March 09, 2009

Analysts And Actionable Findings

Last year, I saw a former boss at a conference. This individual said "You're business must be struggling, after all, most people couldn't understand any of your findings."

Nice to see you, too!

One of the challenges for those of us who analyze data is to provide actionable findings. This is easier said than done.

Findings fall into at least four categories.
  • The results simply aren't actionable. This happens for at least two reasons. Our business leaders ask us to provide information that isn't actionable "tell me how much more multichannel customers spend compared with single-channel customers". More often, this happens because we genuinely provide something that isn't actionable --- interesting (customers who visit the e-mail registration page view another 2.483 pages on the website), but not truly actionable. This is a character flaw of the analyst. All of us possess this character flaw. We dazzle ourselves with our own ability to create metrics, without considering if anybody outside of the analyst community finds our metrics valuable.
  • The results are actionable, the business leader cannot digest the results. This one is painful for any business analyst. Have you ever tried to explain interaction terms in a logistic regression model to a traditional direct marketer? You hold the key to the secrets of your business, but the person you are explaining the results to owns a different set of skills, and may not ever be able to understand what you are trying to explain.
  • The results are actionable, the business chooses to not act upon the results. This isn't quite as frustrating, because the analyst can accept that the business chooses to not act upon actionable findings --- heck, this happens dozens of times a day, every day. Of course, these businesses get stuck, unwilling to learn and evolve. We've all worked in environments like this.
  • The results are actionable, the business acts upon the results. You'll have a very happy analyst team when you see this happening. However, this relationship requires a high level of trust. As an analyst, you'll have had to have been right a lot of times, helping your leader look good time after time, in order for the trust to exist to allow the business leader to act upon your findings.
For me, the most challenging environments are those where you have genuinely actionable results that folks struggle to understand due to legacy knowledge. You're like Galileo, trying to explain to everybody that the Earth rotates around the Sun, while all around you the chorus redirects everybody to the industry script.

It's been interesting to read the blog comments and e-mails, of late. A faction of the direct marketing community is really struggling with the rapid pace with which customer behavior and direct marketing fundamentals are changing. We analysts have a responsibility to somehow communicate complex, non-linear processes to a community that prefers to view the world in a linear manner. We're not going to be successful by making our message more complicated, even if the way customers are behaving is inherently more complicated.