Retailers are experiencing unprecedented swings within the market. As shoppers began staying home and many stores were mandated to shut their doors, U.S. fashion apparel and accessories retailers experienced weeks of steep declines. Those were followed by a spot of light in early April however, as online revenue rose 1% at retailers while pure-play e-tailers’ sales rose 20%, according to Emarsys and GoodData.
With these whiplash fluctuations, it’s easy to understand why many types of measurement and baselines have been turned on their head. While churn is no exception, it can still hold powerful insights for retailers and brands that use it correctly.
Churn rates are typically calculated by taking the number of customers lost over a defined period and dividing that by the number of customers you began with. The resulting percentage is the churn rate.
While churn rates can show what happened in the past, a more valuable way to think of churn in our current market climate is by churn modeling – looking at individuals’ historical transactions to understand their “normal” behaviors and cadence. As long as the normal behavior is maintained, churn probability will remain low. A negative deviation from that “normal” pattern causes the churn probability to increase.
As an example, if a customer usually makes a purchase once a week, a churn model should maintain low churn scores for that customer. However, if that customer hasn’t made a purchase in a week their score would go up. If they don’t make a purchase in two weeks the score would continue to rise – and will rise until the customer returns to their more “normal” behavior.
The Value of Measuring Churn
Churn modeling may not seem useful on its face, given the disruption both retailers and consumers have experienced over the past month or more. However, it will be incredibly valuable as a business planning tool as we turn a corner towards recovery.
First, one of the key characteristics of churn modeling is that if a customer begins to display a new pattern of behavior, a good modeling tool will adjust to this new “normal” state. As more data is fed into the model, it should become more accurate in determining what is “normal” and what is “abnormal.”
As marketers gain an understanding of the new normal, they can quickly take action to keep them engaged, just as they did before the coronavirus turned things on their head. Yes, there may be some period of adjustment. But given that retention efforts typically have a much greater return than new acquisition spend, it’s a worthy investment to understand when and whether customers are likely to stop purchasing. With that understanding, marketers can find ways to incentivize customers to stay engaged.
Today’s churn measurements may also serve as an early warning sign of what’s ahead for brands. While traditional measurements may no longer be accurate in showing how X efforts have produced Y results, understanding the scale of churn and the type of customers with the highest churn rates gives some hint as to how consumers are responding to the uncertainty we’re experiencing.
Some of the traditional value of churn measurements still holds true. Even in on uncertain terrain, marketers using advanced and adaptive churn modeling can use risk of churn as a targeting attribute for campaigns. When a customer is identified as at risk, they can then deploy incentives to motivate these customers to stay engaged or make a purchase.
Also, it’s worth considering that global disruption made the past month or so like hitting a reset button. Looking at churn may have been valuable for understanding which past actions led to which result, but that changed in March 2020 thanks to outside factors. Perhaps a new value derived from understanding churn today is as one more signal – an early sign of what may be coming down the pike.