Understanding the Basic Metrics of Subscription Monetization (Part 1)

MRR, Net MRR Movement, ARPPU

This series is based on actual subscription monetization cases from Alarmy, a service with 2 million DAU.

If you're familiar with app ad monetization, you're likely used to metrics like CPM, Fill rate, and ARPU. To maximize our app's ad revenue, we essentially have to ensure ads aren't empty and are filling properly, while also tracking the unit price per impression. Furthermore, rather than looking at total revenue, it's crucial to look at the Average Revenue Per User (ARPU) to accurately grasp the current situation. These are the most fundamental core metrics in ad monetization, and we optimize key variables by examining secondary metrics strongly correlated with them.

So, what about app subscription monetization?

As discussed in my previous post, 'Making Money with Everyday Apps — Ad Monetization and Subscription Monetization', subscription monetization is very different from ad monetization. Unlike ad monetization—which requires taking the best options from a flood of information and reacting sensitively to constant changes—subscription monetization is a method that demands focusing internally on our own product. Additionally, as an app monetization method, it hasn't been adopted by many app services yet, meaning market data is nowhere near as abundant as it is for ads, and it can still feel like an awkward approach for users. We have to increase revenue carefully, yet quickly, by optimizing internal variables and managing key metrics through our own in-product experiments. Over a total of four posts, including this one, I'll be taking a detailed look at these key metrics and internal variables.

The 'Basic Subscription Metrics' series will span four parts, structured as follows:

1) Explaining MRR, Net MRR Movement, and ARPPU, the core metrics of app subscription monetization. (This post)
2) Explaining Free Trial, a metric that impacts the core metrics above.
3) Explaining Trial to Paid, a metric that impacts the core metrics above.
3) Explaining Churn, a metric that impacts the core metrics above.

[Core Metrics]

01. MRR (Monthly Recurring Revenue)

This refers to the revenue generated by subscriptions. Since the goal of monetization is to make money, MRR is ultimately the metric that needs to be managed. The reason we call this MRR rather than just Revenue is due to the nature of subscription products. Since the subscription 'period' can vary wildly between products, we convert this to a "Monthly" basis to evaluate it. For instance, if an annual subscription generates 36 USD in revenue, the actual amount deposited that month (minus store fees) is the price of the annual subscription. However, because this is revenue for 12 months, we don't recognize it entirely as monthly revenue; instead, we divide it by 12, managing only 3 USD as revenue to ensure accurate revenue management. This concept of converting all generated subscription revenue into a monthly unit is called MRR, and it's a metric you absolutely must track.

02. Net MRR Movement

Because subscription products—much like a gym membership—guarantee revenue for a specific period once paid, it can easily give the impression that 'revenue only needs to be checked once a month.' But that's not true. The act of new users starting a subscription, or existing users requesting a refund, doesn't just happen once a month. It's happening every single day, right at this very moment, which is why MRR, like other metrics, must be monitored daily. At the same time, because subscription products guarantee revenue over a certain period, the fluctuation of subscribed users is relatively small. In other words, the change in MRR (acquisition of new subscribers, churn of existing subscribers) is quite minuscule compared to the total subscription revenue. Therefore, rather than managing the total MRR, it is common to manage only the change in MRR separately, which is called Net MRR Movement.

Let's say an app service has 100,000 monthly subscribers paying 2,000 KRW a month, generating a total MRR of 200 million KRW. This week, 3,000 new users started subscribing, and 1,000 users canceled, making the MRR 204 million KRW. Similarly, the following week it becomes 203 million KRW, and the week after that, 206 million KRW. If you monitor MRR just as MRR in this case, it’s easy to dismiss these changes as minor fluctuations of around 0.05%.

When you look at Net MRR Movement, your perspective changes. +4 million KRW, -1 million KRW, +3 million KRW... The scale of the fluctuations is clearly visible, and doing a deep dive into their causes becomes much easier. The key event variables that impact this Net MRR Movement are as follows:

1) [+] New subscribed users acquired
2) [+] Users who previously canceled their subscription resubscribe
3) [+] Upgrade from an existing subscription to a more expensive one
4) [+] Addition of another subscription to an existing subscription
5) [-] Existing subscribed users churn
6) [-] Downgrade from an existing subscription to a cheaper one
7) [-] Cancellation of some of multiple existing subscriptions, etc.
…

Naturally, revenue metrics like the above are analyzed by OS, by country, and by subscription product. Since each store (Play Store, App Store) provides a payment module and a subscription product management system, you can track various metrics right there. However, because earlier funnel events (such as "purchase screen visit rate") aren't provided by the stores, it's difficult to find clear and sharp seeds for improving subscription revenue using only in-store data. Thus, it's standard practice to use a separate analytics solution to look at 'revenue data' and 'various in-app event data' in a multifaceted way. I'll cover this in more detail when we talk about granular metrics in the next post.

