Fundamentally, churn occurs when a customer stops consuming from a company. But in the ecommerce ecosystem, there is a whole lot more to it than that.
Churn is most commonly discussed as percentage – churn rate, which can be calculated as:
The purpose of churn rate is to indicate the percentage of your customer base that churned over a given time period.
And this conceptualization works great in markets that are contractual or based on subscription.
A service like Netflix, for example, can very easily calculate its monthly churn rate by comparing the number of subscribers at the end of the month to the beginning of the month, subtracting out new subscribers.
However, this calculation is not really applicable to non-contractual ecommerce markets in which shoppers can purchase however much they want, whenever they want.
Our team had the privilege of speaking with Dr. Paul van Loon, the Head of Analytics at Forecast, to gain an industry perspective of customer churn.
Forecast is a management consulting firm that specializes in decision tools, data analytics, and financial modeling.
Paul has extensive experience working with clients to detect and prevent customer churn. He emphasized the complex nature of churn, particularly in ecommerce retail.
In non-contractual ecommerce markets, you will almost never be able to observe the point at which a customer churns. So, a binary classification of churn is not appropriate.
Customer churn is a metric that no one really wants to have but everyone needs to have.
Since churn is the antithesis of retention, it not only affects the size of your customer base, but directly impacts your customer lifetime value.
In addition, understanding churn within your business can help you reinforce your retention efforts by pinpointing where and how customers are falling out.
For years, the primary focus in ecommerce was customer acquisition. This comprises of pumping time and money into ads, search engines, and social media trying to maximize the number of new shoppers that hit your site.
Acquisition is important, particularly for newer brands or retailers looking to grow their shopper base.
But it is not a sustainable way to grow your revenue in the long-term without effective customer retention.
We see across retailers that multi-purchase shoppers consistently contribute more than 2 times the revenue of one-time buyers.
For one brand, multi-purchase shoppers have spent, on average, 4 times as much money as those who only bought once. Multi-purchase shoppers for this retailer also have an average order value 13% greater than one-time buyers.
Yet, multi-purchase shoppers rarely comprise more than one-third of a brand’s total shoppers.
Churn is a critical measure to gauge the health of a business, especially the impact of retention efforts, and is pivotal to grow your ecommerce store.
Does “churn” as it’s understood in a contractual environment even truly exist in a non-contractual environment?
That’s not to say that churn isn’t relevant in ecommerce – our perception of it just needs to be adapted.
Instead of an absolute classification of churn, we need to view it more as a scale, evaluating the health of a shopper.
For this very reason, many ecommerce experts and partners, such as Forecast, are embracing probabilistic approaches to churn. They provide an immensely powerful framework but can seem overwhelmingly complex for the end-user.
Essentially, these models aim to predict the probability that a customer will continue to purchase over time and identify those who may be at risk.
An integral component of this approach is the duration of time between events. An “event” could be any number of actions taken by shoppers, such as site visits, account logins, or purchases, depending on your preference.
And the underlying premise here is pretty logical: if a shopper typically has (or is expected to have) a certain pattern of activity, and begins deviating from that, she may be at risk.
Below, we have listed the number of purchases for a sample of 5 different shoppers, the average number of days between their purchases, and the number of days since their most recent purchase.
While simplified for the example, this will show the power of analyzing churn.
Keep in mind that these are a lot more than just numbers, they are data points summarizing the history and patterns of each of these shoppers. And they are highly indicative of the “health” of the customer relationship.
First, let’s look at Tommy. He has made seven purchases with the brand, and an average of 123.7 days elapsed between those purchases. Tommy most recently purchased 31 days ago.
We can reasonably assume that Tommy is “healthy” for now. The number of days since his last purchase is low compared to his average repurchase time, so at the moment, we can feel comfortable that there is a good chance Tommy will buy again.
Abby, on the other hand, is a completely different story. Abby has made 3 purchases, with an average of 26.5 days between these transactions. But she has not purchased for 324 days.
Given her known purchase behavior, and the time since her most purchase, we would be pretty surprised if Abby were to buy again.
This may seem like pure hindsight – what good is to look at what has happened?
Looking in hindsight allows us to act like detectives. We can identify deviations from expected behavior in the customer journey and make adjustment to future segmentation and campaign themes.
For this purpose, Stacy is an excellent example. Stacy has bought five times from this retailer, an average of 127.8 days transpired between the purchases, and she last bought 103 days ago.
Knowing this, I may want to prioritize Stacy for marketing outreach. Encourage continual top-of-mind awareness, eventually get her back on my website and drive that next purchase.
It’s important to remember that the more observations, the better. We can be much more certain of the purchase patterns of a 5-time buyer than a shopper who has only purchased twice.
