Competing With Data & Analytics
What to Read Next
Already a member?Sign in
This headline tested well and generated lots of clicks. Ergo, all readers must be happy. Right?
The improvements that analytics bring to a process can be downright addictive. Collect data; scrutinize model; refine process. Lather, rinse, repeat. With each iteration, metrics creep upward and results improve. Unfortunately, improvements may get smaller and smaller, while each incremental gain gets increasingly difficult to achieve — and the result may be locally optimal for the tightly defined problem, but not globally optimal for the larger managerial problem.
It’s true that variations of tantalizing headlines may generate more impressions for an article. Or that ever-more-refined images and copy may lead to more clicks on an ad. In each case, data collection and analytical models can improve a metric. That data can lead to deep understanding of what works best in a narrowly defined problem.
However, improving an unambiguous metric is rarely the overall goal. In the case of articles, impressions are an imperfect measure of the number of people who actually read and derive value from an article. For ads, the number of clicks is in a rough indicator of the amount of potential sales that may come from the ad. Neither metric is the true goal.
Metrics can help measure and improve progress towards an overall goal and are critically important to the use of analytics. But intense focus on a narrow measure can address only the well-specified puzzle — a myopic view of the problem.
We are selling analytics short if we stop there. The data and methods embedded in analytical approaches offer more. They offer the ability to explore unexpected relationships that may not only solve the immediate puzzle and climb the top of a local hill, but also find unforeseen options.
Given the hype around analytics, some disillusionment may be inevitable. With gains coming more slowly or requiring more effort, executives may suspect that they have peaked out on a local optimal. Or they may desire more from their analytical investments … or look for something completely different. But how?
- Step back from the specific problem: The blinders and focus that work well to optimize the details of a problem may prevent managers from seeing other options.