How to calculate (and reduce) your churn rate

What is churn rate? 

Churn rate, also known as attrition rate, is a business metric that calculates the rate at which customers leave a product over a given period of time. Customer churn is vital to understand the health and stickiness of a business, as a high churn rate causes a high revenue churn rate and hurts your business' bottom line. The higher the churn rate, the more customers your business is losing.

Churn rate formula

The churn rate formula can be calculated as the number of churned divided by the total number of customers where the number of churned customers is how many people have left your service over the period out of the total number of customers you had during the period.

7 reasons why churn rate can be hard to calculate and understand

Calculating churn rate looks pretty straightforward, but (1) how exactly we define those two numbers can greatly affect the output, and (2) business-dependent external factors may wreak havoc on our understanding of what number comes out.

1. Counting customers is complicated

The total number of customers for a period, say, 1 month, isn’t a well-defined concept because that number will change during the month due to new signups and cancellations.

For any given month, you have three kinds of customers:

1. Those that signed up prior to the month. These customers will come up for renewal in the current month.
2. New customers during the month.
3. Newly churned customers during the month.

If, for instance, your group of new signups is large in proportion to the existing customer base, that can distort your churn rate calculation in a number of ways:

• The number you use for the “total number of customers” in the denominator will be much different on day 1 than it is on the last day of the month. This will mean that no matter what number you use for the “total number of customers”, it will either be pretty distant from the day 1 number, the day 30 number, or both.
• New customers typically churn at a higher rate than customers that have stuck around for a bit. That means that if your company is growing, your churn rate will skew higher than it really is.

2. The moment of churn has multiple definitions

People define the moment of churn in two ways:

1. when the subscription ends and renewal doesn’t happen, or
2. at the moment of the cancellation.

We’ve written about how when a customer cancels, they haven’t churned yet. Customers don’t churn until the end of their subscription period arrives and they don’t renew, because they’ve already paid up until the end of their subscription term. If they’ve only canceled, you still have a chance to win them back before their subscription ends.

By that definition, it’s not actually possible for new signups to churn in their first month, and this does away with the issue of how new customer growth can distort the churn rate calculation. New signups wouldn’t be included in churn or the total number of customers for their first month.

Some call the moment of cancellation churn so that they can have the most current churn number possible. The issue is that churn is fundamentally a lagging indicator and it shouldn’t necessarily be looked at in real-time.

3. Sample sizes can be misleading

When we look at a churn rate of, say, 10%, we’re implying that a churn rate of 1/10 is equivalent to a churn of 1,000/10,000.

Early on and under conditions of hyper growth, our calculated churn rate is just as much a product of our small sample size as it is a number that’s representative or predictive of how well our service retains customers. We don’t have much data in terms of the number of cohorts and how the cohorts behave over time.

This alone can cause wild fluctuations that make it difficult to compare churn rates on a monthly basis. Under these conditions, it’s important to recognize the limitations of the inferences you can draw from your churn rate.

4. Time frames might paint different pictures

You may be looking at customer churn rate over a week, month, quarter, or year.

Also, you want your churn rate calculation to be robust with respect to the time frame chosen. You don’t want your calculation to go from generally correct to wildly incorrect when you move from a monthly frame to a quarterly frame.

5. Customer segments churn differently 

You may have a consumer plan and then an enterprise plan. They’re going to have wildly different churn rates, perhaps with a difference as great as 15% churn monthly vs. 0%.

An aggregated number dissolves the differences between your customers, leading you to misunderstand your churn number if you just take it at face value. For instance, growth in a higher churn customer segment could be mistaken for an increased churn rate overall, which could lead you down the wrong path of trying to fix a non-existent churn problem.

6. Seasonality impacts churn rate

If your business varies based on the season, your churn rate may show changes that correspond with the seasonality of your business that might be hard to understand until you’ve gone through several cycles.

7. Consistency in churn calculation methodology

Several other external factors are likely to be business-specific. What’s important for how you calculate churn rate is that you apply the calculation consistently so that you can compare not only from month to month but year to year as you grow.

4 ways to calculate your churn rate

To calculate your churn rate, divide churned customers over a period of time  by the number of customers you had at the start of that period. While overly simplistic, this allows you to focus on churn by cohort and analyze the cause — instead of debating between overly complex methods to analyze churn.

