What Elements Should Be Included in a Business Intelligence Solution?

With the flood of Business Intelligence software showcased in the last few years, coupled with budget tightening and procurement freezes, the question of what elements should be included in a Business Intelligence Solution becomes more paramount.

BI systems have been defined as being able to do everything from centralizing all of your corporate data from all of your various applications through robust (although nearly always proprietary) ETL tools to being able to provide ad hoc, on demand report developing capabilities for not only business analysts, but also for corporate executives and managers, who may not be a technically savvy as your average 9th grade high school student.

But what does a Business Intelligence tool really need to include? Well, for starters, I'd recommend looking for operating system and database independence and scalability. Sure, you're business might be using Oracle today, but when your CTO decides to cut a great deal with MS SQL Server for a much lower price, you're going to want a tool that can make that transition. And believe me, that happens more than you might think.

I'd make sure your BI tool candidate has all of the analytical tools you'll need. OLAP and MOLAP capability with the ability to develop reports against MOLAP, dimensional model, or relational models architectures are critical. Make sure you have the ability to easily develop reports that use very complex SQL aggregations, correlating sub-queries, and unions. I mean, hey, if we wanted to write SQL, we would have just called a programmer, right?

Next I'd look for something with a web based design interface, with all the bells and whistles, including GIS connectivity (or at least something that can use one of the existing GIS vendors). The days of installing client applications for design tools are past us.

Your might think that ETL would be a major requirement in this decision, but honestly, this isn't as important anymore. Most companies are using an ELT (that's not a typo) solution now where you Electronically Load, and then Transform your data. That way you can use very simplistic data replication tools, and let your database engine do all the transformation rather than doing it on the way to the databases.