If you think that semantics is a huge problem for data warehousing initiatives—and it is—it’s an even bigger problem for our industry at large.
Take the word analytics, for example. It’s a popular term right now: reporting is passé; everyone wants to do “analytics.” And like most popular terms, it’s been completely bastardized. This is largely because vendors want to exploit any term that is popular and bend it to match their current or future product portfolios. But it’s also because we’re too complacent, uninformed, or busy to quibble with exact meanings—that is, until we have to plunk down cold, hard cash or risk our reputation on a new product. Then we care about semantics.
Most vendors use the term analytics to describe what I would call “interactive reporting.” Evidently, any tool that lets users filter data on page or drill down to greater detail qualifies as “analytics.” This isn’t wholly inaccurate since Webster’s dictionary defines analysis (the root of analytics) as: “Studying the nature of something or determining its essential features and their relations.”
“We should use labels to clarify not obscure meaning.”
But let’s get real. We should use labels to clarify not obscure meaning. Reporting tools have given users the ability to filter, sort, rank, and drill for years, and yet we never called them analytical tools. Instead, we’ve used the terms ad hoc query, end user reporting, dashboards, scorecards, and OLAP. To borrow a biblical phrase, let’s not put old wine into new wineskins. Otherwise, we simply spoil something good.
“Analytics is easy to say but hard to do.”
Let’s be clear: analytics is a natural progression of BI from reporting on historical activity to applying statistical techniques and algorithms to identify hidden patterns in large volumes of data. We can use these patterns and relationships to create models that help us understand customer and market dynamics and predict future events and behavior. Thus, analytics is a marked departure from traditional reporting in terms of scope, depth, and impact. It is easy to say but hard to do. This is why so many vendors have jumped on the analytics bandwagon without sporting any real credentials.
One of the few vendors that has marked a steady course in the usage of analytics is SAS. I talked with Keith Collins, SAS’ chief technology officer, last week about this issue, and all he could do is shake his head and commiserate about the bastardization of a term that is near and dear to his company’s heart. On the bright side, there could be worse things than for neophyte vendors to co-opt your semantics. Even if their current products don’t support honest-to-goodness analytics, their product roadmaps do.
In the big scheme of things, the bastardization of analytics is actually a bridge to the future. We use words to propel our thinking ahead as we prepare our bodies to follow. So, give it five or ten years and analytics will have a distinct meaning from reporting. By that time, I’m sure we’ll be bastardizing some new BI term.
Posted by Wayne Eckerson on September 10, 2009