I started my career as an analyst. I am trained in what is generally referred to as “applied analytics,“ the idea that you use statistical methods to control for error in less-than-ideal data. Besides being a hoot at dinner parties, this skill has brought me incredible job stability and satisfaction.
When I started, I had never heard of business intelligence (BI). I was a research coordinator at a hospital and my job was to do anything with numbers that didn’t include the finances. I did everything from creating a method of performance evaluation for administrative staff to analyzing the efficacy of care. Eventually we automated our outcomes protocol and all of that data got dropped into flat files. Suddenly I found myself modeling data to create data marts to support reporting. I was officially a BI professional.
For years, BI was the Holy Grail. But one day, someone starting talking about analytics (of course this is an over-simplification, but bear with me). Then, analytics, data mining, and predictive models became the panacea for organizations, because we are all looking for more efficient ways to understand our data and to increase efficiency and competitive differentiation.
The Difference between BI and Analytics
So, what is the difference between BI and analytics? Is there a difference? How can we ensure that tangible value is a result of either effort?
I believe there is a significant difference. A number of years ago, a boss boiled it down for me in the simplest of ways: “Analytics requires someone’s gray matter.” I started thinking about that and then I realized the fundamental difference between BI and analytics.
In BI, most of my focus is on reducing the maintenance and non-value-added activities so organizations can use their data. That means that I reduce the effort of the analysts on what I refer to as “pedestrian tasks.” Pedestrian tasks are the things that anyone walking down the street could do: pull in data, filter data. Yet most organizations have their analysts doing this work. In an assessment I did in an organization last year, the analyst told me that 80 percent of his job was non-value-added activities. He had a PhD. Most of us would recognize that this is not a good use of resources.
BI is an enablement mechanism for analytics. BI can’t do analytics, and analytics is not BI. They are fundamentally different, with different goals and different teams.
It is true that you don’t necessarily need BI to do analytics (as noted in my first job). Plenty of companies have found great value in their data with nothing more than a team of really smart analysts. They find the data, clean the data and analyze the data. Scalability becomes an issue of course, as does data consistency. For example, this method could lead a health care organization to a situation where four executives have four different numbers for things like member counts or hospital bed days.
The Importance of Experimentation
The biggest critique I hear about building a BI program to enable analytics is that often times BI programs actually get in the way of analytics. A perfect example: focusing on reports rather than analysis provides barriers to analysts. The structure of the data, the frequency of the data and how the data are used are all very different between reports and analysis. Your best bet to make sure that your BI program enables analytics is to build an analytic sandbox. This is an area where your analysts have full control, from what data resides in there, to how long they keep that data.
The important thing is to make sure that when your analysts build something that could benefit a large group of people, you have a process to ensure that it gets brought back into the BI ecosystem to be used more broadly. That means you have to make sure that it follows your standards for data quality and governance. This method of a separate analytic sandbox allows you to take advantage of all that gray matter but still control the quality of the data that ends up in your enterprise data warehouse.
Not everyone is an analyst. As a matter of fact, very few of us have that very specific skill set. Yet our organizations need better information to make informed decisions. To meet the needs of the all of your users, you will have to create a BI program that can deliver everything from dashboards to reports to analytic enablement. Analytics are here to stay, but so is BI.
Laura Madsen, the leader for the healthcare practice at Lancet Software, a BI consulting firm in Minneapolis, is the author of Healthcare Business Intelligence: A Guide to Empowering Successful Data Reporting and Analytics.