Customer Retention: A Framework to Follow and Pitfalls to Avoid

Blog Post
March 22, 2018

The definition of customer retention is simple: a customer continues to transact with your brand. But achieving customer retention? Not so simple. In fact, as Scott Neslin, Albert Wesley Frey Professor of Marketing at the Tuck School of Business, Dartmouth College, explained at a Harte Hanks Marketing Advisory Board meeting, customer retention is downright difficult to achieve in practice and can be expensive to boot (case in point: T-Mobile's "Un-carrier" initiatives that allows loyal customers to keep the same rate plan for the life of their relationship with the brand).

So, what's a brand to do?

Fortunately, Scott has worked with a group of fellow marketing experts to develop a framework for customer retention efforts and outlined the challenges brands can encounter along the way. Read on!

Much of Scott’s talk at the Advisory Board meeting and much of the content in this post was drawn from a 2018 paper he and his colleagues published in the journal ‘Customer Needs and Solutions’: In Pursuit of Enhanced Customer Retention Management: Review, Key Issues and Future Directions.

The Customer Retention Framework


The graphic above illustrates that any single retention campaign should fit within a larger multi-campaign integration, which must align with overall marketing strategy (retention isn't everything!). While it sits at the bottom of the illustration, all retention efforts should start with data, methods and measurement. 

This framework appears to be fairly straightforward. However, there are a variety of challenges along the way that marketers must address to achieve success with customer retention efforts.

Measurement and Survivor Bias

One of the challenges in retention is simply knowing how we are going to measure our retention from the outset. This is no trivial feat, and Scott explains that marketers must be wary of "survivor bias" (Fader and Hardie 2010).

Put simply, those that find and purchase from your brand organically may have a higher retention rate than those that you acquire with promotional efforts. As a marketer, you would look at the numbers below and assume that your retention rates are going up over time due to your own marketing efforts. In reality, your retention rates are going up because those that are loyal to the company make up disproportionately more of your customer base over time, thereby inflating your retention rate.


Listen to Scott explain survivor bias and the graph above in his own words:

Scott says: "It's really difficult to disentangle whether you're really improving retention, or if some of the customers you acquire have a high built-in retention rate, and over time these are the “survivors” that make up a larger proportion of your customer base. It makes one think twice about looking at aggregate retention rate and using that as a KPI."

His advice? At a minimum, you want to examine cohort groups and acquisition channels when measuring your retention rates.

Single Retention Campaigns

Let's dive deeper into the components of single retention campaigns. What do we know about each box? How do they fit together? Where are the challenges in this piece of the Customer Retention Framework?

Who is at risk?

The first step of a single retention effort is to use a predictive model—usually a churn prediction model—to determine who is most likely to churn. Retention efforts should focus on retaining these individuals. But, the predictive accuracy of churn models isn't perfect.

For example, take a look at the data from the studies below that examined various wireless carriers:


The left hand column is the lift they're able to get from the predictive model. The "2.29" means that, in their top decile, those people are about twice as likely to churn than average. This sounds like success, right? We've identified people that are much more likely to churn than others. But, what's the average churn rate? It may only be two percent. In the top decile, then, four percent would churn. On the surface, it makes sense to aim a retention program at this top decile, but ultimately, 96% of them would have stayed with you anyway – potentially a large waste of incentives you might use in your churn management campaign.

Listen to Scott explain:

The takeaway? When the model builders say they've found X% lift, ask: "What percent of those people would stay anyway, without a retention initiative? Can I justify a campaign financially against this number?"

Who to Target

We've used our churn prediction models that explain who is at risk. The next step is to determine who, specifically, to target—and it's not always obvious.

The traditional view is to target those most likely to churn. But Scott explains that marketers should also be integrating other important information about these individuals. Specifically, how much can you increase the likelihood they will stay with you; how will they respond incrementally to your retention efforts.

Look at it this way: Maybe you can predict a particular customers a high likelihood of churning, but what matters is if you can reduce that likelihood. That high-churn likelihood customer may be a lost cause. You might be better off letting that customer churn and focusing on the customers for whom your retention campaign can make a difference.  We want to find the people for whom we actually have a good shot at influencing whether or not they will churn. This is called targeting based on uplift or incremental response.

Uplift modeling and uplift targeting is not easy, methodologically. But it is critical to be measuring incremental response to your efforts versus overall response. The best way to do this is to run tests with control groups to tease out who you are most likely to have an incremental response with.

Who NOT to Target—the Sleeping Dog Phenomenon

When determining whom to target, you want to watch out for the "sleeping dog phenomenon." Scott explains that retention campaign can increase likelihood of churn among those who would not have churned (Berson et al 2000). Customers may be jolted from a mindless habit to active consideration of pros and cons of staying with the company (Wood and Neal 2009).

In one field experiment (Ascarza et al 2016), telecom customers were given an offer for a better telecom rate plan. Even though they were receiving an offer than would save them money every month, the group that received the offer was much more likely to churn.


For best retention results, we would want to identify sleeping dogs in advance to avoid waking them. But there's no clear way to do this yet! For now, marketers should take this phenomenon into consideration when planning a retention campaign.

Multiple Campaign Management—the Recency Trap

Scott explains that the key with integrating single retention initiatives into multiple campaigns is to be forward looking to maximize the return we got now plus what we'll get in the future because of what we do now.

Some marketers fall prey to the recency trap (Neslin, Taylor, Grantham, McNeil 2013). The longer it has been since the customer last purchased, the less likely they are to purchase now. It can therefore be tempting to give up people down the curve in order to concentrate on the most recent buyers that are more likely to come back. But this becomes a self-fulfilling prophecy.

Scott explains how marketing can still have an impact on customers that have not bought as recently, which improves longer term incrementality.

The recommendation here? Adopt of policy of reaching out to customers around four weeks post purchase to avoid the recency trap. It may cost marketing dollars in the short term, but it provides return in the long term.

Strategy Integration

Finally, we come to integrating our retention efforts into our larger marketing strategy. Scott's advice here is that we need to market in a way that builds retention AND builds the brand.

Marketers can lose sight of the bigger picture of the value proposition and how we're communicating when we're focused on retention. For example, doing a lot of promotions and discounts to improve retention can undermine the brand.

Brand equity and retention efforts need to be co-managed. In the diagram below, all connecting arrows need to have positive impact [Sources: Stahl, Heitmann, Lehmann, and Neslin (2012); Luo, Lehmann, and Neslin (2014].


It is possible! But the warning is: don't get caught up in detailed retention efforts at the expense of underlying brand equity.

Bringing It All Together

In sum, our key lessons on retention are:

  1. Simple measures of your retention rate may not be as straightforward as you think. Be careful for distortions like survivor bias.
  2. There are a lot of people doing predictive models to find out who is most at risk of churning—but these may or may not be the people you actually want to target with retention efforts. Consider uplift and incremental response, and be careful of the sleeping dog phenomenon.
  3. To avoid the recency trap, marketers need to be forward looking. Work to get not-so-recent customers back into the purchase cycle even if it does not pay off in the short term.
  4. Retention efforts must integrate with the overall brand strategy. Brand equity and retention must be co-managed.