination of search terms doesn't have results, they will not show up together. A third touch is that the numbers of results are shown next to the search term.  

Faceted search is deceptively complex to get right. We need to first consider what facets to search on. Some of these are obvious--size, color, brand, style, capacity and so on. Others are less obvious and more subjective--like lifestyle, purpose, task and demographic details.

There are many attributes of a product that may be interesting to know, but may not be of value in faceted search. Is a dimension an important search parameter? Perhaps it is for a refrigerator. It depends upon the application. If I have many parameters, how will those be presented to the user? In what order? Shopzilla presents many parameters in the left-hand navigation but some are presented in a new window and then added to the left-hand nav. 

In what order do I present facets? At what level of granularity? Do I use a hierarchy? How do I let users remove terms from the search? Add facets to the search? 

These questions also point to the need to have a solid, well constructed taxonomy and set of controlled vocabularies, since all of faceted search/guided navigation is driven by these terms. 

Best Bets and disambiguation

If users can't locate what they want in guided navigation, they will use the search box. Another challenge around search is that search terms are usually ambiguous. Searchers don't describe what they want very well. They will use a very broad term when they want precise results. One example might be "Gifts". Gifts for what occasion, for whom? Users might add a term to disambiguate such a broad query, or facets can be used after the search results are presented, in order to refine the result set according to various parameters, like men's, women's, children's and price range.

Search can also leverage Best Bets in which full text and parametric search is bypassed and instead a landing page is "hard-wired" to provide promotional items, specific information about a concept or a location where users can further navigate according to topic-specific parameters.

Another approach for giving users what they want rather than what they are asking for is to use so-called "associative relationships" from a thesaurus. These are the "see also" relationships to present related or complementary products. This can be used to facilitate cross- and up-sell opportunities.  

Content related to products

One area that presents great opportunities for retailers to differentiate the online experience is in integrating unstructured content into their product catalogs. This is done by tagging content according to specific product types or at the product level in order to give shoppers additional contextual information. These may be third-party product tests, product reviews, detailed instructions, documentation, warranties, service data or assembly instructions.

These approaches illustrate how site search, content strategy, taxonomy, metadata and information architecture all work together to support the site experience.  

Good search = satisfied customers = increased sales

The bottom line is that customers come to a site in order to locate information about the products they want to purchase. If they cannot locate their products or there is little information to support the buying process, then users will abandon the site and go elsewhere. Content needs to be presented in the context of search terms that represent the user's intent. If we don't understand what users are trying to accomplish then we won't be able to use terminology that resonates with them and they will be dissatisfied with the search experience.  

Site search effectiveness is a key factor in user satisfaction. If users can't find it, they can't buy it. They will be turned off by a poor site search experience if results are inaccurate, if there are too many or too few results, if they cannot further refine the query to get more precise results, if the search interface does not allow them to easily broaden the query, if the search system is intolerant of misspellings or if the system does not understand synonyms.  

In order to tune search, we need to understand users, their tasks and the preferred and variant terminology they use. We need to test terms outside of the search and navigational interface for clarity and intuitive consistent organizing principles and develop use cases for specific types of users engaging in product research and purchasing activities. We need to map their mental model to an information architecture that allows content to be associated through the search process. By getting these details right, online retailers can measurably increase sales of specific items by improving findability. 

Seth Earley is the president & CEO, Earley & Associates, Inc. Earley is the country's leading expert and trend-setter in content management and knowledge management practices. A leader in the field who paved the way for many of today's taxonomy practices and principles, Earley has developed search, content and knowledge strategies and underlying taxonomies for Fortune 500 companies.

Jeannine Bartlett is chief solutions architect at Earley & Associates, Inc., and has over 20 years of experience designing and implementing search, information management, content architecture and security systems. For more information, please visit www.earley.com.