Preventive Churn Management – A Primer!
Churn management is the process of identifying customers who are likely to churn from a company and taking necessary actions to retain by churn analysis.
Churn is a simple concept but seems to be complicated. This is because of two reasons: 1) it is a vital instrument to understand if your business is doing well and 2) someone wants to get paid to make it appear complicated.
This article is to simplify the concept of churn. We will do so through the following:
The article will provide you with a comprehensive understanding of churn and how you can make changes to reduce churn in your business.
Note: There are a lot of articles on churn and how to calculate it. They can be complex and look like a calculus exam. In this article, we will keep the formulas to a minimum. The goal is to enable you to grasp the concept and understand its strengths and weaknesses. So that if you want to learn the formulas and design the SQL scripts later, you will be better equipped to do so.
Churn is the number of customers lost by a company over a specific time period. The number is represented in the percentage of total customers. In simple words, the churn rate is the rate at which your customers are canceling their subscriptions. It is the percentage of customers who stop paying you. This can be of different nature depending upon your business kind, such as:
“The little things matter as much, if not more, than the bigger things”—Adrian Swinscoe
Churn helps to access your business. If you have a subscription service or a SaaS company, the monthly number of customers fluctuates constantly. The goal is to increase the number of customers and reduce dropout. Knowledge about churn helps companies fix the leak in the boat! Businesses often prioritize attracting new customers overlooking the necessity to retain existing ones. They neglect the dent made by churn until it seriously damages the profit. Author Fred Reichheld best know for his research and writings on loyalty business model and loyalty marketing writes that “Across a wide range of businesses, customers generate increasing profits each year they stay with a company. In financial services, for example, a 5% increase in customer retention produces more than a 25% increase in profit.
Why? Return customers tend to buy more from a company over time. As they do, your operating costs to serve them decline. What is more, return customers refer others to your company. And they will often pay a premium to continue to do business with you rather than switch to a competitor with whom they are neither familiar nor comfortable”. (Prescriptions for cost-cutting)
Studies indicate 97% of customers who churn do so quietly, without feedback or clues for dropping out. Even if a company acquires new users and those users do not evolve into paying customers, it is futile for the company in terms of profit. Understanding churn is useful for the company to quantify the value of customers, like the appropriate cost to acquire them and develop ideas to increase retention and customers’ lifetime value.
Here are a few benefits of a customer churn analysis:
Churn rate is the easiest way to decipher if a business is doing well or poorly. This rate, however, does not reflect things like total revenue. It mainly indicates the losses over time, that reveal whether the business is healthy, sick or dying.
For SaaS companies, the churn rate is the most important number that they need to pay attention to. If you obtain $1 million as revenue, loss subscription of $5,000 or $10,000 will not affect you much, initially. But if your churn rate is high and gets higher every month, you are headed in the wrong direction and will suffer over time.
The numbers are simple, it is calculated by the number of customers lost in a given period of time divided by the total number of customers that remain. Churn calculation appears complex because these numbers are constantly in motion. In a world where we are accustomed to real-time information, churn is never actually in real-time. It is a record over a period of time.
If you have a business with 10,000 subscribers on the last day of the month and over that month, you lost 500 subscribers, you have a churn rate of .05 or 5%. Note that if you choose the total number of subscribers on day one versus day 30, you are going to get a different number. That will also give you two different churn percentages.
There are three main types of churn rates:
For many businesses, over the course of, say, a month, there are three different types of customers:
Notice that these numbers will not include the “free trial and leave crowd” if the free trial period is shorter than the period you are looking at for calculating churn. In many cases, you will offer a 7-day free trial. Everyone who signed up before the 23rd of the month and decided not to continue will not affect your number. Most of these folks will drop off before you get to the end of the month. They will not be counted as new customers. While they are important, they do not figure significantly in the churn rate calculations. For this reason, you should make sure that your calculations do not include merely everyone with access. You need the number of those who are paying a subscription fee.
New customers who choose not to renew for the second month have not churned yet. If someone signs up on the 5th of the month and decides to cancel on the 25th of the month, they will not show as churned until next month when they have not renewed for next month’s subscription.
Before calculating your churn rate, it is essential to first define what constitutes an actual churn event for your business. For a software as a service (SaaS) company, you define a churn event as when a customer does not renew or cancels their subscription. For businesses that do not have a subscription model, it is a lot more complicated. Here is how Bahador Khalegi explains in his Medium article, What makes predicting customer churn a challenge? According to him, “A customer can interact with an online store at any time. So, what does it mean to say a customer has churned in this scenario?
A workaround solution is to consider a customer as churner if they have had no (purchase) interaction for the last say 30 days”. This approach does not work for customers with “burst behavior”, that is, customers who have sporadic or bursty interactions. For businesses that have customers with “burst behavior”, Kalgei explains, a machine learning modeling system works best.
Churn is not always simple to calculate. Some of the intrinsic challenges in the concept churn are discussed below:
One part of the idea of churn that confounds many business owners is that it is not a real-time calculation. Because the numbers you are using are bygone numbers, you are only looking at hindsight, not a view of what is happening at the present moment.
