Recurring revenue metrics, such as annual and monthly recurring revenue, are used to evaluate the performance of subscription businesses. Changes in recurring revenue over time can be understood by decomposing recurring revenue into its constituent parts (e.g., expansion, contraction, and churn), which are used to construct various metrics for assessing performance. This article describes:
- The standard recurring revenue decomposition
- Accounting for price changes
- Net Recurring Revenue (NRR) and the role of price
- More detailed decompositions
The standard recurring revenue decomposition
The most widespread decomposition of recurring revenue explains its change between two points in time:
$ Change in Recurring Revenue =
+ $ New
+ $ Expansion
- $ Contraction
- $ Churn
$ Change in Recurring Revenue is the difference in recurring revenue between two points in time. For example, if the company's annual recurring revenue is $2M on 31 December 2022 and $3M on 31 December 2023, then $ Change in Recurring Revenue is $1M.
The components of the change are also computed as differences in recurring revenue between the two points in time, where each relates to a different sub-group of customers, where the definitions relate to customers with:
- $ New: recurring revenue at the second time point but not at the first.
- $ Expansion: recurring revenue at both time points, but more at the second.
- $ Contraction: recurring revenue at both time points, but less at the second.
- $ Churn: recurring revenue at the first time point but not at the second.
Comparing the components over time and by other sub-groups (e.g., geography) is typically interesting. They are usually converted to rates to make such comparisons easy to interpret. Most commonly, this is done by dividing by the first time point’s recurring revenue and expressing the result as a percentage.
For example, if the values of the decomposition for a company are: $2M Annual Recurring Revenue on 31 December 2022, $3M Annual Recurring Revenue on 31 December 2023, $0.8M New, $0.7M Expansion, $0.2M Contraction, $0.3M Churn, then:
- The recurring revenue growth rate is (3 – 2) / 2 = 50%. As the comparison is between two points in time a year apart, this is the annual recurring revenue growth rate.
- The new account revenue growth rate (% New) is 0.8 / 2 = 40%.
- % Expansion is 0.7 / 2 = 35%.
- % Contraction is 0.2 / 2 = 10%
- % Churn is 0.3 / 2 = 15%. This churn rate can be referred to as dollar churn and is different from gross churn to avoid confusion with the churn rate calculated based on the number of customers to churn rather than the recurring revenue.
- Recurring revenue growth rate of 50% = 40% New + 35% Expansion - 10% Contraction - 15% Churn.
Other metrics can be calculated using these metrics as inputs. The two most well-known are:
- % Retention = 100% - % Churn
- The Net Retention Rate (NRR) is calculated as 100% + % Expansion - % Contraction - % Churn. In the example above, 100% + 35% - 10% - 25% = 110%. When this metric is above 100%, the company’s annual recurring revenue grows even when % New is 0%.
Accounting for price changes
As described above, a Net Retention Rate of 105% implies that the company’s revenue grew by 5% without any sales to new accounts. This sounds like a good thing. However, if the price was increased by 5% and this 5% matched inflation, the reality is that, in real terms, there is no growth at all.
The above decomposition can be modified to address price changes:
- At the second time point, recurring revenue is computed for each firm using the prices from the first time point.
- $ Price is the difference between the actual recurring revenue at the second price point and the recurring revenue calculated for the second time using the prices from the first. This calculation is performed excluding customers that have churned.
- All the other components - $ Churn, $Expansion, and $ Contraction - are calculated based on the difference between the initial recurring revenue and the recurring revenue for the second point using the prices from the first point.
For example, with the recurring revenue at the second time point being $3M, if there had been a 10% price rise since the first time point, then the recurring revenue at the earlier prices is $2.7M, and $ Price is $0.3M and % Price = 0.3M / 2M = 15%.
The price effect will usually be distributed over time. For example, if a price rise of 10% occurs on 1 April, but most customers have annual licenses and renew in December, the price effect will primarily appear in December and, on an annualized basis, it may exceed the 10%. Further, as the price effect can relate to new sales as well as renewals, the actual magnitude can exceed the price rise even without the time of renewal effect.
Where price rises are primarily undertaken to keep up with inflation, NRR can be calculated without the price effect.
When price rises exceed inflation by a non-trivial amount, the situation is more complicated in a few ways:
- The other components - $ New, $ Expansion, $ Churn, and $ Contraction - may be smaller than they would otherwise be, cannibalized by the increased price. If there is sufficient information for calculating price elasticity, adjustments can be performed, and another category can be added to the decomposition: $ Recurring Revenue Lost Due to Price Elasticity.
- The interpretation of the $ Price itself becomes difficult, as it in part reflects inflation and actual growth in recurring revenue, and it is arguable that whatever proportion is more than inflation should be included in NRR.
Gross churn =
Net Recurring Revenue (NRR) and the role of price
While recurring revenue can be decomposed into price, new, expansion, contract, and churn, the situation is more complicated for NRR, as the price effect contains both new and renewal effects, making the decomposition:
+ % Price effect excluding price rises for new customers
+ % Expansion
- % Contraction
- % Churn
An additional complication when calculating NRR for periods less than a year is that the percentages need to be annualized, and different methods of annualizing are required. This is discussed in more detail in Annualizing Growth Rates, Churn Rates, and Other Rates.
More detailed decompositions
The components of the standard decomposition can themselves be decomposed in various ways. The standard way is to decompose using various firmographic and acquisition characteristics, such as:
- Channel (e.g., from performance marketing, BDR/SDR, SEO, etc.).
- Sales rep/account manager.
Less obvious decompositions are provided below.
$ Price =
+ $ End of Promotional Price
+ $Price Increase
End of Promotional Price addresses the effective price rise that often occurs when companies are given sweeteners on their initial contract (e.g., 3 months free).
This decomposition can be further decomposed using the new revenue and expansion decompositions.
Refer to the Accounting for price changes for other decompositions of the price effect.
Decomposing new revenue
$ New =
+ $ Completely New
+ $ Completely New In-and-Out
- $ Completely New In-and-Out
+ $ Resurrected
+ $ Resurrected In-and-Out
- $ Resurrected In-and-Out
Resurrected refers to customers who had been customers before the first time point but were not customers at the first time.
In-and-Out refers to subscriptions that started and ended between the two points in time. It appears as both a positive and negative in the decomposition as it does not affect the decomposition, but is still commercially relevant (e.g., it will still count towards the sales team's remuneration). In-and-Out is typically only useful when the billing period is less than the time period being examined (e.g., with annual billing, quarterly In-and-Out will typically be 0, whereas with monthly billing it may be significant).
$ Expansion =
+ $ Expansion At Renewal
+ $ Expansion Not at Renewal
It can be useful to distinguish whether the expansion occurs during renewal as a part of a more general account discussion or at some other time point. Other ways of decomposition expansion further include tenure, channel, geography, vertical, and account manager.
$ Churn =
+ $ Churn Will Return
+ $ Churn Yet To Return
Often, companies that churn return in the future (i.e., be resurrected). Sometimes, they are won back, but often, with small companies, the churn was just a way of avoiding paying for licenses when they weren't required (e.g., if the renewal date was on 1 June, and the small business was going close down for June and July, they may choose not to renew on 1 June, and start a new subscription on 1 August).