Using Predictive Analytics, Without Ever Making A Prediction

Last week I had the privilege of participating on the Advisory Board for the Retail Marketing Analytics Program (ReMAP) at the University of Minnesota, Duluth (UMD). Perhaps the best part of these sessions is the opportunity to meet with the students, many of which will be tomorrow’s marketing scientists.

During a few conversations on this visit, I was asked how to secure an entry-level position that would involve lots of cool predictive analytics. I want to focus on one of the answers I shared — don’t tell anyone you’re doing predictive analytics. What do I mean? Imagine you’re a freshly minted analyst in the following situation:

  • Your manager asks you to quickly evaluate who responded to a promotion.
  • You have many factors to investigate (because you have lots of data).
  • You have very limited time to find a great answer and build a deliverable.
  • The required deliverable needs to be simple and free of analytic jargon.

Sound familiar? What is a young ReMAP graduate to do in her first job? Perhaps her plan could involve:

  • Choosing a predictive modeling algorithm that is resilient to outliers and missing data (i.e., decision trees).
  • Setting the outcome to “response” and rapidly sorting through what factors predict who responded to the promotion.
  • Quickly deploying a few algorithms to make sure nothing too obvious gets missed. In the case of decision trees, try one approach that constructs short, bushy trees (i.e., CHAID) and one that builds relatively tall, thin trees (i.e., CART).
  • Designing simple charts and graphs based on the variables chosen by the modeling algorithms.
  • Building a short deliverable with the most compelling visuals and no reference to predictive analytics.

So our young analyst would quickly identify correlates to campaign response and recommend targeting or other campaign improvements. She would undoubtedly add value through predictive analytics, but not through a scored list. She would not need to mention predictions, techniques used, model validation, or deployment.

So what do you think? Would a new analyst following this process be desirable on your team? Future ReMAP graduates are interested in finding out. I welcome your comments below or you can email me directly. I plan to summarize comments in a future post.


Why just predictive?

Jason - great post! We hire and talk to many grads like the one you are talking about here. There is a shift in the market at the moment. People are more worried about the impact of new data sources (Hadoop, Big Data) and new software paradigms (web vs. download and visual vs. script-based).

The student should also think more broadly about the field. More people are now engaged in the field of fact, we prefer to call it "Advanced Analytics" than "Predictive Analytics". "Advanced" is broader than "Predictive" and in fact 'Advanced Analytics" includes, among other fields - prediction, simulation and optimization.

Hope this helps,

Bruno, Thanks for sharing. I


Thanks for sharing. I agree there are a lot of components to being a good marketing scientist and that includes many of the ideas you brought up.

In the spirit of analytics, perhaps we can agree to a correlative relationship—the graduate who meets your criteria is likely the same graduate I am writing about above.

Thanks again for your contribution. Sorry for my delayed response, I was traveling. Jason

Jason, nice post. I would

Jason, nice post. I would agree that what you have described is important to any successful analyst/statistician/data scientist. However, a major selling point of utilizing predictive analytics is to make sure that the client knows you are an expert in predictive analytics (why did they talk to you in the first place?). Talking about the technicalities of how you solved a business problem will normally stay with your fellow analysts. Clients generally don't care to hear it. They want to know that the problem was solved and the end result is actionable insights they can utilize to better their marketing strategy. In the end, the most important part of a project is what you have described. Any analyst worth their salt will focus on this, be able to communicate it, and know what is important and not important to the client. However, the value we provide is our expertise in predictive analytics and the client should know that. On a side note, we have spoken briefly on decision trees. We both have our opinion on that technique :) Take care!

Analytic detail and trust

Thanks for you comments Ben. I agree that some situations can benefit from sharing analytic details to build confidence and trust in the recommendations. In my personal experience this has been more important for consultants and agencies than internal resources. Trust between CI pros and their business partners is very important, but as we both know, there are many ways to build it.

As an internal resource in the situation described above, I would still encourage our young analyst to create documents that lead with recommendations, which are supported by findings and include the minimal amount of methodology necessary. As a junior algorithm geek myself, I'd certainly allow a pithy, technical appendix, but it goes at the end. And to be clear the young analyst in my blog should not "hide" her approach if asked. That will be an excellent opportunity for her to shine.

Thanks again, Jason

Yes, I think your young

Yes, I think your young analyst, using the approach you described, would be very valuable.

That said, I think an analyst that thinks a problem through and comes up with three of four different tools or techniques to try to solve the problem is valuable because of that thinking... regardless of what the three or four techniques are.

I disagree with the comment above that a major selling point of using predictive modeling is mentioning that you are using predictive modeling. I think people care about outcomes and reasonable thinking, not how you came to the conclusion. The critical idea is that they used problem solving and a reasonable of decent thinking to come up with a solution to the problem.

--> C

P.S. I'm also a fan of UMD's ReMAP program.