This is part one in a series on metrics for subscription apps.
Over the next couple months I’ll be taking a look at different metrics used in subscription based apps, showing how they are computed, and explaining why they are important.
Monthly Recurring Revenue
Monthly recurring revenue, commonly abbreviated as MRR, is the most often cited metric by subscription businesses. It is a benchmark of the size and growth of your subscriber base and your business. It differs from revenue in a few ways.
MRR acts as a de-noising filter, giving a smoother picture of the revenue trend of your business. A smoothed revenue trend is particularly useful for apps that have products with multiple durations: monthly, annual, lifetime, etc. Apple and Google pay you all the revenue from a longer term subscription immediately on purchase, causing large spikes in revenue. These spikes can mask trends caused by shorter term, lower price products. MRR helps clarify those trends.
MRR is the total revenue from active subscriptions normalized to a standard one-month period according to each subscription’s duration.
There is no GAAP or ISO standard for MRR so, it is common for different businesses to compute it differently. There are, however, two important constraints when calculating MRR: the computation should be month independent and it should sum to actual revenue. What does this mean?
Months differ in length. To ensure that the MRR in February isn’t 10% less than in September purely based on month length, the “One Month” duration in the above equation should be standardized. On RevenueCat we use a 28-day duration. The choice is somewhat arbitrary because your interest is in the trend in MRR, not the MRR itself. (Note: Be careful when converting MRR to Annual Recurring Revenue (ARR). Multiplying by 12 is not correct because 12 x 28 does not represent a full year, dividing first by 28 then multiplying by 365 is a more accurate measure of ARR)
MRR should also sum to equal revenue over a sufficiently long time-scale. To ensure this constraint is met, revenue must not be double counted. An example of that would be normalizing the price of a product by an incorrect duration. You always want to normalize by the actual length of the subscription, not a standardized approximation (i.e. 3 months as 90-days).
MRR as a Daily Metric
Because MRR is normalized to a one month period, it is usually presented in monthly periods: MRR for January, February, etc. But there is no reason for this. MRR is based only on current active subscriptions and their durations. The number of active subscribers changes day to day and can be graphed as such, giving you a good overview of the trend of the business.
MRR Momentum and Churn
The MRR on its own isn’t that interesting unless you are bragging to investors and want to give a sense of scale for your business. What is more interesting is the trend.
There are two components that determine the MRR trend: subscribers gained and subscribers lost. This fact may seem obvious but, if you want to keep your MRR trend positive, you need add more subscribers than you lose.
For a given churn and user acquisition rate, MRR will grow until the churn matches the number of subscribers added for a given period. At this point the MRR reaches equilibrium. Unless you have negative churn (few businesses do), you will hit this MRR plateau if they do not continue to increase new subscribers or decrease churn. You can play with the RevenueCat Revenue Model Calculator to get a sense of this relationship.
Generating MRR Reports
To generate an MRR report, you will need a list of transactions, their price, and the duration, purchase dates, and expirations. You can rely on Apple or Google financial reports, but they don’t include the exact expiration date so you will introduce errors based on calendar date renewals. This error may not be significant enough to care about for you.
If you want real-time computation of MRR with ability to slice it by cohort, product, and other dimensions, you might consider using RevenueCat.