Optimize It

Cross-selling has become an important focus for many companies since it’s been shown to generate incremental profit and improve long-term customer retention. Marketers also recognize the CRM perspective that, after offering the best products, successful cross-selling depends largely on an offer’s relevance. But too often database marketers struggle to reach this difficult goal.

Here are some suggestions that might make things easier.

• Redefine and create relevance

  A truly suitable offer combines the right timing, an appropriate product and an appealing offer mix. The interplay and synergistic relationship among these elements can make all the difference.

  But these elements generally are considered separately. In particular, offer timing has been subjected to only rudimentary trial-and-error analysis; at worst, it’s gotten only lip service. The best approach provides one integrated predictive model.

• Build a holistic marketing platform

  Consumer behavior theory points out that a successful CRM philosophy is based on providing value to customers so the relationship can be strengthened and improved, and generating profit from these customers so this relationship is desirable for the business.

  To balance these goals, a company must approach cross-selling within a holistic framework that considers both the customer’s needs and its own. For example, when creating a predictive model to identify customers for a particular cross-selling offer, each customer’s needs must be estimated and the knowledge used to provide relevant offers for attractive products. However, the customer’s preference for a product also has to be evaluated in terms of what’s important to the company.

  Without considering both sides, the firm risks spending heavily on ineffective cross-selling programs or making less profit than hoped from resulting sales. When an integrated approach is put in place, both sides win.

• Discover real optimization

  Marketers use models to improve their programs’ efficiency, but they’re not statisticians or programmers. And so they need a means to make optimal yet practical decisions.

The term “optimization” is rather abused in the DM community since most of the so-called optimization engines deliver nothing but simple sorting or order ranking. They can’t handle the very real, involved demands of many marketing decisions. One has to look at the budget, different product mixes, customer portfolio, and both short- and long-term effects.

Take the example of a bank where marketers had a budget of $200,000 for one quarter. Because the debit-card product team used $50,000 of it, marketing had to spend no less than 25% of the total promoting debit-card products, regardless of the predicted next-best product.

Marketing also had an annual target for home equity lines of credit and was hoping to reach 20% of the goal in the same quarter. Finally, at least 50% of the best customer segment was to be targeted.

The team had to use a predictive model to optimize the overall campaign profit subject to these constraints. Simple sorting or order ranking can’t easily solve such a problem.

Fortunately, today’s analytical techniques provide solutions that go beyond sorting. One breakthrough approach for database marketing is to use a powerful tool called mathematical programming to address such issues. MP has long been used by brand marketers, supply chain managers and logistics controllers to solve complex, practical business problems.

In this case, the bank’s marketers balanced and met their demands through MP techniques of linear, integer and goal programming. The result? Marketers could focus on cross-selling that was relevant, streamlined and cost-effective. In a word, optimized.

HONGJIE WANG is vice president of customer analytics and manages the analytical team at database marketing firm Fulcrum’s Fairfield, CT office.

NL

For more on ROI, subscribe to the MarketingROI newsletter by Richard H. Levey at http://subscribe.chiefmarketer.com/subscribe.cfm.