How flexible does business intelligence technology need to be? Should it allow fully flexible ad-hoc data analysis, or does that overwhelm users? Are they perhaps happier with simpler, more prescriptive analytic paths? My answer is a resounding “It depends”.
On the one hand, it’s clear that some users really care about business intelligence flexibility. They don’t want the “right” dimensional hierarchy, carefully worked out in advance. They don’t even want fixed drilldown paths smartly calculated on the fly, ala’ Endeca (which, after all, ultimately didn’t succeed). Rather, they want to be able to truly choose aggregations and roll-ups for themselves.
Supporting this view is the rise of in-memory business intelligence. For example:
- SAP HANA is selling in impressive quantities.
- Further, HANA and alternatives are generating a lot of buzz. For example:
- Multiple clients have asked me for help positioning their products against HANA and Exalytics.
- Kognitio’s pretense to be HANA-like is getting them some sales too.
- QlikView has had considerable success.
But why would anybody pay up for the speed of in-memory BI? Analytic RDBMS offer blazing speed for broad ranges of queries. Parameterized reports let you do drilldowns in memory. So only if you need great flexibility do you need to keep a whole analytic data set permanently in RAM.
On the other hand, mobile BI is hot too, notwithstanding that its limited form factor detracts from analytic flexibility. And despite the problems with analytic applications, there’s clearly still an appetite for them. If nothing else, prebuilt data models, as offered by Teradata, are important to buyers, as I was recently reminded by a Teradata competitor sighing that it needed to develop some application data models of its own.
What actually triggered me to write this post was a chat with Bruce Armstrong, CEO of my client PivotLink. PivotLink is a SaaS (Software as a Service) BI stack vendor, now focused on the vertical market of retailing. In support of this focus, PivotLink is selling first and foremost a kind of retail-specific analytic application, because that’s what its customers want. Thus, PivotLink’s users don’t seem to care much about flexibility. Right?
Wrong! As I ran through cutting-edge BI features I thought PivotLink should or shouldn’t bother to implement, Bruce politely took notes. But when I raised the subject of making geographical aggregations customizable,* things got lively. Of course his users demand that! And they demand other kinds of flexible drilldown and dimensional hierarchies as well. Even for PivotLink users who want prebuilt analytic applications, BI flexibility is of high importance.
*Not all outlets in the same geographical area will have the same demographics.