But that’s just the start.

Steve Noble, a data specialist at Shopify, outlined 4 basic ways to calculate churn rate: (1) Simple, (2) Adjusted, (3) Predictive and (4) the method he ultimately settled on. We’ll walk you through these increasingly complex churn rate calculations.

You’ll see the same constants throughout these examples:

• Churned Customers is the number of customers that churned in the time window.
• n is the number of days in your chosen time frame. When calculating over a month, n = 28, 29, 30 or 31.
• Customers is a list of the numbers of customers on any given day i, 1 through n. For example, Customers_1 is the total number of customers you had on the first day of the window.

1. The Simple Way

The simplest way to calculate churn rate is by dividing the total number of churned customers over the period by the number of customers you had on the first day of the specific period.

The good & the bad

The main pro' of the simple version of calculating churn rate is its simplicity. The churn rate formula is easily understandable and quickly calculable. You only need to know 2 quick numbers to figure out your churn rate for the month, and all you need is those two numbers for each month to be able to compare month-to-month churn.

The problem with this simple churn rate calculation is that it has a hard time dealing with significant growth. When you have a lot of growth, both your churn and total customers can go up. If your total number of customers goes up more, your churn rate will go down, even when you have more customers churning out of your product than the previous month.

This isn’t an issue if you’re an established company with a significant customer base and stable growth month on month. But if you’re a new company with substantial new customers each month, this can lead to a strange interpretation where you can lose more at-risk customers per month, but your rate will get better.

Example

Table 1 is an example from the Shopify post illustrating the shortcomings of the Simple Way.

To calculate churn rate, begin with the number of customers at the beginning of August (10,000). In this example, you lose 500 (5%) of these customers, but acquire 5,000 new customers throughout the month, of which 125 (2.5%) churn out. This gives you a churn rate of 6.25% for August.

625 / 10,000 = 0.0625

You're then starting September with 14,375 customers. This month, you see the same behavior, with 5% (719) of existing users churning, 5,000 new customers joining, and 2.5% (125) of those customers churning. Your simple churn rate for September comes in as 5.87%.

844 / 14,375 = 0.0587

Wait, what happened? You’ve seen the same behavior in both months, 5% of existing customers and 2.5% of new customers churning, but the outcome is two completely different churn rates. It looks like your churn rate has gone down, but the underlying behavior has remained the same.

Your high growth has distorted your calculation. In August, 125 churned customers were added to the numerator, but the 5,000 new customers that joined in August didn't get added to the denominator—which means that the churn rate is artificially high. In the following months, the growth is less proportionally to the existing customer count, so the effect lessens.

2. The Adjusted Way

To account for significant monthly growth, we can take the midpoint of the customer count for the month, rather than using its value on the 1st of the month.

Here we’re dividing the number of churned customers by an adjusted average of the number of customers throughout the window.

The good & the bad

This approach manages to deal with the growth issue by normalizing changes in total customers over the time window. Now you have a more stable platform to base your churn rate on, with the time window for your total customers the same as your time window for churn.

However, though this approach to churn rate calculation does deal with the growth issue, it fails to scale with different time windows. Using the same calculation and the same data, you’d get very different answers for daily, weekly, monthly, and quarterly churn.

Example

Building on the data above, we add the figures for October (Table 2).

Now we see the churn rate as the same, even with a different number of customers at the beginning of the month.

August becomes: 625 / 12,187.5 = 0.0513

September becomes: 844 / 16,453 = 0.0513

October is: 1052 / 20,505 = 0.0513

Quarter: 2521 / 16,239.5 = 0.1552

Bingo! Problem solved. We can all go home for tea and medals.

Not quite so fast. The main problem with this approach to churn rate calculation is that it makes assumptions about the data. If you calculate this over 3 months, you come out with a churn rate of 15.52%. Divide this across the 3 months, and you get 5.17%, very close to the individual monthly customer churn rates. So far so good.

But what if you don’t have exactly the same numbers across each month? Let’s make August a bad month for our imaginary B2B SaaS company. This time, it only gets 100 new customers, 2 of which churn out.

The behavior is the same in terms of churn (5% of existing customers and ~2.5% of new customers), and when calculated individually, each month shows the same churn rate of 5.13%.

