Many people reach out to me asking for advice in helping them evaluate various BI platforms for their companies. In my mind I am obviously the wrong guy to ask as I am not shy to admit my fairly heavy bias towards Microsoft. However, over the course of the last twenty years, I have been exposed to just about every major BI player on the market and I think that my bias is well justified as it is actually based on a very pragmatic and rational set of criteria.
Having said that, I find it very interesting, however, that a lot of companies go about selecting their Business Intelligence platform in a very non-strategic, irrational and tactical way. I, therefore, decided to compile a quick list of things that I think would be important to consider as these evaluations take place.
The changes that are taking place on the hardware front are very important. Majority of the top enterprise BI vendors will say that some sort of a BI appliance (Hana, Exadata, and PDW just to name a few) is absolutely necessary to catch up with the ever increasing data volumes. I happen to think that as a general rule this argument is completely ridiculous. As a rule of thumb, hardware is commoditized on a daily basis. A $20k server today will likely have 32+ cores and up to a TB of RAM. Few years ago, similar hardware would require a $1M investment. Today I routinely load hundreds of millions of rows of data in a PowerPivot model that runs on my laptop allowing me to do the kind of analysis that required something like Teradata + Microstrategy just a few years ago. My recommendation would be to exhaust the commodity hardware options before an appliance is even considered.
Self Service Vs. Enterprise
Much of my time is spent fighting philosophical battles with various IT organizations around this topic. IT wants to be in control and own the content creation (they want to build all dashboards and reports themselves). At the same time, I have yet to see an IT shop that did not have a huge backlog of reports and an upset customer (business) who is starving for data and reporting tools. I therefore recommend reframing the role of IT in the business intelligence strategy of a company this way: IT will own data, security, governance and plumbing required to wire up all the pieces of the BI architecture, business will own all (well, most…) of the content creation. IT would be required to have someone on staff who can assist business customers if they get stuck but whatever BI tool an organization is looking at, it should be self-service focused first and foremost.
Over the last few years the BI landscape has changed very significantly. One of the major changes was the movement away from the “best in breed” and towards the “platform buy”. Generally speaking, this makes sense. A platform buy will sacrifice on some features and usability for the potential gains in interoperability. In reality, however, out of the four big vendors, IBM, Oracle, SAP and Microsoft, only one can honestly say that all of its offerings are wired together natively.
Bill of materials
Many people don’t think about this aspect of the decision, but I think it is very important to consider how many different products and installs will have to be stood up in order to get the BI platform implemented. All but one vendor will typically have a separate tool (or most likely several different tools that came from different acquisitions) for different aspects of BI. So an ETL, Dashboarding, Reporting, Analysis, Master Data, Data Quality, Predictive Analysis, Data Warehousing, Semantic Layer, Self Service tools, Data Exploration, Drill-though, Report bursting, Knowledge management, Social, and many other elements of business intelligence platform will often require a different infrastructure investment, install management and most importantly a separate license to obtain.
This is obviously not a complete list of things to consider for someone who is trying to evaluate a BI platform but I think it is a good start. When I am involved in an evaluation process like this, my number one goal is to spend as much of the budget $$$ as possible on content – dashboards, reports, analytical views – something that my end users will get to use on a day-to-day basis as opposed to adding more hardware to the data centers or software on a balance sheet.