Is your churn rate skewed? Watch out for two big factors.

by Brigitte Hackler - Sr. Financial Analyst

Churn metrics are getting a lot of hype in the subscription world lately. ARR is still the KPI king, but the “leaky bucket” analogy illustrates that you can’t grow a successful business by adding new ARR alone — you have to retain your customer base and generate renewals. The concept of churn is simple at first glance. As you start to dig in, though, and you’ll quickly realize its complexities.

Say an investor asks the common question “What’s your churn?” Are they asking for number of customers churned or dollars churned? Churn rate? Net or gross? Over what time period? An average rate over a trailing twelve months? Should you include month-to-month or just annually contracted customers? What about off-cycle churn? What does this all even mean?

If you’re confused about what to track when it comes to churn, you’re not alone. I’ll define the two big issues that can skew the data, take a look at what to track to avoid those issues, and describe how churn ties in with retention and renewal rates.

Defining Churn

The churn rate, defined broadly, is total losses in a period as a percent of what is available to lose in that period. In SaaS terms, the formula is:

The complicated part? Defining these inputs. There are a lot of factors to consider, and it’s important to understand how these factors influence the meaning of the output churn number. It’s hard to correctly interpret the instruments on your dashboard if you don’t understand what’s feeding the data behind them.

Churn metrics are divided into two camps: number of customers (logos) churned and dollars (ARR or MRR) churned. You should look at both and compare. They’ll generally follow the same trend; however, if you’re losing very few customers that happen to be your biggest customers, the two metrics will diverge. For simplicity in the formulas below, I’ll just use ARR dollars churned.

When defining lost ARR, you should include both ARR dollars from customers that are lost completely, and downgrade ARR dollars. Including just these numbers gives you your Gross Churn Rate. Subtracting out expansion (upsell) ARR yields your Net Churn Rate. You can therefore have a negative net churn rate (a wonderful thing) if your expansion revenue is larger than your lost and downgrade revenue in a period.

Both of these churn rates tell you useful information. Your gross churn tells you how much ARR you’ve acquired and now lost, signaling customer dissatisfaction. Net churn lets you know if you’re making up for these dissatisfied customers by upselling other customers. Gross churn is often used to calculate your customer lifetime value, another key metric.

Two Factors that Can Skew Your Churn

There are two concepts that can potentially misrepresent your churn rate. The first concept is cohort analysis and the idea of comparing apples to apples: the lost ARR dollars in your numerator should only include ARR dollars that are also in the denominator at the beginning of the period. So if you’re calculating churn over the whole year, look at the customers and ARR you had at the beginning of the year, and determine what is lost from that cohort. Including ARR that is both won and lost within the year will unfavorably inflate your churn, as those lost ARR dollars won’t be included in the beginning ARR in the denominator.

The second concept is renewal timing. To calculate your most accurate churn number, you should only look at ARR dollars that are up for renewal over the period of time that you’re considering. Including all ARR dollars in a monthly churn calculation, such as annual contracts that aren’t up for renewal in that month, will understate your churn rate because you’ll be dividing by a bigger number than what was really eligible to be lost in that month.

The simplest churn calculations ignore these issues and are therefore distorted. But from an operations standpoint, if you really want to understand your churn and what’s happening within your business, you should take care to track the right data so that you can adjust for these potential distortions.

Two Data Points to Track:

1. Varying Contract Lengths

To ensure accurate churn data, track your customers and ARR dollars in cohorts by contract length, and calculate churn over the corresponding period of time. Looking at churn metrics by cohort will give you a more accurate picture and will help avoid those two skew problems.

Calculating churn by month for monthly contracted customers will yield a monthly churn rate. You’ll undoubtedly want to compare this to your annual customer churn rates. To annualize your most recent monthly churn rate, use the following formula:

When stating your churn rate, be clear about your time period; a 5% monthly churn rate translates to a 46% annual churn rate, which is less than ideal, as Tomasz Tunguz points out in his post about maximum viable churn rates.

2. Off-Cycle Churn

If you break out churn by contract length, and calculate it over the corresponding period length, you’ll in theory be comparing apples to apples; all of your annual ARR dollars twelve months ago should have been up for renewal in the past year, so comparing what ARR was lost in the year accurately reflects the churn rate. The piece that can throw this off, however, is the dreaded off-cycle churn.

Off-cycle churn happens when you lose a customer before the contract period is up. This could happen because the customer has to break its contract, you can’t deliver on the contract, or you know the customer won’t renew and decide to mark the deal as churned earlier than the renewal date. All of a sudden, you need to start worrying about comparing apples to apples again; if you lose an annual customer mid-contract, be sure to flag it as off-cycle churn so that you can adjust your churn calculation. Lost ARR dollars from off-cycle churn should also be included in the beginning ARR balance in your denominator or not included in lost ARR in the numerator.

Retention and Renewal Rates

Your retention rate is how many customers or dollars you’re keeping in a period, rather than how many you’re losing. Therefore, it should align as the counter to your churn rate (so long as you keep your churn calculations nice and clean).

Renewal and retention rates should in theory be one in the same. I think of renewals as how many customers or dollars were contractually renewed in a period, and retention as how many were not booked as lost.

One issue that impacts both retention and renewal rates is when you are booking your renewals or losses. Say you have a customer up for renewal in May, but you’re still working on closing the contract in June. It is important to develop a policy for defining how to handle these situations. You don’t want to mark the customer as lost if there’s still a good chance of signing a renewal, albeit late. However, this will cause your retention and renewal rates to diverge; you haven’t marked the customer as lost, so they by default are “retained”, yet they haven’t officially renewed. Additionally, if they renew in a period different from their original “renewal period”, you’ll have an extra level of complexity in matching up dollars renewed to dollars eligible to be renewed.

Like churn, renewal and retention rates can be subdivided into net and gross rates. If an account renews at a higher dollar amount, do you include this expansion revenue in your renewal rate? Consider looking at both, by comparing net and gross renewal rates.

Your net renewal rate could be above 100% if upsells associated with renewals are outpacing your churn.

There’s more to churn than meets the eye, and it’s easy to get lost in what-ifs and complications. Dave Kellogg has a good post that provides more details about the issues that can complicate churn calculations. Overall, if you understand what data is going into your calculation, you’ll be better set to interpret it correctly, and hopefully plug that leaky bucket.