A few weeks ago, Tricension attended a “State of Data Science in KC” meetup event. The event drew a large crowd of executives, data scientists, and aspiring data scientists from the community, and did an excellent job of covering the Kansas City job market and detailing the work being done around town. Early in the evening, an audience member asked what I feel is an incredibly important question – “What is the most challenging part of a data science project?” The general consensus amongst the panel was that actually defining the problem/business challenge can be especially difficult for businesses as they plan for a project. While the panelists emphasized that a well-defined problem is the key to the success of a data science project, answers on how to define a problem, get executive buy-in, and set project expectations often go un-discussed within the data science community. This is concerning, given that the clear definition of a business challenge is so absolutely vital to a project’s success. Although each business case is specific and nuanced, Tricension has found a reliable process to put data science in context and set a project up for success. We wanted to share that with you today.
Prepare the Your Data Foundation Be honest about where you’re at within the data hierarchy of needs. Before you can hire a data scientist to start building models and using cutting-edge algorithms on your data, you need the data to be available and manageable. You should already be using data to measure your performance on key metrics. Why get into the business of making predictions about the future when you don’t even understand your business’ recent past? Only after putting reliable data management systems in place, will you see significant ROI from hiring a data scientist. Without this kind of support, a data scientist will struggle to succeed or be forced to do work better handled by a systems architect or business analyst.
Relate Your Business Challenges to Money and Value When trying to discover business challenges, I go straight to the money and the value. How does your company make money and what stops you from earning more? A frank discussion centered on the business model and the sales cycle often uncovers business challenges with a significant ROI. Second, what value do you provide to customers and how can you deliver even more value? What blocks you from providing value? By exploring these questions, we can help your business leverage data science to become more valuable to current and future customers. Lastly, we also seek to automate or eliminate processes that don’t add value for you or your customers.
Don’t Be Afraid To Experiment Start small if data science is new for you and your company, but be rigorous in your evaluation of the work being done. If you want to get started and see how data science can impact your business, limit the work to a department or functional area with well-defined business challenges, a data-driven mindset, and the ability to store, search, and measure their current data. Treat data science as an experiment with a well-defined scope and target. With success, you can expand your efforts into other areas of the business. If the experiment isn’t successful, you can cut your losses or make strategic adjustments. Data scientists are inclined to view their work as experiments because of their empirical nature.
I’d like to give a big thank you to the organizers of this event. Everyone that I spoke with was thrilled to learn about the great work being done in Kansas City, including big wins such as improving outcomes for mental health patients and identifying customer segments from anonymized data, to name just a few. I can’t wait to take part in more events like it in the future. At Tricension, we are here to help you manage your IT needs and help you turn your data into actionable ways to streamline and improve your business. If you’d like to explore the value of data science for your company, please give Eric Levy a call at (816)336-1050 or by email firstname.lastname@example.org to set up a complimentary data science assessment.