I have spent the last 20 years of my professional life either implementing or leveraging Business Intelligence (BI) platforms in various industries. Over the last 5 years I have had several conversations regarding the relevance of BI and the promise of data science. While my experience would lead one to believe that I would be less than impartial when comparing BI to data science, recent advances in technology, think NoSQL solutions, have forced me to reconsider its relevance.
So, let’s start at the beginning, my beginning. It seems like only yesterday (late 90’s) that I was initially introduced to this exciting new technology. This was the brave new frontier of the data driven culture and represented the must see destination for any organization seeking to effectively leverage an ever increasing mountain of data. As you might expect, there was considerable hype around this burgeoning new industry. The promise that you could empower the entire business to develop their own analytics through a miraculous and easy to use interface. It did not take long after BI implementation to realize that there are relatively few people in a company, not currently indoctrinated in the data dark arts, that were willing to expend the time and energy necessary to understand the data well enough to build compelling analytic insight. Eventually, the hype passed and BI was appreciated for its true ability to quickly build and distribute easy to understand descriptive analytics that answer the “what” has happened questions.
Fast forward to 2012 and there appears an article in Harvard Business Review titled Big Data: The Management Revolution. Soon thereafter, C-Suite hearts were all a flutter with the notion of accelerated growth and new business development using armies of data scientists. These are highly trained, and expensive, resources capable of reaching into and wrangling disparate data sources to find patterns in the data that ideally lead to actionable outcomes. No longer did you need to know the question, the data could direct you to the next best opportunity. Data science practitioners scoff at the very notion of BI. Why would you want to spend your time focused on what has happened when you can direct your attention to what will happen.
As you might suspect, these two disciplines are related and definitely not mutually exclusive. Any CEO would want to possess the data science crystal ball that would make their organization more nimble, facile, and dare I say, omniscient. But that does not mean that you should not stay well grounded in the here and now. Once the results of advanced data analytics are leveraged to expand product offerings or optimize marketing efficiency, you still have to run the business and ensure that the newly formed strategies derived from data science are performing up to expectation. This is where BI reigns supreme for its ease of development, multi channel distribution, and intuitive out of the box data visualizations.
Most of the large BI providers offer a free version of their software that enables developers to explore
the usability of their products. A few noteworthy examples include Tableau,
these tools provide comprehensive data integration and visualization functionality, they
limit your ability to share and collaborate. There are also open source offerings including Birt,
JasperSoft, and Pentaho. It’s
important to bear in mind that these open source solutions are not as
intuitive and can require specialized resources to fully leverage the functionality.
BI solutions can be expensive to acquire and implement. The cost can vary widely depending on implementation size (number of developers and report users). There are opportunities to minimize the upfront expense by limiting deployment and as of late more are providing SaaS offerings eliminating the need for costly hardware. We at the MacLaurin Group have many years experience evaluating data warehousing and BI solutions in a variety of industries and recognize that there is no “one size fits all” solution.
BI providers have made it much easier to deploy their solutions across a variety of data sources including structured (i.e. data warehouse) and unstructured ( think key-value pair ) data. While variety may be the spice of life, it is important that the source for your BI data be a logical and accurate representation of the business which lands squarely in the realm of the data warehouse. Afterall, the BI solution will become the institutional source for the “truth” but only if it provides reliable and accurate information. This can be a challenge with less structured or non-curated data sources.
As the introspective CEO trying to determine if the organization should bother investing in BI, data
science or both, I would say that you would first need to assess whether or not you have the appropriate
level of automation to leverage the prognostications of those data science resources. As for BI,
data heavy companies will find a practical and affordable solution that will provide benefit even once
you do decide to invest in data science.
So, is BI relevant in the age of data science? If you think its’ important to have a comprehensive view into how your organization is performing, then most emphatically YES! Irrespective of your views on data science.
For more information regarding BI, feel free to listen to this podcast.