MINING YOUR CUSTOMERS FOR ALL THEIR WORTH
by Chris Watkins, Vision Associates D S *

INTRODUCTION

Data mining combines the best of modern artificial intelligence techniques, statistical analysis and the large scale processing of data to tackle a variety of bottom line business questions. It enables companies like yours to discover new information and new solutions by mining the vast wealth of data your company now keeps. This qualitative and quantitative data is about your business's products and processes and, most important, about your customers.

We would like to tell you about how you can extract the most from your rich reserves of client and customer data using data mining techniques. We believe that data mining can empower your company to maximize the value of its customer information. Read on to see how data mining applications can enhance your efforts to attract new customers, and to retain and fully utilize your existing customer base. Find out how data mining can help you better distinguish between customers that are worth your efforts, and those that are not.

GRAB 'EM: USING DATA MINING FOR CUSTOMER ACQUISITION

Which of your competitor's customers can be "stolen"? And which ones would you want to have, anyway? What business would not want to know this information? The answer is in your own data. For example, credit card companies often have large amounts of information on potential credit card customers, as well as, data on what happens after some of them actually become customers.

These companies, and yours, can use data mining techniques to find the characteristics of the best customers and to identify the types of customers you have been most successful in attracting.

Armed with this information from your own data, you will be able to score prospects in two ways. First, you can score them by their likelihood of being grabbed by considering the past success of grabbing prospects. Second, you can score prospects by their potential profitability -- how much they are likely to spend with you versus how much they will cost you to acquire them. With these two scores you can find the best prospects for new customers with the least amount of effort.

HOLD 'EM: USING DATA MINING FOR CUSTOMER RETENTION

You have lots of data on past clients, such as, product interests, account balances, demographics, number of contacts, etc. Mine that data to avoid losing your existing customers unnecessarily. You know which of your customers you have lost in the past and which of your customers have stayed with you. Applying data mining techniques to this large amount of data allows you to identify which elements in the data are statistically significant factors.

With this information in hand, you can then develop a classification model for your customers. This model can be used to determine which of your current customers are more likely to defect from you. You can then focus on customers who are in danger of defecting since they have the same or similar characteristics to the ones who have already defected. Once you identify the customers who are most likely to jump ship, you can concentrate your precious marketing dollars on this group.

SQUEEZE 'EM : USING DATA MINING TO INCREASE CUSTOMER VALUE

You certainly know that it is usually much more expensive to acquire new customers than to maximize the amount of business that you get from your existing customers. You can cross-sell more products and services to your current customers, but you need to know what they are likely to need and want. You can get to know this by applying the association technique of data mining to analyze the buying or ownership patterns of your current customers.

You can also examine the sequential patterns of buying in your current customer base. These patterns will tell you what a particular set of customers will be likely to purchase next from you so that you can customize your advertising and promotions to those customers accordingly.

PASS 'EM ON: USING DATA MINING TO KNOW WHEN YOUR EFFORT EXCEEDS YOUR REWARDS

This may sound like a heretical idea -- getting rid of customers -- but there are customers that you might be better off without, or at least spend minimal effort to keep. Data mining can help you to decide who is worth keeping, from a profitability standpoint, and who is not.

Your data has information on the lifetime value of your customers. The lifetime value of a customer takes into consideration the monetary value of the customer's purchases, purchase frequency, and the cost of doing business with the customer. The value of customers to you in future benefits is based on their net lifetime value versus your opportunity costs in keeping these customers.

A variety of data mining techniques can help you decide which customers you want to keep and which you do not need to hang on to. Time series analysis can tell you whether certain customers are becoming less valuable over time. Value prediction can tell you what the value of your customers is to you at a particular point in time. Segmentation can tell you if a customer belongs to a customer group that is projected as decreasing in value or as being more costly to maintain in the future.