Enterprise Search: Rely On Strategy, Not Keywords

The word Google has become synonymous with the concept of search for most people. And why not? Google is accessible, easy to use, and in most cases compiles fairly accurate results for the user. The use of Internet search tools like Google, or even Yahoo! or MSN for that matter, has become so commonplace that it has formed the basis of most users' perception of search. Not surprising, most users expect enterprise search tools to provide the same Google-like experience. The reality, however, is that Web search and enterprise search are not at all similar. As David Haucke, VP of global marketing for ISYS Search Software (ISYS) explains it, "While enterprise search can encompass information from the World Wide Web, its focus is on searching information contained behind the firewall."

Don't Depend On Federated Search
Enterprise search is all about the ability to find an organization's electronic information wherever, or however, it may be stored. Much as an Internet search tool is designed to search the World Wide Web for relevant websites, enterprise search is intended to give a company's employees the ability to recall key information from internal documents, emails, databases, and more. Sounds simple, right?  Unfortunately, enterprise search is anything but simple.

"Many organizations have attempted to provide enterprise search by relying on a federated search infrastructure," says Alex Marshall, VP of technology for Autonomy Corporation. A federated search system is intended to enable users to search varied and numerous databases with one query. It is also intended to display the results of the search neatly packaged in order of relevance. While users dream of such a simple search engine in the enterprise, federated search has yet to prove itself as a comprehensive, easy-to-use solution.

"While federated search may be an appealing quick fix, this actually exposes the enterprise to significant risk," says Marshall. "By federating searches to native search engines, one assumes that these engines are capable of effectively searching their own data. Given that many repositories rely on outdated, end-of-life search products for their native search, this is not a reasonable assumption to make." According to Marshall, native search engines rarely search all the information contained in a repository. This can lead to compliance issues, as well as inefficient and incomplete search issues. For example, the FRCP (Federal Rules of Civil Procedure) mandate that companies must be able to locate and produce all ESI (electronically stored information) in the event of pending litigation. Since a federated search strategy may not search all of the data contained in the company's repositories, a federated search threatens compliance.

Develop An Enterprise Search Strategy
"Industry analysts are forever encouraging customers to develop a strategy for enterprise search — long before they even get to the stage of selecting a vendor and implementing the technology," says ISYS' Haucke. Developing the strategy involves gaining a thorough understanding of the problems to be solved, the data types to be included in the search, and the number and roles of employees who will be affected. Companies must also take into consideration the resources required, both financial and technical, as well as any other infrastructure requirements. "The outcome of strategy development is a plan that outlines all of the company's key requirements," continues Haucke. "Too often, we see customers reviewing too broad a spectrum of vendors.  It's analogous to a solo bookkeeper reviewing both QuickBooks and a high-end JD Edwards financial application for his business." Understanding in advance what the company will require from an enterprise search solution will both streamline the vendor selection process and ensure the actual implementation and post-launch period proceed as smoothly as possible.  

Autonomy's Marshall agrees that an enterprise search strategy is the important first step. Without it, companies can unintentionally — and easily — create silos of information that can again expose a company to compliance risks. "Having multiple search infrastructures is not only costly and cumbersome, but also introduces more risk as information is copied, moved, and often stored in perpetuity," says Marshall. "Each of these systems has to be tapped into separately for enterprise search, knowledge management, and e-discovery, and indexing and analysis of information is performed redundantly. All of this can cost a business significant amounts of time and many millions of dollars when different systems can't be automatically related to one another."

Above all, an enterprise search solution that doesn't work or is not adopted by the company's employees can be a high-cost failure. If you're a business that maintains fewer than 1,000 files, then you're probably not going to run out and buy an enterprise search engine.  But thanks, in part, to regulatory compliance and updated amendments to how companies treat electronically stored information, enterprise search has moved from a nice-to-have technology to a must-have. A well-defined strategy can help companies to avoid choosing the wrong approach and ending up with a project that is abandoned not long after implementation.

Start With A Proof Of Concept
Enterprise search solutions can be impacted by countless variables, so starting with a POC (proof of concept) to test the solution in a company's actual operating environment is highly advisable. "By starting with a pilot or proof of concept, you learn a great deal about the search technology, your own infrastructure, the unknown challenges, and the ways in which your end users interact with search," says Haucke. "The lessons learned during the POC can then be fed back into the search solution and translated into the search strategy in other areas of your business." A company's intranet is one of the most applicable environments for testing how an enterprise search solution will perform. Regardless of whether a company has a generic intranet or is using a platform such as Microsoft SharePoint, the corporate intranet tends to be the largest source of centralized content.

While a POC may be the best way to test an enterprise search solution, Autonomy's Marshall advises there are pitfalls to avoid here as well. "A poorly formulated POC can actually hide inferior technology, confuse buyers, and raise the risk of making the wrong software choice," says Marshall. Here are some of the top pointers to keep in mind with regards to evaluating a POC. First, be sure to use information that has not been seen by the vendor, and be wary of vendor requests to retest or use different data sets or to process the data off-site. Second, be sure the search engine provides you with a complete set of results, not just the first results the engine locates. Third, ask the vendor to perform a search for a word that appears only on the last 10 pages of a document to ensure the engine doesn't run on the basis of only recalling items where the search criteria appears close to the beginning of the document.

 To gain a greater understanding of the concepts behind enterprise search, visit ISMinfo.com/jp/5724.

Overall, companies must remain cognizant that the enterprise search does not boil down to simple keyword search. And despite the multitude of tools on the market to address enterprise search, end users are still not well versed in how to select the right one or how to use it successfully. In fact, a recent report released by Delphi Research indicates that even with enterprise search tools in place, users are still spending at least the equivalent of one full workday searching for electronic information. Taking the time to develop an enterprise search strategy based on specific company needs before engaging search vendors and following through with a POC can help to ensure your employees don't experience this frustration.