But when calculated as a quarter, you get a 3-month churn rate of 13.72%, which, when divided across each month, is 4.57% (see Table 3).

August: 502 / 9799 = 0.0513

September: 605 / 11,795.5 = 0.0513

October: 825 / 16,080.5 = 0.0513

Quarter: 1932 / 14,084 = 0.1371

Now our monthly churn rates no longer tally with our quarterly churn rate, even though they use the same data. This is because we’ve changed the time window we’re calculating. This approach assumes that churn is spread evenly within the time period, with a linear distribution. But churn is never this helpful. A good churn rate ratio should expand or contract well with the length of time it measures, and still deliver comparable results.

3. The Predictive Way

Any good churn rate calculation should give some actionable advice. In this example, Shopify has tried to incorporate a predictive element into the equation. They’re trying to determine a weighted average churn rate, so that rate*customers will predict the likely churn rate on any given day.

Inactive Customers is an array of how many customers active on day i are inactive on day i+n, i.e. one month later. If you have 1000 customers on September 1, you then look forward in time to see how many of those 1000 have churned on October 1. You sum that up, then divide by the total customers in September.

The good & the bad

It seems awesome to be able to predict churn. Having a weight that you can multiply with customers to get predicted churn would be great for planning your finances. Who doesn’t want to do that?

Well, you have probably noticed a critical problem with this approach: “...you then look forward in time...”

This requires two months of data to run one month’s calculation. To determine your churn rate for this month, you have to wait until the end of next month. That isn’t good for a metric that is supposed to keep you up-to-date on your company’s success. If you have a number of accounts canceled in September, you won’t have this information until October.

The flip of this is that when you do get to the end of October and have a churn rate, it’s now from a month ago. It’s not current. You can no longer report churn rates to your employees for the prior month, you are instead telling them what happened a month ago.

This approach has all the same problems as rolling metrics, and you know you should stay away from those.

Calculations in SaaS metrics are supposed to take all your data and transform it into easily understandable, actionable numbers. This churn rate calculation makes your numbers more complicated and less actionable.

4. The Shopify Way

Instead of roughly taking the average of the first day and last day of the month as we do with the Adjusted Way, we can take the average of every day in the month to get a more accurate churn rate calculation.

You divide your number churned by the average of your customer count between days 1 and n.

The good & the bad

This deals with the issues that plague the other variations. You can use it in periods of high growth, and it scales nicely across different time windows. You can also use it in a timely manner, getting an up-to-date churn rate.

But there are always going to be variations in your numbers that a single calculation can’t account for: newer customers churning at a higher rate the older customers, differences in cohorts, in plans, in size of accounts. None of these are captured in this churn rate formula, and by using it, companies could have a false sense of security that the number they get each day, week, month, or quarter is the whole story of their churn.

Why you should simplify your churn rate calculation

As Noah Lorang at Basecamp points out, SaaS analytics shouldn’t be rocket science. One of his “three secrets” is to “make it easy.”

When you reduce complexity on your churn rate calculation, you get the following benefits, which can’t be underestimated.

• It’s easily understandable — anyone in your organization can understand that number. This is critical for a key metric. If no one understands your number, they can’t act on it.
• It’s easily comparable — the more complexity you add and the more cases you attempt to account for, the harder it will be to compare your churn rate calculation across different periods. You create consistency by taking the simple and straightforward path.
• It serves as a starting point for deeper analysis — you’re able to easily comprehend what your number accounts for, what it doesn’t, and where you need to dig in to learn more. With more complex calculations, your first step will be reminding yourself how to calculate churn rate.

That’s why we at Paddle use the Simple Way with a monthly time window.

We keep the churn rate formula simple so that you can spend your time taking a deeper dive into the number, analyzing churn by cohort, and so on—not spending it trying to calculate how we arrived at our number. This churn rate calculation method has worked for thousands of our customers, and it can work for your B2B SaaS company (or any other subscription business) as well.

All of your top-line metrics are just headlines. They’re not the story. The story is buried deep within the numbers. You need to be looking in-depth at the how and why of your churn rather than trying to account for every variable within your churn rate calculation.

Your deep dive into the numbers is where you’ll find out about your business, and how you’ll be able to make actionable decisions to improve customer retention rate.

