Search—It’s All About The User

In 1992, James Carville coined the phrase, “It’s the economy, stupid”—a saying which became a cornerstone of Bill Clinton’s successful presidential campaign. Mr. Carville correctly identified that, among the many factors at play, concerns about the economy trumped all other issues of the day.

Often, in the enterprise search world, industry pundits and certainly the search vendors (Google included) talk more about many factors other than the most important one: the user. There is, of course, relevance of results, which we certainly believe is very important, and many other factors such as comprehensiveness, security, ease of administration and total cost of ownership. All of these factors matter to enterprises, just as, say, healthcare and foreign policy mattered to economy-watchers in 1992.

But at the end of the day, users know best. Increased usage trumps all other factors, because it presents real evidence of the value of search, taking search relevance, usability, and comprehensiveness into account. Business users will not use a system if it doesn’t successfully search the important company repositories and return the information they need.

For this very reason, search vendors such as Google are most successful when they keep the user in mind. Here are some examples:

Social search features:
As smart as search systems might be, they can be much smarter by tapping the collective intelligence of users across the enterprise. For this reason, we launched our social search feature, “user-added results.” User-added results lets users across the organization recommend links for any given search queries that appear at the top of search results. Every user tends to be an expert of a given area; with this feature, they can now share their expertise easily across the organization.

For example, at Northern Trust, a global investments and banking firm, many of their 11,000 employees needed to access business forms and other documents on a frequent basis, which were housed in intranet pages, Web servers, a content management system and in Lotus Notes databases. But often users would use common names or acronyms, rather than “official” ones, for the documents they needed. “We need to share niche employee expertise about terms, acronyms and financial instruments and bubble it up to the rest of the organization,” explains Marty Byro, manager, Enterprise Partner Experience at Northern Trust. “User-added results lets trusted experts share their knowledge by recommending keyword matches and promoting pages.” For instance, when an employee now searches for “UDC,” the first result links to the latest version of the “unified distribution checklist,” thanks to a link uploaded by a fellow employee.

Searching content management systems:
As content management systems have grown in prevalence, users across enterprises are looking for easier—and more relevant—means of searching content inside them. This is no surprise, since most content management systems focus much more heavily on the upload of information and the workflow, as opposed to the findability and download of information. Consequently, many organizations have turned to enterprise search systems as a means of increasing return on their investment in content management systems.

The goal of increasing ROI drove Mercer, a global consulting firm that had heavily invested in content management systems, to turn to search. Their core repository, Mercer Link, had nearly 2 million documents, and was vital to the productivity of the 19,000 employees distributed all over the world. The firm had several criteria in mind when selecting a solution, including fast, accurate search of multiple content sources, single sign-on for users, ease of use, cost-effectiveness, security, and scalability.

Ultimately, by focusing on users, Mercer is seeing business rewards. According to Haroon Suleman, Mercer’s lead enterprise search architect, “Even if our 19,000 employees save just a small fraction of time searching for information, then our increase in productivity pays off the search solution very quickly.”

Self-improving relevance:
For users, nothing is more important than search relevance. Numerous studies show that not only do most users not click to the second page of search results, but that many users don’t even scroll down the first page. When search precision studies are done, the first result is given a score of 1.0, the second result 0.5, the third result 0.33, and so on. This means that the 10th result would get a mere 0.1 in terms of precision.

The goal of increasing relevance led to one of the most recent enterprise search features: “self-learning scorer,” an algorithm which guides search results to automatically improve over time. The self-learning scorer analyzes user clicks and behavior and uses this information as an additional input to rank the relevance of search results.

For instance, if most users in the enterprise click on the fourth result for “product roadmap,” the search appliance understands that this is probably more relevant than the top three results, and therefore pushes the fourth one up.

To further ensure that users get the most relevant results, they can use a free, downloadable tool called “side-by-side.” This tool allows a search administrator to test any two search configurations and view the search results side-by-side. The search administrator can choose any number of queries, and let users across the organization run each query and view search results across both configurations. They can then vote on the one they deem most relevant.

According to Sébastien Oliver, project manager at BP, “Typically, people type in one or two keywords at most and these are usually not very precise. What they expect the search engine to do is read their minds to find out what they are looking for—if it does not deliver, they quickly stop using the tool.”

Universal login:
A key requirement of any enterprise search solution is security: it should only show results that the user is authorized to see. But in increasingly complex enterprise environments with multiple systems and multiple logins, getting to secure search results in a user-friendly manner can be challenging.

“Universal login” dynamically generates a single login page for each user, and securely passes the login information to back-end systems, even if using heterogeneous authentication protocols. Administrators can further define credential groups to support multiple credentials across multiple systems, minimizing the number of logins required for the user.

Distributed search:
In some cases, users might be spread apart geographically, along with the data. In these cases, it’s important for users to be able to view data in their local region, but also easily be able to access data from other regions as well. Such was the case with MTCSC, a defense contractor specializing in integrated logistics support, engineering and enterprise solution development. One of MTCSC’s most valued customers was faced with reducing or eliminating the threat of improvised explosive devices (IEDs). IED relevant data consisted of unstructured documents, structured relational data sets and data sets which contain detailed geospatial relevant information. The customer previously had no adequate way to provide a single user interface that could seamlessly search these various IED related data holdings within the Department of Defense intelligence community.

MTCSC deployed a worldwide solution, which currently crawls 2,500+ data sources in support of five combatant commands. This solution provided an immediate capability to search websites, file shares, relational database platforms and content management systems such as Microsoft Office SharePoint Server (MOSS), and provided a unified view of relevant information. Currently, the enterprise solution indexes in excess of 40+ million documents worldwide.

Integration has also been performed with Google Earth Enterprise and Google Maps to provide a geospatial representation of information. The integration provides a unique way to view the intelligence information that is more meaningful for end users. These capabilities ensure the overall mission to deliver relevant information to a counteractive measure to the armed forces as rapidly as possible with efficiency and success.

As can be seen in the above example, while data had been scattered geographically and in separate silos, user-centric search tied with geospatial visualization enables users to view results from multiple search appliances spread across geographically, improving response time and effectiveness.