2006 Jan 01
Customer Value Dashboard
David M. Raab
DM Review
January, 2006
Customer value models are powerful tools to improve business strategy, tactics and reporting. But they need a lot of input data, which is not always fully available. Some people object to the models for this reason. Their concern is that incomplete data will yield inaccurate results, leading to worse decisions than no model at all.

There are ways to overcome data limits. One is to model only areas where detailed data is available. For example, log files can capture a precise history of a customer’s behavior on the Web. Marketers may build a customer value model to help with Web-related decisions even if non-Web behavior is modeled in much less detail.

Another option is to use aggregate values. A model might apply the same average payment rate to all customer segments when actual rates by segment are not known. Sensitivity analyses can determine whether variations across segments would really make a difference in the relevant outputs. If so, special studies can sometimes discover the actual values when these are not available from regular reporting systems. Even without such studies, managers can estimate the different segment values and use simple math to ensure the weighted average of the estimates matches the value of the actual average itself.

One more approach is simply to model with less detail. Again, sensitivity analysis can determine whether working with a simpler model affects the utility of these results.

Which, if any, of these approaches makes sense will depend on the situation. Often a useful model can be built using the available data. Sometimes it cannot.

But even when a true customer value model is not possible, customer value metrics can still be used. Such metrics present the inputs to a customer value model for users to evaluate directly. Although they do not provide the long-term value projections of an actual model, they can still give important insights into business performance and trends.

These metrics are based on the fundamental customer value equation: customer value is equal to acquisition value plus future value. Acquisition value is the net cost to acquire new customers. It includes prospecting costs less any revenues from the initial acquisition transactions. Future value is customer value per period multiplied by the average number of periods in a customer lifetime. Value per period is the net of customer revenues less costs. Periods per lifetime is derived from retention rates.

Standard accounting systems do not measure customer value directly. Yet nearly all can link transactions to customers and transaction dates. This means they can determine which customers purchased during current and previous time periods. (It doesn’t matter whether these periods are days, weeks, months or years. Nor must the previous and current period be the same length. In some cases, it may make sense for the previous period to include activity over the past year, while the current period looks only at the current month.) Once customers have been classified this way, they fall into four categories.

  • purchased in both periods (active customers)
  • purchased in neither period (inactive customers)
  • purchased in this period but not last period (new customers)
  • purchased last period but not this period (lost customers)

These can be displayed neatly in a two-dimensional matrix:

no purchase last period purchased last period
purchased this period new active
no purchase this period Inactive lost

All the inputs for a simple customer value model can be derived from these values.

  • acquisition value per customer is the costs and revenues of new customer transactions, divided by the number of new customers. The costs include all prospecting expenses associated with the transactions, including costs to contact other individuals (that is, non-responders). Although a precise analysis might address some subtle issues of response attribution and cost allocation, simply charging all current-period prospecting costs to current-period new customers is usually adequate.
  • customer value per period is the costs and revenues of active customer transactions, divided by the number of active customers. As with acquisition costs, there are some general accounting issues to address, such as allocation of overheads. The “correct” answer will vary depending on the particular analysis. But, again, simply including all period costs will probably serve in most cases.
  • periods per active customer is calculated by extending the retention rate for however many periods makes sense. Anything from three to ten years may be reasonable depending on the business. The retention rate itself is calculated as the number of customers who purchased in both periods (actives) divided by the number who purchased last period (actives plus lost).

The specific customer value figure that results from this calculation can be intriguing. But it has little practical use. The real value comes from examining the input figures themselves, and in particular how they change over time. The matrix described earlier can be interpreted as illustrating stages in a customer life cycle: from new through active to lost to inactive and possibly new again. Viewed over successive time periods, the values in each cell illustrate this flow.

From this perspective, changes in relative values indicate whether a company is growing or shrinking its customer base and whether retention rates are improving or getting worse. Changes in financial values show whether revenue is growing by adding new customers or expanding relationships with existing ones. Comparing past period financial values of lost customers with current values for actives give some insight into the quality of customers being lost.

A complete set of relevant metrics would include:

no purchase last period purchased last period
purchased this period
– number of new customers
– new customer value
– value/new customer
– active customer value (last period)
– number of active customers
– active customer value (this period)
– value (this period)/active customer
– value (last period)/active customer
– retention rate (nbr active customers / nbr active + nbr lost)
no purchase this period
– number of inactive customers
[these customers have no value figures because they have no transactions this period or last period; any promotion expenses would be allocated to new customers]
– number of lost customers
– lost customer value (this period)
– lost customer value (last period)
– value (this period)/lost customer
– value (last period)/active customer

Because period-to-period changes are so important, a matrix is less effective at presenting the data than a columnar format showing several periods simultaneously. This is similar to any traditional report comparing current and past results.

January February March April
new customers 5,000 6,000 5,000 4,000
active customers 40,000 40,500 41,850 42,165
lost customers 5,000 4,500 4,650 4,685
inactive customers 100,000 104,500 109,150 113,835
retention rate 89% 90% 90% 90%
new customer value ($20,000) ($24,000) ($20,000) ($16,000)
active customer value (this period) $80,000 $81,000 $83,700 $84,330
active customer value (last period) $72,000 $72,900 $75,330 $75,897
lost customer value (last period) ($7,500) ($6,750) ($6,975) ($7,028)
value / new customer ($4.00) ($4.00) ($4.00) ($4.00)
value/active customer (this period) $2.00 $2.00 $2.00 $2.00
value / active customer (last period) $1.80 $1.80 $1.80 $1.80
value / lost customer (last period) ($1.50) ($1.50) ($1.50) ($1.50)
retention rate 90% 90% 90% 90%

A natural extension, also common to such reports, is to calculate differences between one period and the next. Exception reporting can then highlight changes that might need closer examination. Add some attractive graphics and you have the basis for a customer value dashboard.

January February March April % change
new customers 5,000 6,000 5,000 4,000 -20.0%
active customers 40,000 40,500 41,850 42,165 0.8%
lost customers 5,000 4,500 4,650 4,685 0.8%
inactive customers 100,000 104,500 109,150 113,835 4.3%
retention rate 89% 90% 90% 90% 0.0%
new customer value ($20,000) ($24,000) ($20,000) ($16,000) -20.0%
active customer value (this period) $80,000 $81,000 $83,700 $92,763 10.8%
active customer value (last period) $72,000 $72,900 $75,330 $75,897 0.8%
lost customer value (last period) $7,500 $6,750 $6,975 $9,370 34.3%
value / new customer ($4.00) ($4.00) ($4.00) ($4.00) 0.0%
value/active customer (this period) $2.00 $2.00 $2.00 $2.20 10.0%
value / active customer (last period) $1.80 $1.80 $1.80 $1.80 0.0%
value / lost customer (last period) $1.50 $1.50 $1.50 $2.00 33.3%

In the hands of an insightful reader, such reports are a powerful tool for understanding the business. Drill-downs to finer detail add even more value. These might present the same metrics for subsets such as specific products or customer segments. Or they might break the aggregate metrics down into components such as different categories of costs.

These refinements are common to many types of reports. What’s important is not such details, but basing the report on customer value metrics. This incorporates the customer value perspective into corporate reporting even when customer value modeling itself is not a practical option.

* * *

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.

Leave a Reply

You must be logged in to post a comment.