The Pitfalls of Churn Rate
12–11–2015 | Reid DeRamus
The tectonic shift from linear TV viewing towards digital streaming has triggered an explosion of online streaming video services. The pace of new entrants is poised to accelerate as both traditional brands and emerging networks look to find their place in an increasingly crowded ecosystem. Large investments in content and marketing may draw users in, but the real challenge will be retaining and building relationships with audiences.
There is no silver bullet when it comes to improving user retention — it’s an ongoing cross-company effort and must be ingrained into the team’s culture. However, a flawed approach to measuring retention undermines any initiative or cultural focus on improving it, so ensuring the right metrics are being used effectively is imperative. Churn rate is the most commonly used measurement of retention, but there are inherent shortcomings that often result in misleading analysis and unsound decision making. Let’s begin by defining churn rate and retention, and then dive into the limitations of relying on churn rate as a measurement of retention.
What is “churn rate” and what is “retention”?
Churn rate is an important growth metric and measure of business sustainability. A high churn rate indicates that customers are cancelling quickly and the service is burning through its total addressable market, becoming increasingly reliant on costly customer acquisition to fuel growth. Soaring churn rates are typically accompanied by product deficiencies, an inferior marketing strategy, lack of product/market fit, or competitive pressure, among many other dynamics.
For subscription businesses, retention is typically based on successful billing cycles, and is expressed as a percentage of a cohort that reaches a particular billing cycle. An example of this would be isolating new paying subscribers added during a given month, and then observing the proportion that are successfully charged the second time, the third time, and so on. This is effectively a subscriber decay curve, which forms the basis for measuring average customer lifetime and, as we’ll discuss later, is independent of churn rate.
Methodologies for calculating churn rate
While there is no universal standard, there are two primary methodologies for calculating churn rate; either methodology is effective as long as the same formula is used consistently over time. The first methodology calculates churn rate as the number of paying subscribers that cancel their subscription during a period of time (numerator), divided by the total number of paying subscribers that had the opportunity to cancel during that period of time (denominator). The denominator is calculated as the number of paying subscribers at the beginning of the period, plus the total number of new paying members added during that period of time.
Pay Cancels / (Beginning Pay Balance + New Paying Subscribers Added)
An alternative method is to look at cancels relative to the average number of paying subscribers (vs. the total number of paying subscribers in the methodology above). Here the denominator becomes the average of the number of paying subscribers at the beginning and the end of the period. This methodology will always result in a higher churn rate relative to the other methodology — the numerator is the same, but the denominator will always be smaller because the average pay balance takes into account cancels during the period.
Pay Cancels / (Average [Beg. Pay Balance , End Pay Balance])
Shortcomings & common misuses of churn rate
Churn rate has become a universally applied retention metric in large part due to it’s simplicity: calculating churn rate requires a few inputs, can be measured upon launch, and uses a straightforward formula. This simplicity also presents a few key dependencies — cancels in the numerator and size/growth of the business in the denominator — that occasionally over-influence trends in churn rate, producing confounding and deceptive results.
Formulaic sensitivity to absolute cancels
A key limitation of the churn rate formula is its sensitivity to fluctuations in cancels, which are inherently linked to the number of recently added paying subscribers. Customer decay curves are almost always exponential, not linear, with the probability of a subscriber cancelling being much higher during the initial months of their subscription. The image below shows the difference between a linear and exponential decay curve, with the Y-axis representing the probability a subscriber is retained at a given point in time (X-axis).
To improve churn rate, the growth of the subscriber balance (denominator) has to outpace the growth of cancels (numerator). This is especially challenging when there is an influx of new subscribers. As a greater portion of the subscriber base shifts towards recently added members — many of whom cancel during their initial months — the growth of cancels accelerates. The image below illustrates this relationship: January was an exceptional month for subscriber growth, but the influx of new members led to an outsized increase in cancels and churn rate.
Formulaic dependency on size / growth of business — evaluating churn for a company over time
Secondly, the denominator of the churn rate formula is dependent upon the product’s audience size and growth trajectory. This presents challenges in two primary use cases: 1) comparing churn rate over time for an individual company and 2) across comparable companies within a particular industry (discussed in next section).
