Blogs

Again and again, clients engage us to help them develop a strategy for their enterprise search requirements. When we sit down to talk with their users – people who regularly run Google and Bing searches out on the web – the question never fails to come up: What makes enterprise search so difficult, when it’s so simple on the public Internet? After all, we’re talking about searching the information assets of just a single organization!

So why is search so difficult for corporate entities? Three reasons: 1) the lack of historical data to help tune performance; 2) how and where information assets tend to be stored within an organization; and 3) the complexity of many of the usage scenarios for enterprise search.

First, the advanced search capabilities on the public Internet are actually quite complex. Billions of dollars have been invested to give all of us the impression that it is simple. In addition, we’ve collectively been using several common search utilities for a number of years now, so there’s a rich history from which to mine statistics on performance and accuracy, which can then be used to tune the search utilities. In the corporate setting, little, if any, historical data is available.

Second, certain characteristics of the information assets within an enterprise are guaranteed to dramatically increase the scope of search requirements. Web pages or documents on shared drives are the easy part. Those pesky document management systems are much more difficult, as each has its own access controls. It’s common for one of our clients to list out 15-plus different systems where they have content that they would like to be able to search: e.g. a system with contracts, a system with reports, a system with images, a system with proposals, a system with internal policies, etc., etc. Each has different meta data; seldom has common terminology been used across these different systems (customer vs. client, employee vs. associate). Worst of all, secure or confidential content is mixed in with generally viewable content, with a proprietary access control database housing the logic.

Third, the usage scenarios for enterprise search tend to be more complex, often mixing structured data needs with unstructured content. For example, a sales person looks up a client via a database inquiry by client number, then needs to see all of the proposals issued over the last 3 years. Pretty simple, right? But what if the proposals span three different geographies (U.S., Europe, and Asia), and the client names used are the operating divisions within each geography? Maybe, just maybe, the customer numbers have been used consistently throughout the period that’s of interest, but more likely it’s just been done uniformly across all geos within the past year, after the organization completed a master data management project.

For all these reasons, search capabilities for corporations will continue to be complex.

The key takeaway: If you’re embarking on a program to implement enterprise search, do some prioritization. Do you want to focus on the low-hanging fruit, such as documents stored on network drives? Or do you want to target frequently referenced material in a particular repository? Often the best way to begin is by conducting an extensive survey of your users, asking them what they search for, the frequency, and how HOW MUCH TIME THEY SPEND SEARCHING (the latter numbers will be critical when you get ready to make the business case for your enterprise search initiative). Your findings will help you target the high-value use cases with a pilot and then build upon that experience.