The Analytics Error Nearly Every Organization is Making

Blog /Analytics-Error-Nearly-Every-Organization-is-Making

Michael Lewis’s Moneyball is a fun book on analytics, as well as an enjoyable movie. If you’ve seen the film, you’ve seen Oakland A’s General Manager Billy Beane, played by a dashing Brad Pitt, sit in a closet sized office with his data analyst, played by a not quite as dashing Jonah Hill. Billy fires off creative, data-oriented questions, and Jonah lobs back snappy answers. Hollywood banter aside, these scenes show how analytics should be done—the ideal distance between the data needed and the decisionmaker seeking insight is about the width of a desk.

Unfortunately, that’s rarely the case in a Fortune 500 enterprise. Consider a brand manager who needs to know if her last promotion brought about the intended result before deciding whether or not to run another promotion. She has sales data from the POS records up to the most recent day, Nielsen data on her market share from last week, shipments data, inventory data, and marketing data on the promotion’s cost. Some combination of this information might provide the answer. She asks her analyst what can be learned.

He agrees the answer should be discoverable, starts muttering about data extracts and data mappings and data quality, and finally promises to get back with her after speaking to his shared services representative. Shared services schedules a workshop with IT. The IT folks want requirements. The term “self service BI” is tossed around a bit. The IT applications people explain that challenges as complex as promotion analysis require BI applications, and what’s needed is essentially a BI app store, where people like our brand manager can provide specifications for quickly created apps. Procurement gets involved, and an RFP goes out to vendors.

Weeks go by, then months, with our poor brand manager still making gut-level calls on new promotions.

The devil is in the distance. Our decisionmaker is separated from data-driven insight by several organizational layers, each filtering through different agendas and developing various interpretations of what she needs.

Instead, our analytics solutions should start with the user experience, the brand manager’s experience, and work backward to the technology. Technology must  focus on eliminating distance, eliminating delays, inventing only what her questions require, and bringing data science right to her office. The result is a data analyst’s response to each query almost as fast as the brand manager can think them up.

This does not have to look like Jonah Hill with a laptop, thankfully. Many variants are possible. But the experience must be responsive and empowering. Answers to new questions should rarely (or never) be on the other end of multi-month projects.

Sure, a baseball club is different from a large products company, but every enterprise properly focused on user experience will find a way to close the distance between decisionmaker and data science, and make their analytics play like Moneyball.  

Fecha de publicación: 15/10/2015