Three Proven Steps To Data-Driven Marketing

Data-driven marketing is a complicated business, but there’s still a simple way to break down the market and develop a strategy, one that’s easily taught to all members of your marketing org.  A couple of years back I posited this same model when developing the marketing strategy for BlueKai, and since that time most companies in the space have adopted the model.  It’s based on three stages: data in, adding or creating value, and data out.

Data in refers to the inputs of data, primarily first-party, second-party and third-party.  These data sets are used for developing audiences to be used by themselves or as models for targeting and/or personalization of your go-to-market.  These data sets are a means of proving or disproving a hypothesis, and make up the foundation of your data-driven marketing strategy.  (Note I don’t refer to this as the foundation of a “data strategy,” because that is a far more grandiose exercise.  This discussion is focused on marketing.)



The second stage is creating or adding value.  This is very often referred to as “the black box,” but it doesn’t have to be.  This can easily be transparent if you want it to be, and I tend to err on the side of transparency.  In this stage you are either activating data through media, and creating interactions with customers (either through exposure or impressions), or you are creating more value by doing things like look-a-like or spend-a-like modeling, all of which are then activated through media.  This stage can also refer to site optimization, dynamic creative and any of the multitudes of other things that can be done to create value.  The vast majority of the ad and marketing tech ecosystems operate here.

The third and final stage is data out.  This is where you read the impact, measure the value and optimize your efforts.  Data out can easily be overlooked, but the goal of this stage is to close the loop on measurement and deliver a true understanding of your campaigns, whether they are ads, emails or other kinds of interactions.  Data Out is also where you create the feedback loop and (if done right) deliver learned data back into your campaign efforts.

If you believe in this construct, and it seems most people do since I’ve seen it in the presentations of many people in the landscape as well as in the different strategic presentations of marketers I speak to, then you are able to determine what partners you need at which stage of the process in order to deliver a truly customized, effective campaign.   This understanding can allow you to create a “marketecture” of the system you are going to need. 

Much of the time this process is couched in a discussion about the desired “day in the life” of the target audience, as you identify and convert them from a prospect into a customer.  I also refer to this as the shift from the unknown audience, to the known audience, to the loyal customer.  The journey is different for every new customer, but the steps required for speaking to and engaging with him or her are the same, just maybe in a different order.

This simple 3-step construct seems to work and I encourage each of you to apply it to your thinking and see if you agree.  You may refine the definitions a bit, but I think the rule of three applies well here.  Let me know!<

1 comment about "Three Proven Steps To Data-Driven Marketing".
Check to receive email when comments are posted.
  1. James Smith from J. R. Smith Group, August 21, 2015 at 3:06 a.m.

    Cory: In my experience, there are three main embedded concerns.  First, knowing what "questions" to ask of the data. This requires extensive knowledge of the target market, brand, and obviously consumer behavior in addition to the dataset's basket of variables. Second, without the superimposition of that framework, what is the data telling you?  More of a discovery process; a search for correlates within the data.  Modeling is a great tool, if you know what marker variables to dump into the model. 

    Third, the marketplace is dynamic, not static, so snapshots are nice but we are really dealing with an active, ongoing process across time.  Results generated from March data might not totally hold in a September data sweep.  (For example, impact of gasoline price fluctations.)

    Your thoughts?   

Next story loading loading..