People have widely differing patterns of purchase behavior. This is obvious, and most of the time marketers are very aware of it and base their marketing strategies on this fact. It is why marketers develop segmentation schemes and target markets, why they provide a choice of products and features, and why they target and personalize direct communication. However, there are times when standard data analysis tends to ignore the existence of individual differences. This article will give a specific example that is of direct interest to many where group performance is tracked over time. But even if it doesn’t seem relevant to your situation, you should remember the broader point: forgetting that people are different can have significant consequences for marketing strategy.

The example concerns loyalty and attrition analysis. It is commonly observed that customers with long tenure as a loyal customer are much more likely to be retained as loyal customers than are new customers. Figure 1 shows a typical attrition pattern, where the attrition rate declines as customer tenure increases. This is particularly easy to see with "subscription" services, such as magazine subscriptions, long distance service, cable TV service, direct deposit of paychecks, etc. But a similar concept could be applied to brand loyalty for consumer packaged goods or automobiles.


Figure 1: Typical Attrition Pattern

It would be very easy to look at Figure 1 and conclude that a period of intense retention marketing early in a customer’s tenure would "get them over the hump" – keep them loyal until tenure-based loyalty "kicks in" after seven or eight periods. But even if a strong retention incentive can be provided for new customers, this program will not necessarily help. To understand this, let’s consider why the attrition pattern in Figure 1 looks the way it does.

There are two factors that can contribute to attrition rates that decline with tenure.

True Increased Loyalty: The first factor is the obvious one: as customers gain experience with the product they become more loyal.

Loyalty Differences: The second factor has to do with differences across individuals. As a simple example, let’s assume that there are two groups of customers: "loyal" customers who have a constant attrition rate equal to 5 percent, and "disloyal" customers who have a constant attrition rate equal to 30 percent. If 1,000 new customers of each type are acquired at time zero, what will happen to the average attrition rate of this customer group over time? As time passes, relatively few of the disloyal group will remain as customers. Therefore, the average attrition rate will decline to the level of the customers that are left: the loyal customers. (See Figure 2.)


Figure 2: Attrition Rate Table

Why does it matter? To simply understand which customers are likely to leave, it doesn’t matter. Customers with low tenure are more likely to leave whether or not the explanation for this pattern has more to do with true increased loyalty or loyalty differences.

However, if the marketer is trying to intervene with new customers to reduce attrition, it matters a lot. Let’s examine the two scenarios.

True Increased Loyalty: If individual customers are truly experiencing increased loyalty over time then getting them to stay for an initial period of time can increase the number of long-term loyal customers. Once they get through the initial high-risk period, they become loyal customers.

Loyalty Differences: However, a new customer retention program aimed at the loyal and disloyal groups previously described would not have any long-term impact. The loyal customers were likely to stay, and the disloyal customers will likely leave after the retention program has ended.

In the latter case, when one opportunity goes away another one emerges. While the presence of strong loyalty differences among customers limits the ability to use new customer retention programs, the fact that a group of loyal customers exists opens the door for other targeted programs. One possibility is targeting acquisition programs towards loyal customers, as discussed in my April 2002 column, "Modeling to Reduce Attrition." (http://www.dmreview.com/master.cfm? NavID=55&EdID=5016).

While it requires some advanced modeling, it is possible to quantify the extent to which loyalty differences are driving the decline in attrition rates. For this there is a specific data requirement: observations on multiple tenure spells for some customers. If there is a group of loyal customers, then these customers should have multiple spells of long tenure. Disloyal customers should have multiple spells of short tenure. Therefore, loyalty differences are quantified by looking for tenure correlations across spells.

In this kind of study, it is common to find a substantial amount of differences between individuals. This is not surprising – the premise of this article is that people are different. Less obvious are 1) the implications of the finding and 2) the ability to quantify the underlying characteristics of customers. Rather than jumping to the "obvious" conclusion, we need to remember that the world is complicated. As Einstein once said, "Things should be made as simple as possible, but no simpler."

Register or login for access to this item and much more

All Information Management content is archived after seven days.

Community members receive:
  • All recent and archived articles
  • Conference offers and updates
  • A full menu of enewsletter options
  • Web seminars, white papers, ebooks

Don't have an account? Register for Free Unlimited Access