Recency, frequency and monetary value (RFM) analysis is one of the most widely used tools in the direct marketing and loyalty marketing fields. The tool grades customers based on their purchase behavior on three attributes: Recency of purchase (time elapsed since the last purchase by the customer), frequency (the number of times that the customer has bought your product in a predefined time period) and monetary value (the amount of money that the customer spends on your product in a predefined time period). A combination of these three attributes, or RFM, has been found to be a pretty accurate indicator of a customer's likelihood of responding to a direct mail offer. Empirical evidence also suggests that the model has achieved significant success in the direct marketing field. This model has been adopted by the loyalty marketing industry as means of predicting customer value and, by a stretch of definition, customer loyalty. The objective of this article is to scrutinize the validity of one of the variables used in the model namely Recency and highlight some of the dangers in adopting it as a measure of customer value, let alone customer loyalty.

Interpretations of Recency
Recency provides the answer to a simple question which is: how long has it been since the customer last bought my product? Direct marketing professionals are familiar with a related concept of "recent purchaser's hot list." This hot list of recent purchasers does not refer to buyers of your product but is more an indication of customers who are in the spending mood. Responses from mailers sent to such hot lists have traditionally outperformed other standard lists, since the mail caught the prospect when he was in a splurging mood. What we will be calling recency in this article will be recency as applied to a particular product or service as defined in the first sentence of this paragraph.

Current Application of Recency
This concept of recency has now been adopted into the loyalty marketing sphere to identify the most valuable customers, so that the marketing manager could then direct a major portion of his marketing dollars to these most valuable customers. It is this method of identifying the value of a customer which raises a host of questions. Is recency an appropriate tool to identify the value of a customer?

Issues
Let us look at some of the questions in detail: Is a customer who buys more frequently than others a more valuable customer? Definitely yes, since he is giving us more business and is interacting with us on more occasions. Is a customer who brings you more value than others fit to be classified as MVC? Most certainly, because, it is monetary value which is the prime attribute in assessing a customer. Now, let us look at a third question: Is a customer who bought last week more valuable than a customer who bought three months back? The answer is that we do not know. There is no way that we can say with certainty that the former customer is more valuable than the latter. And that, I believe is the primary issue in using recency to predict customer value.

What is the Problem?
To understand the issue fully, let us look at an automobile dealer who has a database of all his customers for the past two years. Let us further look at two customers in the database: Customer A who bought a car in the last month and Customer B who bought it 12 months ago. Now if we were to apply the recency logic without a modifier, it would appear that Customer A is more valuable than Customer B. Yet intuitive logic would tell you that perhaps Customer B would be a safer bet in the short term, though it is conceivable that both customers become equal over a longer time interval. Why would this be? For the simple reason that Customer B will be near his repurchase cycle and so is likely to buy in the short term.

Where is the Problem?
The problem lies in two spheres: what do we expect recency to indicate and how we define recency. By itself, recency is a good indicator of the currency of the customer ... that is it is good enough to show whether the customer is still with you or deserted to another brand. In other words, if the customer has not bought from you in quite a while, then it is reasonable to assume that he has switched brands. Thus recency is more a "negative indicator" than an affirmative one.

Even as an indicator of customer defection, we need to develop other parameters to identify defecting customers. The best way to do this is to identify the "mean time between purchases" (MTBP) and then define the time intervals for recency measurement in terms of MTBP. It is now possible to weed out all those customers who have defected. All those customers who have recency score which is greater than two times MTBP are definitely gone. You don't have to waste your time trying to assess their value to your organization. But all this helps you to identify your least value customers. What about the most valuable customers?

Identifying the MVC
Sadly when it comes to identifying the most valuable customer, the case for recency is not really strong. And with frequency, which is measuring time dependant behavior in any case, recency seems to be a redundant factor. Perhaps the answer lies in addressing some other aspect of customer behavior which approaches the relationship in terms of the depth and breadth of the customer's engagement with your brand. Engagement measures neither time- series behavior nor his commercial value. It is more a measurement of how deeply the brand habit is entrenched in the customer. As a result engagement brings in a new dimension to the assessment of customer value. Needless to say the definition of engagement and the calibration will have to be done on a case by case basis. But knowing how many touch points the customer is using or how many of your products he is buying is a much better indicator of his loyalty than recency.