Customer Lifetime Value
Once the CDP is in place and all customer data is aggregated, the next step in using VRM to anchor a transformation in customer loyalty is to calculate the customer lifetime value (CLV) as a prelude to segmenting customers.
Here are a few quotes from Wharton professor Peter Fader on CLV, sourced from Retail Touchpoints:
The differences [between customers] are staring us in the face. And not only are customers vastly different from each other, but their lifetime value varies by orders of magnitude.
If you can come up with a lifetime value measurement, it’s like a number shining over each customer’s head. CLV provides a different framework for running a retail or brand business. Rather than focusing on product, you would want to figure out what you could do for these really valuable customers. What products and services should you offer to enhance the value of those customers, and to find more customers like them.
The truth is you’ll only get so far with innovation and efficiency; to succeed, you need to think of your customers as individual entities.
If you can quantify the CLV, you can figure out what kinds of discounts you should be offering — or, what kinds of value-enhancing activities you should offer instead of discounts.
Focusing on CLV doesn’t mean ignoring the remaining 80% of customers, but it does mean paying more attention to the high-value 20%.
There are many ways CLV can be wrongly calculated. Many brands just use an average of transactions done in a period of time to estimate CLV. A flawed CLV calculation will lead to an incorrect identification of Best Customers.
The right way to calculate CLV is to look at the recency and frequency of transactions, and then estimate future transactions for each customer. This is the model we use in VRM, building on work done by Peter Fader and others.
Here is an explanation of a CLV model proposed by Fader, Bruce Hardie and Ka Lok Lee in a paper entitled “RFM and CLV: Using Iso-Value Curves for Customer Base Analysis”:
The challenge we face is how to generate forward looking forecasts of CLV. At the heart of any such effort is a model of customer purchasing that accurately characterizes buyer behavior and therefore can be trusted as the basis for any CLV estimates. Ideally, such a model would generate these estimates using only simple summary statistics (e.g., RFM) without requiring more detailed information about each customer’s purchasing history.
In developing our model, we assume that monetary value is independent of the underlying transaction process. Although this may seem counterintuitive (e.g., frequent buyers might be expected to spend less per transaction than infrequent buyers), our analysis lends support for the independence assumption. This suggests that the value per transaction (revenue per transaction × contribution margin) can be factored out, and we can focus on forecasting the “flow” of future transactions (discounted to yield a present value). We can then rescale this number of discounted expected transactions (DET) by a monetary value “multiplier” to yield an overall estimate of lifetime value:
CLV = margin × revenue/transaction × DET
Once the CLV has been correctly calculated, the easy next step is to segment customers and thus get a clear idea on the Best Customers.
Tomorrow: Part 12
