Mining for Customer Lifetime Value
The most successful companies have taken this strategy to the next level - acquiring and retaining the most profitable customers. These are the customers who will add the most to the bottom line, or the highest customer lifetime value.
CLV is determined by calculating the net present value of all future profits expected from a new customer - the total financial contribution (revenues minus costs) of a customer over his lifetime with the company. When you can predict CLV, you add a powerful tool to your decision-making processes. How do you predict future customer revenue streams, then determine which customers will add the most to the bottom line? More companies use data mining to find the answers. After all, the best indicator of future behavior is past behavior.
The explosive growth of information technology has allowed businesses to collect and store tremendous amounts of historical customer transaction data, as well as information collected about customer attitudes and opinions and purchased demographic, geographic and psychographic data.
This information is increasingly stored in data warehouses, central repositories of enterprisewide data collected to support decision-making. The sheer size and complexity of these data warehouses often preclude the use of traditional statistical analysis and decision-support methods, so data mining is employed.
Data mining automates the extraction of relevant predictive information from these large databases. This information is then used to make better-informed and more profitable decisions. This is an iterative process that produces more refined and more accurate information with each cycle. Data mining helps unearth information gems, such as event associations and sequences and customer clusters, that have remained hidden to the marketing analyst. These gems may not be intuitive - they may even be counterintuitive - and, therefore, more valuable and more likely to offer a unique opportunity for competitive advantage.
Before adopting data mining, companies must determine where they are in their data-usage life cycle and where they want to go. What are their data assets? How comprehensive and accurate is their information? How easily can they integrate customer, transaction and business information from across the organization? How fast can they access information? Do they have the internal resources - IT staff expertise as well as technology infrastructure - to begin data mining, or should they outsource their initial efforts? The answers will decide the best opportunities (or not) for data mining.
CLV analysis can be integrated with data mining applications to improve profitability at each stage of the customer life cycle. Predictive modeling applications often are used to identify prospects. Historical customer data are analyzed to ascertain the attributes belonging to a company's best customers - those with the highest CLV. This model is used in acquisition programs to target prospects with similar attributes. Acquisition costs go down, and campaign effectiveness increases when fewer and more valuable prospects are solicited.
When you are able to predict long-term customer value, you also are able to determine whether you are willing to spend more to acquire certain types of customers. Some target segments may merit special acquisition offers. If analysis shows that these prospects have the highest potential for adding to your bottom line and they respond best to special introductory discounts, it may be in your best interest to provide the discount.
Companies also can use data to minimize risk. By evaluating credit histories - using historical transaction data as well as data obtained from credit bureaus, companies can avoid acquiring customers who may not be creditworthy or avoid making offers to current customers that may result in default.
Data mining is used in retention programs to predict customers who are most likely to defect before they are lost. Further analysis will determine the value of these customers, and whether it is profitable to retain them. Cost savings occur two ways: Retaining a good customer is cheaper than acquiring a new one, and funds are not spent to retain or win back customers who won't add to the bottom line in the long run.
With the right information, it is possible to tailor retention programs specifically for the most valuable customers. Some respond best to loyalty programs or frequency marketing programs. Other segments may appreciate being assigned to a dedicated customer service team that provides specialized, personalized attention. And then there are those who wish for minimal contact.
To retain customers and build loyalty you must understand how and when they wish to communicate with you, and you must be able to provide service through all the channels they desire, including phone, fax, mail, e-mail and the Web. It may be cheaper for you to serve customers over the Internet, and you may offer them incentives to do so, but if you have a valuable segment that prefers to visit a brick-and-mortar facility, you must provide them with that opportunity.
Data mining can help companies predict what their customers desire and what they will do. Armed with this information they can refine and optimize customer interactions, offering customers exactly what they want at just the right time. Information obtained from data mining can be used to develop models that predict the "next logical product" that a customer will need or want, and assist in developing cross-sell and up-sell programs. In addition, "early adopter" customers can be flagged for special offers on new or technologically advanced products.
Using CLV analysis to determine your most profitable customers and data mining to inform business decisions regarding acquisition, retention and optimization of these customers will boost profits as well as customer satisfaction and loyalty. n
• Richard Hebert is president/CEO of iSKY Inc., Laurel, MD.