Because churn has a lag time, it can be unreliable as a predictor. If you have lost 5% this month, you cannot assume that you will lose 5% next month. In fact, based on the changes in the numbers used in the calculations, a loss of 5% this month and 10% next month might actually be the same churn rate. The real issue is that the numbers in any churn prediction are constantly changing. Since the denominator (the total number of customers) changes every month, as does the number of churned customers, the resulting percentage is inherently unreliable. Even if unreliable, nevertheless, it is vital! This is not to say that you should not calculate it, but simply do not consider your churn rate an absolute number for every month.
Later in this article, we will share a formula that can help you find a predictive number for your business.
For the most part, the data that any company has is not in a format that can be used for quality churn calculations. An engineer needs to extract-transform-load (ETL) the data into a new format so that the numbers can perform well.
The best way to handle this is to have someone create the necessary SQL scripts for you. Unless this is your field (and chances are you would not be reading a primer if it was), an expert will probably have a much easier time creating what you need.
If your company has 10 subscribers and you lose one, you have a churn rate of 10%. You have the same rate if you have 1,000 subscribers and lose 100, but the net effect and its meaning are very different. In other words, if your sample size is small, the churn rate will appear skewed and needs to be looked at differently.
Machine learning models work to predict things accurately. If you have a massive non-churner base and very few churners, a machine may interpret the predictions in favor of the majority, the non-churners. This will teach you nothing about the churns and negate information that you might glean from the losses. There are several ways to balance this case, like oversampling the minority or undersampling the majority to help deliver better numbers. To learn more about this situation and what to do, take a look at this article.
There are many common mistakes people make when calculating churn rates, specifically for SaaS businesses.
This is by no means a comprehensive list, but it represents most cases where the churn management is not done well.
“A customer who leaves before you have recouped your cost of acquisition is worse than not signing up a customer at all. The fastest-growing subscription businesses know this, and are focusing their efforts on developing strategies to combat churn”.Karl Stjernstrom, The SaaS Churn Bible
Up to this point, we have been discussing customer or logo churn, that is, the number of people you lose over a period of time. This is a vital number, but it’s not the only churn number you should be looking at. The other number is revenue churn, the amount of revenue lost because of churned customers.
Revenue churn is important because it can tell you a great deal about what type of customers you are losing. For example, if you have a subscription at $25 and another at $300 per month and you lose ten people at $25, you are not much affected. Losing the same ten people at $300 will be a much impact on your profits.
Part of revenue churn is also the margin that different products might be worth. For example, if one of your subscriptions is for a white label product that gives you just a 10% margin and the other product is yours and gives you a 50% margin, the loss of higher-margin customers will be more damaging to profits.
So, churn rate is not exclusively logo churn; it’s also the revenue lost from that churn.
As we said at the beginning, this is a primer. We want to introduce you to the concept of churn and discuss how you can avoid it.
In this section, we will get into some of the different formulae that have been created to calculate this figure. Do not get bogged down in mastering these, right now.
The most basic churn rate formula is the place to start:
Users at the start of the period –
Users at the end of the period
Users at the start of the period
This formula is used to calculate the monthly and annual churn rates. All you need are the numbers for that period. You can even calculate churn for a day, a week, or a quarter.
Everyone wants to have some benchmarks to know what their probable churn rate will be. This lets you have an expected of your churn rate over a given period.
Stephen Noble at Shopify developed this formula. The principle is that every day on which a user does not churn needs to be factored into the equation. This gives a probable monthly churn.
Number of Churns
The user at the start of the period X # days in period +
New users X 0.5 x days in the period)
Days in the period
Probable monthly churn
These two numbers will give you your baseline churn rates. These numbers will help you make the most decisions in churn management planning.
Every manager in subscription companies, including SaaS companies, worries about churn. At the heart of everything they do, is an effort to retain the customers they have and the new ones that they win. Getting new customers is expensive, much more expensive than keeping them. That means increasing profitability requires that you do everything you can to retain your customers.
Cohort analysis a fancy way of saying, “Find out where your customers are coming from and how you are losing them.”
Earlier we mentioned that not all churn is created equal. Losing a bunch of people from your basic package might not have nearly the impact that losing a handful of people from the VIP package.
In cohort analysis, you look at the details of your customers’ activities.
The way to visualize the timeline is to see the days when you acquired the new us and then to track out how long those users kept the app and continued to use is.
Suppose that after one week, you have only retained 5% of the original customers. It is those people that you need to drill down on. See where they came from and what they have been doing in the app.
Let us say you have a basic drawing app that lets users sketch on their mobile devices. They can share their art on social media and invite friends to do art with them.
Do the customers that remained after week share their art? Did they invite friends? When, during that first week, did they take these actions and other major actions?
This tells you, for example, that most of them shared their first drawing on the second day. You will push notifications at the end of the first day reminding users that they should share their work.
This is the most basic use of churn rates when you tie them to activities and customer losses.
There are many other actions that your company can take to reduce churn beyond doing specific actions on specific days.
Originally Published March 5th, 2020, Updated December 29th, 2020
Surojit has over 15 years of experience in quality and implementations. He is a promoter of an extremely light and efficient Agile process to fit business needs. In his prior role as product owner, he built a robust product in a very short span of time.
Churn| 5 MIN READ
Jun 10, 2021
What is a Customer Exit Survey and what is a customer exit interview? Do th...
Churn| 6 MIN READ
Nov 26, 2020
Identifying at-risk customers can help you in saving a fortune before it is...