In today's competitive markets, it's all about focus - reaching the customers that deliver the best return for the least cost. But traditional marketing programs are often more about coverage than focus.
The secret lies in understanding the shape of a curve.
Analysts have long been familiar with the law of diminishing returns - increasing returns in the early stages of a project that diminish as the project matures.
But a concept called uplift modeling goes deeper than this and divides the curve up into three parts. The first part tells the familiar story of increasing returns up to the point where they start to tail off. The second part implies neither gain nor loss, while the third part actually delivers increasing losses - the project returns less than its costs.
Let's apply this to a set of customers. Some will change their behavior positively in response to a marketing campaign; some will be indifferent and some will respond negatively - for example they may shop around in response to the campaign and pick a competitor.
According to Forrester Research, the bottom 20 percent of customers can drain a company's profits by as much as 80 percent while the top 20 percent can generate up to 150 percent of profits - a huge asymmetry that it is important to understand.
But how do we know which customers are which?
Whereas traditional predictive analytics in the consumer space is about predicting the response of customers to a specific campaign, uplift modeling is about predicting the change in their behavior. In other words, where do our customers fit on the curve?
The gap between the two approaches comes into play when customers with a high predicted response score are low-scoring in terms of lift (if the customer was always really likely to buy, then the marketing treatment didn't make the difference).
To demonstrate just how important this is, by applying uplift modeling U.S. Bank was able to increase cross sell revenues by over 300 percent and reduce mailing volumes by 40 percent in a well-publicized case study.
In today's tough market conditions, predictive analytics has a huge role to play in growing revenues and reducing costs bringing financial institutions much closer to their vision of a world of truly personalized financial services.
Portrait Software (www.portraitsoftware.com) is a leader in this field and their products are in use in many leading companies and industries around the world including financial services.
For more details please see the white paper at:
Registration is required on the company website