Secrets to Site Search Success

Paul Bruemmer

Most of us use the search box on an ecommerce site to find the product or service we want — if we can find a search box, that is. When we do, most search attempts will yield either no results, or too many irrelevant results.

Forrester reports that 93 percent of ecommerce sites either don’t provide site search facilities, or use software that fails basic tasks. Jupiter MediaMetrix reports that:

“80 percent of online users will abandon a site if the search function doesn’t work well.”

As over 50% of online buyers use a site’s search functionality to find products, it appears that ecommerce sites continually let customers down.

Site Search Failures

Most ecommerce site search tools fail shoppers in one (or more) of five areas:

    1. No search – the site fails to offer any kind of search tool.


    1. Weak technology – the tool uses weak technology that’s either unforgiving or too aggressive. Search results yield too many options, or none at all.


    1. Inadequate knowledge base – the tool is based upon inadequate background knowledge – for instance, common language or defined relationships.


    1. Poor content structure – the tool must attempt to search poorly managed content, including inconsistent tagging, redundant titles, and unusable descriptions.


  1. Bad interface design – the user is forced to deal with unfriendly user interfaces (providing unhelpful input, inflexible output, and no means to take action).

If your site’s search function fails in any of the above areas, it should be investigated and fixed as soon as possible. And if site search is missing, it should be included in next year’s budget.

Due to the fact that online buyers want user-friendly search functions that work, and many online retailers don’t provide them, a site search is an excellent up-sell for Webmasters who design or upgrade sites. Let’s take a look at some tools.

Six Site Search Resources

There are many providers of site search tools, and new products are always entering the market:

    1. AltaVista Search Engine 3.0 is designed with a scalable, 64-bit multiprocessing architecture that integrates easily with back-end databases and business applications. It helps users find data in 30 languages and in over 200 file formats, including databases and ecommerce catalogs. AltaVista has over 1,200 customers, including HP, Arriba, Nordstroms, and Ticketmaster. It also has partner agreements to integrate its enterprise software with the BEA WebLogic Portal, Oracle’s Enterprise Portal, and Novell’s Portal Services.
  1. Ask Jeeves initially offered its natural language search technology to corporate customers as a service. Now it sells JeevesOne software for enterprise Web solutions. Jeeves Solutions President Claudio Pinkus said more companies prefer to operate their own software applications than buy search services, because potential gross profit margins are higher (90 percent versus 60 percent). Customers search for information through an Ask Jeeves-type question box, which allows retailers to learn information about their customers through the questions they ask. JeevesOne starts at $100,000 with add-ons at $50,000, and you can request an online demo. Customers include Dell, Nextel, and Ford.
  2. EasyAsk will accept keyword or natural language input, matching these requirements to your entire site’s product catalog. Over 90 percent of searches are accurately satisfied, because EasyAsk uses all possible information, including product category, requested attributes (price, color, size.), product description, and more. Prices for the software start at $30,000, and an interactive product demo is offered online. Customers include Lands’ End, Coldwater Creek, and by eVinyard. EasyAsk also recently announced a wireless question/answer technology that lets mobile users find information when away from the desktop.
  3. Inktomi Search Software offers a powerful and scalable natural language search engine for thousands to millions of pages. Options include a CCE (content classification engine) for the creation of directories and display of search results with categories. You can also adjust relevance rankings and index database content, spider SSL-protected servers, and search multiple language documents. Licensing costs vary by database size, starting at $2,995 for 3,000 documents with no penalty for updating to a larger index. Try Inktomi Search Software for free by downloading a 30-day evaluation copy. Customers include Boeing, CNN, Yale University, and many more.
  4. Mohomine builds licensable and OEM software products and solutions for ebusiness unstructured data management. Mohomine software enables enterprise applications to access and intelligently use unstructured data. Customers include Qualcomm, Peoplesoft, and Sourcebank.
Free Search Tools

Smaller ecommerce sites should also provide site search. If the above solutions prove too expensive, consider these two free site search options:

    1. Atomz offers a free search engine service for smaller sites of less than 500 pages and does not have an ad-viewing requirement. You can quickly add a powerful search engine to any Website with just a few lines of HTML and no programming.
  1. FreeFind is another free search engine, but requires that users view ads. A professional version is available without advertising.
Evaluating Site Search Technology

Those evaluating site search technology will want to look at price, platform, capacity, ease of installation, and maintenance requirements. This requires coordination between the company’s top administrative and technical managers, their product or department managers, and the Webmaster, to create a plan with a timeline.

The next step is to screen available search tools based on your requirements and compatibility to your existing systems and site infrastructure. This involves defining preliminary and end-user requirements, checking a database of site search tools, and choosing the most appropriate options for further evaluation.

It’s helpful to develop a technical requirements document for end-user needs, administration, cost of ownership, vendor reliability, hardware and OS compatibility.

Once technical requirements are established, evaluate the options based on this document. Some product features you might consider are:

  • Advanced machine learning: content-based information retrieval and organization.
  • Customization: ability to personalize the tool to your site’s characteristics and users’ needs.
  • Topic-specific knowledge mining: client-defined content.
  • Automation: scalable, machine-driven knowledge management procedures.
  • Information extraction and summarization: the ability to extract meaningful content.
  • Handcrafted taxonomies: the ability to classify information in a way that’s relevant to the methods your users utlize to search.
  • Parametric searching: the retrieval and organization of information by textual qualities and properties.
  • Hosted ASP models: the elimination of the need for companies to purchase, maintain, and upgrade knowledge management technology in-house.

After you evaluate your selected options, install test versions of the final candidate products and perform automated and manual user tests. Evaluate the results of the top three; then select, customize, and install your new search tool. As a last step, don’t forget to publicize the new search tool and feature it prominently on the site.

Billions of dollars in online purchases are lost because of a lack of search tools or badly designed tools that yield poor results. Don’t fall into this trap. eCommerce sites can turn one-time visitors into loyal, repeat customers by helping online buyers find exactly what they’re looking for.