When analyzing churn for an individual company, the expectation should be for churn to decrease over time. However, an improvement in churn rate is not dependent upon an improvement in retention. This means that churn rate can (and often will) decrease despite no change in average customer lifetime. All else equal, this also means that profitability on a per subscriber basis remains the same (i.e. customer lifetime value is stagnant). That churn rate can improve without an underlying improvement in customer lifetime diminishes the purpose of the metric — to evaluate retention over time.
To illustrate this concept, the image below shows metrics for a subscription business during its first five years, with two key assumptions: 1) static annual gross pay adds and 2) static customer decay curve (this is equivalent to no improvement in customer lifetime). Despite no actual improvement in retention, churn rate decreases precipitously in the first few years before a more modest descent in the outer years. This contradiction has the potential to deceive the company into thinking its retention initiatives are working and that they are extending their average customer lifetime, when in fact it is remaining the same.
Formulaic dependency on size / growth of business — comparing churn to competitors
Churn rate is also used to evaluate a company’s ability to retain customers relative to their peer group. As we saw in the example above, the expectation should be for churn rate to decrease over time as the business matures. For that reason, it’s important to consider the growth stage of the businesses when comparing their churn rates.
Churn rate is elevated and more volatile towards the beginning of a product’s lifecycle primarily because a greater portion of subscribers have been recently added. The exponential shape of the customer decay curve tells us that recently acquired subscribers are more likely to cancel. As the business matures, loyal and longer-tenured subscribers — whose probability of cancelling is next to zero — represent an increasingly larger portion, helping stabilize churn rate at lower levels.
The table below shows churn rate for three businesses in different periods of their product life cycle, all of which experience the same number of cancels for a particular period of time. The launch company has a churn rate nearly double that of the more mature incumbent, but has a much more attractive growth rate. More importantly, the launch company added six times as many new paying subscribers and experienced the same cancels as the incumbent. This indicates that the launch company has much better retention rates and a much longer average customer lifetime.
In addition to growth stage, the structure of the relationship between the consumer and product has a significant impact on key metrics. Many traditional subscription businesses, especially cable and telecom providers, have multi-year agreements with hefty penalties for early cancellation, helping produce churn rates around 1.5%. Long-term guaranteed revenue streams and fierce competition have also enabled customer acquisition costs to skyrocket to several hundred dollars per subscriber.
On the contrary, most digital subscription businesses have (and promote) a “cancel anytime” policy for their month-to-month plans. The lack of commitment by the consumer produces a higher churn rate, but it also facilitates customer acquisition, enabling these businesses to achieve scale while paying a fraction of what their traditional counterparts pay. These are structurally different business models, making comparisons of key metrics challenging.
Misused to determine customer lifetime value
A common methodology for calculating customer lifetime is to simply divide 1 by the churn rate. For example, a monthly churn rate of 5% would imply a 20 month average lifetime (1 / 5% = 20 months). The key assumption here is that the probability that a subscriber cancels in a particular month remains constant in perpetuity. This is fundamentally flawed — as noted in the sections above, the probability that a subscriber cancels is much higher during the first few months and decreases exponentially, moving towards zero in the outer months. Determining customer lifetime is the backbone of the customer lifetime value calculation; using an oversimplified methodology to determine customer lifetime amplifies the margin of error, increasing the likelihood of inaccurate results and ill-advised decision making.
Summary
Improving retention is frequently the lever that can drive the greatest gain in customer lifetime value, a top priority for subscription businesses alongside subscriber growth. There is no silver bullet — building relationships with your audience and extending customer lifetime is an ongoing effort and must be a primary initiative across the company.
Prioritizing retention as a core goal has to be accompanied by a strong approach in measuring retention. For churn rate to be an effective metric, it requires an understanding of its limitations. It should be one component of a more holistic approach to measuring retention that includes segmenting cancels (i.e. by subscription duration or engagement), analyzing retention rates, and leveraging user feedback (i.e. net promoter score and exit surveys).
