Data-driven marketing is like a three-legged stool. It symbolizes three marketing levers working together: offer, creative and audience.

In fact, omnichannel marketing is thriving because of marketers' skill and discipline in producing relevant new product offers, richer varieties of creative approaches and vast amounts of new data for segmenting and targeting audiences.

  • A balance between these business drivers—offer, creative and audience—is essential to our success as marketers. Yet, introducing new offers is often slowed by test and learn-based launches.

Our marketing data ecosystem is inherently backward-looking, driven by past transactions and behaviors. It is no doubt incredibly effective. But given the rapid pace of product line extensions and new product introductions, some recent buyers may no longer represent the “next best customer” to target. And while the truism that there are few ‘brand’ new ideas in marketing, when it comes to new offers and creative tests, marketers must still test and learn more about audiences that have never seen, bought or expressed an interest in their new campaign. Yet time is of the essence. Is there a way to accelerate marketers’ successes in learning about channel, incentive, audience and message in their drive to relevance and engagement as quickly as possible?

Consumer behavior patterns offer the most powerful data. Recency and category spend tell us who’s in market. However, marketers launching new product offers, with the goal of seeking to expand new customer acquisition, need to identify new audiences that need their new product. To accomplish this, they need to target differently.

Propensity models can aid marketers through powerful transactional and behavioral data to infer new market prospects. This data is generated from speaking to known consumers and asking about their purchase histories, behavior propensities and plans for future product engagements. And they can be created quickly. But how?

Given a new targeting need, and a wealth of powerful data fed to skilled analytic teams, predictive statistical techniques provide the new rank ordering of prospects and customers that’s needed to fine-tune targeting.

In contrast, predictive models find look-alikes to current responder pools, or to a pool of recent buyers acquired in a recent marketing campaign. In other words they optimize your existing audience, prospects and customers that you’re already in contact with.

Propensity models represent the opportunity to find new audiences. With each propensity model, existing prospects and customers are newly rank-ordered to meet marketers specific objectives. Based on tried and true regression techniques, propensity models represent a beneficial new dimension of data that marketers in all industries benefit from. To illustrate how they work, let’s take a look at a propensity model created to support new offers in the insurance market.

Case in point

An auto insurer’s customers were switching providers annually to get the best price; this trend has been gaining more traction and creating more insurance company ‘defectors’ in recent years. The auto insurer wanted to know: “Who is most likely to buy auto insurance – and stay with us?” To meet their needs, a propensity model was developed—the Auto Insurance Switcher. The client found that by using this model along with other data from Epsilon’s TotalSource Plus database, and their own transactional database, they could create a new strategy to improve ROI and retain more loyal customers year-over-year.

The convergence of technology improvements and an ever increasing proliferation of online channels have created a multitude of customer touchpoints. This increase in consumer interactions has resulted in an abundance of data flowing into the marketing-consumption ecosystem. Given the rapid flow of information, many marketers struggle to make sense of it all, knowing they need to leverage as much of it as they can if their new offers and creatives are going to take root and grow new, loyal customers.

Data-driven marketing professionals have always turned to predictive analytics and information technology to filter large amounts of new data in order to make it actionable. Keeping in mind offer, creative and audience—the three-legged stool of marketing—these market leaders now can access highly targeted propensity models, to succeed even in untested or new product offerings for prospect, cross-sell, and loyalty marketing.

Laura Lucido also contributed to this piece. 

Topics: analytics, data, propensity models, Topic, US, Data, Direct

Join the discussion...