Less Mad Men, More Mad Skills: Making Predictive Analytics Work for You


“In my heart, I know we cannot be defeated, because there is an answer that will open the door. There is a way around this system. This is a test of our patience and commitment. One great idea can win someone over.”

This great quote, from Don Draper in Mad Men, shows how marketing was once based on gut instinct. But, the nagging question remains, “Will my gut instinct and great idea lead to actual conversions?”

Predictive analytics brings us into the modern era of marketing with tools that enable us to be methodical and informed about not only how we make decisions, but also the potential results of those decisions for our businesses.

Predictive Analytics 101

When it comes to predicting customers’ futures, there are key terms that you will typically hear. Beyond these terms, you may encounter other predictive-analytics terminology such as regression or decision trees. However, before predictive-analytics models are built, many data scientists employ advanced diagnostic-analytics techniques like clustering. Clustering involves grouping people who are similar to one another but very dissimilar from those in another cluster or group of people. Clustering determines how people are ‘similar’ based on a number of different demographic or behavioral traits. Clustering is a more objective approach to audience segmentation.

Once clusters of people are uncovered, predictive analytics is the natural next step in providing higher-order insights and in uncovering opportunities. Predictive analytics leverages algorithms to help you discover answers through:

  • Predictive Modeling: “What will happen next?”
  • Forecasting: “What if these trends continue?” or
  • Simulation: “What could happen based on our changes?”

An Iterative Oracle

Predictive analytics can help marketers predict the futures of customers or prospects. How likely are they to convert? Churn? Will they move forward in the sales funnel?

But, it should not only be used to predict future customer behavior. It can and should be used to assess the probability of different potential futures, and then, based on those results, recommend marketing communications that optimize a conversion for that customer. For example, in early March, much of the United States was enthralled in filling out March Madness NCAA Basketball tournament brackets before the games began. And then, they crossed their fingers, hoping for a good outcome. But, if they had access to predictive analytics, their projections could not only be based on much more than their own analyses, but also adjust based on new and developing information. With each basketball game upset and each win, one could change predictions and brackets accordingly. You could also adjust predictions based on current performance in a particular situation. I’m sure many people who picked North Carolina before the championship game started would have switched to Villanova after seeing their dominant performance in the first half. Predictive analytics can be used to predict future customer behavior, but it can also — and more importantly — help a marketer adjust and optimize a customer’s current experience.

Let’s Get to Work: Predictive-Analytics Guidelines

When starting a predictive-analytics program, one of the biggest mistakes made is asking a key question that is not aligned with real, meaningful business objectives. Ask yourself whether the challenge you seek to resolve aligns with what your CMO thinks and cares about. Don’t waste your time and business resources on a project for which there is no strategic end or executive buy-in.

Another mistake is not having a clear vision of what will be done with the results. Too often, the predictive model is not actionable once created. If there are no clear mechanisms for actioning in the real world, then the results of your efforts may end up as a PowerPoint presentation and nothing more.

Mapping the Journey: Analytics Maturity

Predictive analytics is a step in the natural progression of analytics maturity. Analytics maturity starts with descriptive analytics, establishing a baseline: “How are we doing? What is happening?”

The next step is diagnostic analytics: “Why are these things occurring in my data? Which audiences are driving those?”

Now, you are ready to make predictions. Predictions, however, are not the end goal. Predictions set you up for prescriptive analytics in which the system prescribes the best situation or decision given all possible outcomes.

Finally, there is a new area, called cognitive analytics, where machines and algorithms combine artificial intelligence-based cognition with the prescription. At this level, they actually make some decisions on your behalf.

When you reach this level of maturity, you are starting to use your data effectively to make predictions and business decisions for your company. As you collect more data, and improve existing data, you are able to reduce error. The end goal is optimization — to perform better than you might have done using instinct alone.

Create Your Future

In his book, Competing on Analytics, Thomas Davenport sums it up it well: “Any company can generate simple descriptive statistics about aspects of its business … but analytics competitors … use predictive modeling to identify the most profitable customers — plus those with the greatest profit potential and the ones most likely to cancel their accounts.”

Rather than a reactive approach, predictive analytics allows you to take a proactive approach to marketing and optimization — eliminating the guesswork in marketing. Instead of just hoping for the best future, you create your future by taking control of it.

This post taken from http://blogs.adobe.com

Profile Status
Profile Info

Kalyan Banga224 Posts

I am Kalyan Banga, a Post Graduate in Business Analytics from Indian Institute of Management (IIM) Calcutta, a premier management institute, ranked best B-School in Asia in FT Masters management global rankings. I have spent 14 years in field of Research & Analytics.


Leave a Comment

4 × 2 =