David Raab Article Archive

Lifetime value is one of those concepts that are easy to understand but hard to execute. The challenge is a practical one–how do you gather the information needed to calculate a customer’s profitability? Think of the data required:

– promotion costs. Not just the initial media expense, but also the cost to handle replies, fulfill any premiums, and provide ongoing communications. And don’t forget to allocate costs related to non-respondents, bad debt, and returns. Bear in mind the allocated figure can change as you get more responses or fewer payments.

– gross profit margins. Total purchases are just the start; you also need to know what you paid for each item, packing and shipping expenses, and costs for customer service and billing. If you’re in an industry with high fixed costs, such as airlines or telecommunications, you have to deal with allocations yet again.

– projected future behavior. Cumulative past purchases are not enough; you really need to predict future purchases as well. This means looking at how similar customers have already behaved–assuming you can figure out who is truly “similar” in the ways that count. This takes still more data, on demographics, lifestage and behavior patterns, as well as the tools and techniques to analyze it.

– projected changes in behavior. Lifetime value is anything but static: the more you watch it, the more you will attempt to change it through adjustments in pricing, service, promotions, and so on. So you need still more tools to measure the impact of past policies and estimate the impact of new ones.

If you’re not worried by now, you haven’t been paying attention. Clearly where so much data is involved there has to be a computer somewhere nearby. But just as clearly, a process this complex will not have a simple software solution.

In fact, there is no single “solution” to the challenge of lifetime value calculations. Instead, different industries have evolved solutions that make sense given the data they have available and the economics of their business. These solutions are embedded in those industries’ marketing and operational software.

Consider the magazine industry, where fulfillment service bureaus like Neodata and CDS capture transaction statistics from the subscription systems and feed them into models that project future circulation levels. These models are organized primarily by subscription source, which circulators have found are the most meaningful predictors of subscriber behavior. Each source has a set of assumptions regarding renewal and payment rates, as well as the number of bills, renewal notices and bad debt copies per subscriber. Add some unit costs for the magazines, bills and renewals, and maybe an assumption about advertising revenue per paid subscriber, and the lifetime value per new subscriber by source is a mechanical byproduct of the modeling process. Magazine circulators have made this a science for decades–in fact, they did it by hand before computer models were widely available.

Catalog marketers also speak in terms of circulation, but they share little else with their magazine cousins. Most catalog fulfillment systems maintain a running total of purchases for each customer, primarily for Recency-Frequency-Monetary Value segmentation. This is often labeled “lifetime value”, although it does not usually incorporate any projection for the future. At best, the magazine systems provide a report of that groups customers by their initial order date and shows each cohort’s lifetime purchases. This gives some sense of the flow of purchases over time, although it’s usually left to the marketer to build any type of projection curves. The magazine systems generally do not record the cost of catalogs sent per customer, let alone other expenses such as customer service or billing. They often contain actual merchandise costs and frequently can show profit margins (that is, revenue less cost of goods) by catalog or product line. But they generally do not calculate margin on a per customer basis.

Margins are a much greater concern–bordering on obsession–to retailers. General merchandisers and grocery stores both use price adjustments as a major marketing tool, to attract traffic during sales and to move excess merchandise. As a result, their database marketing systems often do track the difference between what each item cost and what the customer actually paid. These systems also record the costs of promotions and calculate profitability, net of both promotion and merchandise costs, per campaign. Products in this group include Retail Target Marketing Systems Archer (414-798-1705), RMS MarketExpert (203-656-3411), S2 Systems CRM (972-458-3800) and STS Systems Open MarketWorks (514-426-0822). These systems also will often divide customers into segments based on lifetime purchases and show the profitability by segment. They still do not project future purchases by customer, however.

Perhaps the most advanced lifetime value calculations belong to portfolio management systems such as Fair-Isaacs TRIAD (800-999-2955), AMS Strata (703-267-8760) and CCN-MDS Strategy Management/PROBE (404-841-1400). Customer Management Services Customer Portfolio Manager (919-969-5233) and Exchange Applications ValEX (617-737-2244) also arguably fit into this category. Originally built for credit card marketers, these products are now applied throughout financial services, in telecommunications and elsewhere. They include statistical modeling to predict customer behavior, consulting to define contact strategies and policies, and software to integrate the strategies and models with business operations such as billing and pricing decisions. Implementation generally involves identification of key business decisions–which might be credit limits, collection policies, and interest rates in a credit card situation–and development of models to predict the results of applying different policies at each decision point to each individual customer. Because the value of earlier decisions can be affected by later decisions–for example, accepting high-risk customers may be profitable only if aggressive collection policies are applied–marketers must define “strategies” that are sets of decisions or options to be applied as a package.

The portfolio management systems also provide extensive support for testing alternative strategies over time, which they typically refer to as “champion/challenger”. The systems gather a combination of operational data, such as account balances and payment rates, and user-provided assumptions for promotion and contact costs.

The comprehensive nature of the portfolio management systems makes them arguably the most complete implementation of the lifetime value concept. They explicitly incorporate models to predict future behavior and, even more important, they attempt to change lifetime value through business actions. This understanding–that lifetime value is is something to manage, not measure–is the key to customer-centered marketing.

David M. Raab is a Principal at Raab Associates Inc., a consultancy specializing in marketing technology and analytics. He can be reached at draab@raabassociates.com.