How exactly do the areas of data and predictive analyses differ from business intelligence and what should IT leaders be doing differently?
As we all know by now, the Big Data bus is rolling. Now is the time to get a seat on that bus if you want to help lead change in your company.
Big Data is not the same thing as business intelligence. But how exactly do the areas of data and predictive analyses differ from business intelligence and what should IT leaders be doing differently?
I had the opportunity to talk to Joe DeCosmo, Director of advanced analytics and insights in West Monroe Partners' Technology Solutions and Enablement practice, about predictive analysis. Here is the result of that interview.Question: For those companies that are just considering the area of Big Data, can you explain the difference between the traditional concept of business intelligence and the newer concept of predictive analysis, and why now is the time to go for it? Answer: Business intelligence has traditionally been about reporting on historical trends and current business performance. The best BI applications provide easy access to business users to allow them to quickly gain business insights.
Predictive analytics goes beyond these insights to creating business impact. Predictive models and optimization algorithms allow you to not only predict the most likely outcome, but what's the best that could happen. Predictive analytics lets you standardize and automate this process, so decisions are fact-based instead of judgment or interpretation.
With the volume and variety of data we have today, this shift from insight to impact is more important than ever. The most effective data-driven organizations today have honed their ability to quickly mine and model all of this data to find the most meaningful patterns or combinations of data to predict the next best action or outcome.Question: What are the first steps a company needs to take to get serious about business analytics? Answer: As Ken Rudin, Director of Analytics for Facebook, says, "find a needle, move the needle." In other words, rather than trying to build an entire business analytics practice at once, start small. Work with your business users to find a current business problem where you believe you can have a measurable and meaningful impact. Define your key metrics, gather the appropriate data, and then iterate through the analytics quickly to find predictive patterns. Involve your business users throughout the process, so the analytics aren't a mystery, and concentrate on building a simple, useful model that can be deployed quickly.
Throughout this process keep in mind that there is no such thing as a perfect predictive model, so instead of trying to build one, concentrate on finding a model that's good enough to "move the needle."
This deliberate, start small approach will increase the odds of success and user adoption, and start to lay the groundwork for building a culture of analytics-driven decision-making.Question: It looks like IT is the most logical department to take on predictive/business analysis duties. Do you agree? Answer: For the most part, I think that's true. The best department to take on analytics is where there is the ability to collect, cleanse, manage, and analyze an organization's data. IT traditionally has owned the first three activities, but analysts might be anywhere in the organization. It's not uncommon to find folks with analytics skills embedded in finance, marketing, even sales or HR. This means that the users of the data are separated from the producers of the data. Obviously, this isn't an ideal situation. Ideally, these groups would come together in a new group under an umbrella of "data science" or customer intelligence. Question: What skills should IT pros look to develop or sharpen in order to excel in that area-other than the ability to work closely with business executives on what the business goals they want to achieve? Answer: You've hit on the first part of my answer related to developing business acumen. This is critical to be able to relate to your business users. More on that later!
Otherwise, you must have a solid understanding of multi-variate statistics and quantitative analysis. And don't overlook presentation and visualization skills, so that you can tell meaningful, compelling stories with your data.Question: You mention the three Rs of predictive analysis. Can you explain those? Answer: Sure, the three R's are Reliable, Repeatable, and Relateable.
Reliable refers to the accuracy of your predictive model. As I said above, the model doesn't have to be perfect, but needs to be accurate enough to have a business impact. The art of analytics is knowing when a model is "good enough", so you don't burn cycles searching for the perfect model.
Repeatable refers to repeatable results and repeatable process. Models need to be able to replicate their results across customers, time periods, and markets to be most useful to the business. Poorly built models may look good when they're built, but will be susceptible to small changes in data and won't hold up over repeated applications and measures.
Repeatable process means building a predictive analytics framework that can be applied to different business problems. At a high level, we use a Define - Diagnose - Predict framework where you define your objectives through the right data and KPI's, mine the data to explore and diagnose the problem, and then build a predictive model that can be deployed going forward. Such a process turns analytics from a one-off project to a meaningful, ongoing business function.
Finally, predictive analytics must be relatable to the business users. Simply put, this means being able to present and explain analytics in business NOT statistical terms. This is critical to convincing your users to trust and believe in your models enough to put them into practice. Unfortunately, while this might be the most important of the three R's, it's often the least appreciated when it comes to analytics. It's not taught in school, and it really only comes with experience. The bottom line is that no matter how good a model is, if the user can't relate to it, the model will be either ignored or underutilized.
By concentrating on these three R's, you'll be well on your way to building an effective, efficient and successful analytics practice.