What Makes Search So Tough?

By James Watson posted 12-16-2010 10:48


As we look toward 2011, one of the top projects on many of our clients’ agendas is enterprise search. Candidly, search capabilities have been available for many years, and nearly every organization we consult for has one or more systems in place for search. So why the renewed interest?

Well, for the most part, users are not happy with their existing search capabilities. Top of their complaint list: having to execute a search in multiple systems when they are not exactly sure where the content they’re looking for may reside. Looking for last year’s budget? First, check email, then the network drive, then SharePoint, etc. Then give up and call someone you know is a pack-rat and well organized and ask them for the file. 

When we frame the problem for customers, we break “search” into three critical elements: User Experience, Information Organization, and Technology. 

User Experience: Simplicity is the key, but keep in mind the user’s role. For example, a call center representative vs. an IT project manager. Call center staff search in fairly consistent patterns; the top 20 call types represent the majority of their volume. So why not provide saved searches with frequently asked questions that they can select from? By comparison, project management personnel are likely to be involved in a wide variety of initiatives, so it would be much tougher to pre-populate or anticipate any search patterns. Thus a thorough initial understanding of the requirements and types of users is key to the user experience.

Information Organization: Indices, tags, and metadata are critical to improving the efficiency of any search mechanism.  But what if your content isn’t indexed? A couple of options: 1) Go back and index all of the content (good luck), 2) index just the critical stuff, or 3) start today and begin capturing metadata on new content (day forward). The latter is the most practical and cost-effective approach. With any option, you will need to decide upon an indexing structure; what is the minimum amount of index data you can get by with? How can you automate the capture of metadata so users aren’t burdened with the task? 

Technology: Not surprising, the current search technology in the market is extremely mature (far more mature than the other elements). We’ve all encountered very powerful search algorithms, such that we find ourselves saying, “How did it figure out that?” Yet the effectiveness of most search tools is limited without metadata and tags to work with. So if you have 10 years’ worth of content sitting on shared drives, and only cryptic file names, dates, and author names to work with, your ability to leverage the technology will be limited to full-text indexing. Full-text indexing is better than nothing, but it tends to return search results cluttered with garbage – which in turn leads to a poor user experience and often abandonment (heck, just pick up the phone).

Given this brief review of the critical elements, it’s important to recognize that all three components must be addressed to be successful. Just focusing on the technology won’t get you very far. Nor will asking users to go back and index content so it can be leveraged by the search engine. Best-in-class organizations distill out the interdependencies and build a pragmatic roadmap to address their enterprise search needs.

#InformationGovernance #enterprisesearch #userexperience #ElectronicRecordsManagement