The challenge, then, is to do this at scale. Recognize patterns of activity among your base of shoppers and act on it at the customer level.
So next, we’ll explore the journeys and typical behaviors across all shoppers for this retailer.
Here, we have the rate of repurchase graphed over time since purchase for this retailer. This tells us that the base repurchase rate, at time 0, is 27.94% for this brand.
This isn’t entirely fair to shoppers who purchased recently. For example, a shopper who purchased a week ago has not had as much time to buy again as someone who purchased a year ago.
But even looking back at shoppers acquired more than a year ago, the repurchase rate is 35.01%. Just over a third of shoppers who first purchased from this retailer more than a year ago have bought since.
What is more telling, however, is the nosedive in repurchase rate as time transpires. After 30 days, the repurchase rate is just over 12%. And by day 60, it’s under 10%.
In a similar way, we can visualize the cumulative percentage of repurchases that occurred over the time since the previous transaction.
This allows us to answer: “What percentage of repurchases occurred within X days?"
The above graph shows us that, for this retailer, 26.45% of repurchases have occurred within 10 days of the previous transaction. After that, 54.86% of repurchases occurred within 30 days, and 67.43% of repurchases occurred within 60 days.
We can use this exact methodology to flag shoppers when they reach certain thresholds on this progression - maybe 25%, 50%, or 75% - and act accordingly to drive their next purchase.
So far, our approach is assuming that all purchases in the customer journey are equal, that a first-time buyer is as likely to purchase again as a second-time buyer. But as the data will show, many times this simply is not the case.
The greatest decline of shoppers occurs between the first and second purchase in the customer journey, with only 25% of first-time buyers making a second purchase.
After the third purchase, in particular, there is substantially less drop-off from one purchase to the next. By the third purchase, nearly 38% of shoppers go on to make a fourth.
We see a far more complete picture when we combine the concepts of days between purchases and the customer journey.
Not only does the repurchase rate increase along the customer journey, it tends to stay higher even as time passes.
This tells us that the more a shopper has bought, the more likely she is to purchase again, regardless of how recently she last purchased.
In fact, we see that a customer who has made 5 or more purchases with this brand but has not purchased for 28 days, is more likely to repurchase than a first-time shopper who bought today.
But what explains this progression through the customer journey? We spoke with Tim Peckover, who is the Content Marketing Manager of customer loyalty program provider Smile.
Acquisition, as we understand it, does not truly end at the first purchase.
The key to diminishing churn and sustaining profitable customer retention is to do this at scale. Purchase patterns can be used to compose more precise customer segmentation.
Observe typical behavior of individual shoppers and shoppers as a whole, identify deviations and at-risk customers, and strike strategically to repeatedly generate value.
An individual’s “health” as a shopper should inform your approach to the customer relationship.
But what sort of events should you look at in this process? As we mentioned earlier, an event can really be any action you can track: a site visit, account login, transaction, or any other observable activity.
Peckover highlighted that it's best to compile all of these events in order to get a 360° look into the shopper's standing.
At-risk shoppers may need more direct outreach, discounts, and special offers to reel them back into your funnel.
Shoppers who are not at-risk probably do not require these same touch points. Top-of-mind content and branding may be more efficient in this spectrum.
You can inject this analysis directly into your existing outreach strategy, and provide precisely targeted touch points based on a shopper's activity patterns and current status.
Here, we overlay our distribution of repurchases onto the shopper behavior we present above.
And, by simply matching a shopper's recency with the retailer's overall repurchase patterns, we can model the likelihood that a shopper would have repurchased in the amount of time since her most recent transaction.
Essentially, answering “How likely is it that an active shopper would have repurchased by now?"
The higher the percentage, the more anomalous the behavior, and more likely it is that she has churned.
For Stacy, as an example, there is a 76% chance that if she were going to repurchase, she would have done so already, based on this retailer's overall shopper purchase patterns.
This approach can profoundly impact the way retailers identify customers for outreach, and help choose the right messages and offers to drive a longer customer relationship.
The customer relationship is inherently a matter of personal connection.
In-depth shopper analyses are great for making initiatives more effective, but you must never lose sight of the human component.
Customer churn is one of the most intriguing topics in ecommerce: it is easy to understand, but can be very difficult to incorporate into your business strategies.
And while churn as it’s understood in contractual businesses may not be applicable to non-contractual ecommerce retailers, it is nonetheless crucial for these retailers to acknowledge customer churn in order to solidify their retention efforts.
Studying customer churn empowers us to locate pitfalls in the customer journey, identify at-risk customers, and scale a profitable ecommerce business.