The Predictive Marketing Roadmap: From One to Omni Channels

July 10, 2013

In our blog “How to Get Started with Predictive Marketing”, we talked about just one, simple example of predictive marketing: how to boost revenues by differentiating the number of emails sent for different behavioral customer segments.

 Now let’s look at the bigger picture of Predictive Marketing. What if you really buy into the value of big data for marketers? What is the end goal and how to get there?

 The goal of predictive marketing is to personalize the customer experience across all channels, in order to drive customer engagement and lifetime value

 Predictive marketing is about more than just big data or predictive analytics. In fact, there are three important steps to any predictive marketing campaign:

  •  Step 1: Collect and prepare customer data
  •  Step 2: Analyze, model and predict customer behavior
  •  Step 3: Trigger the right customer action at the right time

The Downside of Omni Channel

Now, you could decide to start by collecting all customer data, across all channels and to figure out what to do with that data later. The downside? It takes a long time to get access to all customer silos (from an organizational as much as a technical point of view) and even longer to figure out what to do with all this big data.  Plus, after collecting all this data, you still have not achieved any of your business goals – which only comes with acting on the data.

 One Channel at a Time

An alternative approach would be to start to collect, analyze and act upon information from just one customer data source and to expand from there. The big advantage is that you can get this setup very quickly (in minutes, hours or days in some cases) and still achieve a tangible ROI. Define just one campaign right from the beginning of your Predictive Marketing project.

 Here is what a potential roadmap could look like:

1.     Email behavior

By collecting just email engagement (opens, clicks, opt outs), you can learn about the level of enthusiasm of your customer base and predict who might unsubscribe if you send too many emails.  In our blog “Six Predictive Email Marketing Tips” we recommend how you may want to adjust email touch to minimize opt outs and to preserve millions of revenues for the future.

 2.    Clickstream data & online purchases (from web tag)

Now adding web behavior will give you many new options to personalize email content. For example, you could run reactivation and retention campaigns.


  • Abandoned on-site search
  • Abandoned AdWord clicks
  • Abandoned Cart
  • Lost customer reactivation


  • New customer welcome
  • Special messages to VIP customers
  • Replenishment orders
  • Preventative touch for at risk customers

 3.    Tie back to your transaction system

After covering most online channels, it’s time to connect back to your legacy order processing system and perhaps your point of sale systems. From a data integration perspective, these are often more difficult. Whereas email data and web data can be collected in days, you probably need weeks for offline data. However, by now you should have a powerful ROI to prove to the organization that this investment is well worth the time and money.

Special circumstances

The above are just an example. It is fine to choose a different channel to experiment with. Perhaps you have a direct marketing program that you would to optimize, or perhaps the priority is to do more cross-sell through your customer service reps. The main point is that in order to achieve a successful business outcome, you need to complete the cycle of data collection & preparation, data analysis and prediction and campaign execution. Leave out one of the steps and you will have all the data but none of the ROI.  Starting small and dreaming big is the key to success.

The key is to start with a specific campaign you want to run that will drive business results and then work backwards from there to decide which data you need.