Because this is a non-GAAP metric, there are a number of methodologies used by public companies to report retention. Five9, which makes call center software, uses an invoicing-based measure. To arrive at an “Annual Dollar Based Retention Rate,” they divide their current monthly invoicing by their invoicing from the same group of clients from a year prior, then they calculate a 12 month average. Or: 12 month average of (Recurring Invoicing in current month) / (Recurring Invoicing from same month one year ago).
Hubspot, which makes demand generation software, uses MRR. They divide the aggregate contractual monthly recurring revenue of their customer base at the end of each month by the aggregate CMRR of the same group of customers at the beginning of that month, and then arrive at a twelve month weighted average. Or: (12 month weighted average of (CMRR at end of month) / (CMRR at beginning of month)).
12 Marketo, which similarly makes marketing automation engagement software, looks at aggregate GAAP revenue. They compare the aggregate monthly subscription revenue of their customer base in the last month of the prior year fiscal quarter, to the aggregate monthly subscription revenue generated from the same group in the last month of the current quarter. Then they calculate a weighted average “Subscription Dollar Retention Rate” of the four fiscal quarters within the year. Or: 4 quarter weighted average of (GAAP subscription revenue from last month of quarter) / (GAAP subscription revenue from last month of same quarter one year ago).
In spite of differences in the underlying metric and approach to crafting this ratio, they have a lot in common.
- Look at a group of customers that were present at t = n and understand how much recurring revenue you were receiving from them
- Look at the same group of customers at t = n+1 and determine how much the recurring revenue from that group of customers has increased or decreased.
Doing this on a dollar-basis with upsells included is particularly powerful. This takes into account both your effectiveness in reducing churn and your effectiveness in growing your customer relationships over time. As a result, it translates directly into the expected lifetime value of each new customer you acquire.
For example, in the simplified lifetime value model below, GC is yearly gross contribution, d is the discount rate, and r is the yearly net dollar retention.
Lifetime Value = GC * (r / (1 + d – r))
I prefer methods involving ARR (annual recurring revenue) or MRR (monthly recurring revenue) as the base metric. You would measure ARR net retention using the formula below.
Net ARR retention = ARR t = n + 1 / ARR t = n
Since these point-in-time metrics reflect changes to subscriptions immediately, they provide a better leading indicator of net retention. If a customer churns near the end of a period, the entire impact is reflected in the impact on ARR, while the impact on GAAP revenue for that period is prorated and therefore significantly smaller. The full impact on GAAP revenue would not appear until the following period. That said, regardless of how a company calculates net retention, what is crucial is the underlying performance itself. Below we mine some of the publicly available data in order to explore how widely performance ranges and how important retention is.