Data-driven is one of those flashy terms that corporations like to throw around during press conferences. You hear statements such as, “Our new data-driven culture will lead to a 12% increase in net profit.” When sentences like that get thrown around, one has to wonder if the speaker is the victim of bad data. While using big data to support business decisions is the new norm, and a survival requirement for doing business online, the way the term data-driven is used misses the boat in many ways.
For one, it implies that all decisions made before the age of big data were not driven by quantitative or qualitative information. This is obviously and patently untrue. Good businesses have always used data as a tool. More importantly, the term imbues data itself with mystical powers to transform a mediocre or failing business into a success. This misunderstands the nature of data, the limits of data, the reasons businesses succeed and the inevitability of bad data.
Even if you’re doing everything right, asking the important questions, focusing in on the relevant areas, none of this guarantees that the analytics are going to point in the right direction. The results are only as sound as the data you analyze. While most businesses assume that there will outlier data that needs to be ignored, fewer ever stop to wonder if there might be a more systemic problem with their data. Is there a flaw in how the data is collected? Is there information that we don’t have that might make this result look more or less cheery?
Take this as a hypothetical. A small software company gets a sudden influx of orders from New Zealand, where they’ve never sold anything before. The payments all clear and the software company executives think they’ve somehow opened up a new market. They begin marketing their software widely in Auckland, where most of the orders came from. Seems straightforward, right?
What the software company doesn’t realize is that all those orders came from employees of Company Q. For internal reasons, Company Q had employees purchase the software and reimbursed them for the costs afterward. What looked like dozens of discrete customers was really nothing more than a bulk license order from one customer. It was also an order with little chance of being repeated. The software company is likely wasting its marketing budget due to bad data, or at least incomplete data.
If the software company had been aware of the context, or asked the right questions during the purchasing process, it would have viewed the sales opportunities in New Zealand much less optimistically. Data-driven decisions can be good decisions, but they depend on the quality and completeness of the data itself. Assuming that becoming “data-driven” cures all problems is shortsighted and unrealistic.