What are you hoping to accomplish with an analytics and/or data science implementation? Can you define it both qualitatively and quantitatively? In simple terms, with the analytics piece, you’re probably looking for answers to questions you have, and with the data science piece, you’re probably looking for new insights and questions you didn’t even know you had. Abstract that, though, and what you’re really looking for is change: a change in the amount and quality of information you have about your business or your customers, and a change in the depth and breadth of insights your information systems provide.
But what drives change? Think back to your algebra days. All changes are really just functions—inputs determining outputs: X Δ Y. You have a goal—Y—and you have to understand the variables to plug in to reach that goal—X. The X will determine the Y. The input will determine the outcome. It’s really that simple. What’s not so simple, though, is determining the X part. What X will drive the best Y? As in algebra, the Y is totally dependent on the X, i.e., reaching your goal is totally dependent on how you choose to get there.
Define your goals clearly. For example, you could define a goal as, “We would like to have better customer information to create more personalized offers, at the most opportune times, to increase conversion rates and drive revenue increases across product lines.
Once you’ve defined your goals, the next step is to figure out what changes in your information will drive the biggest changes in your business and deliver the greatest benefit. For our example above, the change might be to implement a rapid-start analytics initiative to gain quicker, deeper insight into customer behavior across channels. In essence, you’re always thinking about the biggest Xs that will drive the biggest Ys. You’re always asking, “Which inputs can I use to drive the best outcomes for my business?”
Once you’ve identified your X Δ Y, you’ll need to get the right tools and architecture to make your plans a reality. There are a few must-haves. First, make sure the tools match your data needs. Data mining technology will give you deep insights into the business and will also help you harvest that low hanging fruit. Statistical analysis tools will help you perform sensitivity analysis and uncover hidden relationships in data. Machine learning technology will help you predict events and automate decisions.
Also, make sure the technology can grow with you so that you can ensure that you can get the answers you need, when you need it, both now and in the future. Whether your platform is on-premise, in the cloud, or a hybrid environment, it needs to fit your business needs, not the sales quota of your vendor.
Finally, embrace change. Shift your paradigm to build a data culture. If you’ve built good systems, you can trust your numbers. Do that. Gut instinct is fine, but it needs to be backed by hard numbers. Once you’re up and running, you’ll start receiving answers to questions you didn’t know you had. Use that information to effect change and meet your goals.
Operationalize analytics. Make analytics a repeatable process—complete with good data governance and self-service analytics capabilities—that enables you to solve your problems more quickly. You can build analytics models, one at a time, as you go along, but the better way is to find a technology vendor whose products come with a complete set of tools with pre-built models you can customize.
With operationalized analytics, you’re able to realize one of the biggest benefits of an analytics culture: you’ll be able to find those Xs that help you reach your Ys more quickly than you could before—and more quickly than your competition. With the ability to develop faster, deeper insights and act on those insights more quickly, you can reduce your opportunity cost and show a clear ROI—and get a leg up on the competition.