Four extra things to consider when calculating churn

When you do go ahead and dig into the numbers, there are some things to bear in mind.

1) The differences between B2B vs B2C

B2B and B2C companies deal with churn slightly differently, mainly down to the fact their target audiences (and their behaviors) differ:

B2B SaaS businesses often have a far more niche target audience and business model, normally with strong specifications in mind. They typically experience a relatively low churn rate, due to factors such as higher prices, specialized accounting departments or services to deal with B2B churn, and annual subscriptions

The world of B2C is far broader meaning it can be harder to attain (and retain) each customer’s attention, particularly with all the competition in similar markets.  They typically experience a higher churn rate than B2B, and greater competition within their industries.

2) Not all churn is intentional

Sometimes, churn has nothing to do with how happy your customer is with your product or service. 

They might be sat there minding their own business, unknowingly having their card declined. This could be thanks to weaknesses in your payments infrastructure causing poor payment routing, or something as mundane as the customer’s card having expired. 

This is known as involuntary (passive) churn. Unlike voluntary (active) churn, customers who churn involuntarily aren’t looking to cancel their subscription; they would have stayed! So unless you recover them, it’s simply money down the drain. 

It’s so important that you focus on both areas of churn if you want your business to succeed - which is what every business owner wants, right?

3) Pre-churn vs post-churn

You also need to classify your churn strategies by considering how to re-engage users based on pre-churn events and post-churn events. 

Say what? 

Well, ideally you want to catch the customers that are either wanting to cancel or involuntarily cancelling through payment failures in a pre-churn scenario. That way, it’s easier to avoid or at the least reduce the chances of them leaving. 

Whether you butter up your near-voluntary churners with top customer service and support to keep them happy, or whether you make sure your payment acceptance is on fleek for all users, make sure you get on it before the subscription renewal date. 

But, all is not necessarily lost if you’re a bit late in the game and the churn has already happened. You can still shoot your shot with cancellation offers or the opportunity to pause payments for those that need a little push (back) in the right direction. As for involuntary churners, you can look into payment recovery options such as account updaters and payment dunning so you don’t miss out on that revenue. 

4) Churn rates vary by industry, and that’s okay

Know that all of the above depends on the business and service you’re offering. Some businesses will have a higher churn rate than others and be just as successful, and some will have to tackle churn in different ways. 

The ‘average’ churn rate for SaaS businesses is around 5% - and we’d like to stress the word ‘average’ a little more. Av-er-age. 

This can be compared to education services, for example. The average (yes, average) churn rate for this field is just under 10%. That’s down to the fact that the industry is very much seasonal – churn is affected by the school year. While 10% sounds high, in the education industry it’s considered healthy.  

That said, a high churn rate can often be a strong indicator that something needs to change.

Churn rate FAQs

Churn is a tricky subject, but we get many of the same questions on churn rate again and again. Here are some of the most asked questions, and our answers for them. 

Why is churn so important?

Churn is a super important metric for businesses, because it can point you to what’s going right, what’s not-so-right, and what’s very wrong with your business model and its processes.

A low churn rate in a subscription business means customers are happy with the value you’re delivering. A high churn rate, on the other hand, could mean that you’re failing to fulfil the promises you made to your customers.

The faster a customer churns, the less amount of revenue you will have earned from them. If a customer churns quickly, you may find that you spent more on acquiring them in the first place than they ended up spending with you. Not ideal.

What is a good churn rate?

According to our churn rate studies, average churn rates are everywhere from 2% - 8% of MRR. Therefore, a churn rate at the low end (2%) would be considered “good”. By company age, 10+-year-old companies have a 2-4% churn, whereas younger companies range from 4% - 24%.

What is a negative churn rate?

A negative churn rate occurs when added revenue from new customers (expansion revenue) surpasses lost revenue from churned customers. A negative churn rate is usually caused by activities such as = upgrades, service options, add-ons, etc.

Does churn rate affect retention?

Yes, the churn rate is the inverse of retention. When customers are not retained, they churn by default. Understand your customers better using this cohort analysis excel template.

How can I track churn?

Services like ProfitWell Metrics and Paddle Retain help you track churn and deploy means to manage and reduce churn.