Trolling through Google recently, I ran across an old column in E-Commerce Times that asked, "Search-Based Applications: Smoke and Mirrors or Real Innovation?" What struck me most was how relevant that question still is today, when we are looking for information within an ever-widening array of technologies, from mobile devices to the cloud. Often we are not quite sure where search ought to live, begin or end.
As is so often the case, part of the issue is vocabulary -- namely, just what is an SBA anyway? I dislike arguing semantics because it can take time from the real issue: getting maximum value from technology. But, it's important because how you approach it has significant ramifications on how you build it.
More than just presenting results
SearchContentManagement defines a search-based application as software that can "query disparate structured, semi-structured, and unstructured content sources and return the results of the query in a single, unified view." This is accurate and powerful because it encompasses all types of content, but I prefer to focus on how all that content is indexed rather than how it is presented.
Simply overlaying a macro search engine on top of multiple repositories fails to account for the different security or access schemes that are at work in each one (i.e., your permissions might be set at one level for documents and another for CRM records, so you may not even have access to some of the information you think you do).
Further, this approach won't bring together the different vocabularies that likely are at work: A search on "ABC Corp." might register a hit in one system but not in another, in which it is listed as "A.B.C. Corp." or "ABC Corporation." So, you may be presented with a list of search results that are incomplete in some way.
I never 'metadata' I didn't like
The preceding reasons are why "It's OK, we'll just do a Google search of everything we have" is the wrong answer.
The right answer concerns the way you approach your metadata, which is the umbrella term for those "magic" descriptors you apply to each piece of content. Effective metadata has a lot to do with the efficacy of the thesaurus you install to reconcile the tags that don’t quite line up from one system to another, so that even a close-but-no-cigar search returns the results you need.
Metadata is the fuel that makes search go, and I daresay, its sophisticated management is at the root of the innovation the E-Commerce Times article described. One big question you have to answer is whether you want to embed it directly into your ECM solution or add it as an overlay -- either can work, but there are business processes, technical architecture, and cost factors to consider, and here it's especially crucial to avoid being sucked into the inevitable discussions about semantics.
Isn't every application search-based?
Call it a search-based application or a rutabaga, your goal remains the same: To expose as much enterprise information as possible to as many people -- and systems -- as you're able, accommodating and unifying your security and access policies to the greatest degree you can. The challenge is that there is no single thing known as "information." Rather, there are documents, databases, images and tweets and a whole range of other items to contend with.
Many of today's search utilities can span multiple siloes, automatically tag incoming information, permit the tagging of content snippets from existing material or online, or track and synthesize more free-form, crowdsourced types of input. These kinds of capabilities are what give offerings like HP Autonomy their muscle and an SBA its greatest traction -- and they’re also why I contend that pretty much every (non-desktop) application is an SBA, for I’m hard-pressed to think of one that doesn’t involve looking something up.
Enterprise collaboration tools hindering collaboration?
Building the case for enterprise search
Getting started with enterprise search