03. ARPPU (Average Revenue Per Paid User)

Meaning Average Revenue Per Paid User rather than general DAU (Daily Active User), ARPPU is a similar concept to ARPU in ad monetization, but its importance as a monitoring metric is lower than that of ARPU. Because ad revenue is heavily influenced by the ad market, the range and frequency of ARPU fluctuations due to various variables are quite large. On the other hand, for subscription revenue, the 'price per subscription product' is effectively the ARPPU. Since that price doesn't change often—and the entity changing it is the app service provider itself—it falls into the category of a highly stable metric. Therefore, it's hard to call it a metric you need to monitor daily; it's better to just check in occasionally to ensure there are no issues.

Nevertheless, because ARPPU represents the unit productivity of our subscription products, we must make efforts to find subscription product attributes that can generate a higher ARPPU. We pinpoint the right attribute values through various experiments, such as figuring out whether annual or monthly subscriptions yield a higher ARPPU for our service, or whether raising each price ultimately increases or decreases MRR. These 'pricing tests' and 'annual/monthly tests' are experiments that many subscription services run regularly. The pricing tests and annual/monthly tests we ran at Alarmy were also quite interesting, so I'll cover them as a separate topic next time.

The MRR, Net MRR Movement, and ARPPU explained so far are the very last metrics in the entire subscription monetization funnel. As the most fundamental metrics, you've essentially acquired the necessary information to understand and assess the current state—how much current revenue is fluctuating and whether the unit price is at an acceptable level. Now it's time to take it a step further and look at the preceding events that influence those revenue fluctuations, along with the variables that impact them.

Acquisition of New Users and Churn of Existing Users

I briefly mentioned it in the Net MRR Movement section above, but the most important things among those are the 'acquisition of new subscribed users' and the 'churn of existing subscribed users,' and you really only need a solid understanding of these two. One is a plus (+) and the other is a minus (-), so you might wonder if there's anything special to understand, but the reality is quite different.

To generate a single subscription, various user onboarding flows, nudge messages, and promotions are executed all over the place, whether users realize it or not. Including the main onboarding when the app is first launched, the strategically placed entry points, and the attractively designed purchase screens, each functions as a crucial funnel. Among them, the tactics of 'free trial' and 'auto-renewal after n days' are already established as major funnels, with Free Trial CVR (free trial conversion rate) and Trial to Paid CVR (purchase conversion rate) being the metrics that indicate the health of those funnels. We'll dive deeper into these details, along with other specifics, in parts (2) and (3) of this series.

Churn isn't simple either. First, unlike the churn of typical goods, a subscription churn is not a 'refund' concept. If I had to spell it out, it means: 'I can't do anything about the payments made so far, but I want to cancel so I don't get charged from the next billing cycle onwards.' Therefore, depending on how long the subscription period is (monthly, 3 months, annual, etc.), the variance in the churn rate itself is massive, but more importantly, the impact of that churn rate on revenue differs. For example, even if an annual product churns in the middle, they've already paid for a year, so it doesn't take as big of a toll on total MRR as the churn of a monthly product. (This is also the secret to how an empty gym stays in business without going bankrupt.) In short, you have to be careful because churn becomes a meaningless metric if you don't look at it segmented by subscription product. Also, from a revenue perspective, 'retention period' (how long a subscription was maintained after starting) is critical for subscriptions. So when managing the churn rate, it must be managed on a Cohorted basis. For instance, grouping user cohorts by month to manage metrics like average M3 churn rate (= the churn rate in the third month after subscribing) and M6 churn rate (= the churn rate in the sixth month after subscribing). Beyond this, 'refunds' must also be managed separately from 'churn'... Phew... It's complicated. For 'churn', which cannot be easily dismissed as a simple minus (-) action, let's explore it in more detail in part (4)!

The 'Basic Subscription Metrics' series will span four parts, structured as follows:

1) Explaining MRR, Net MRR Movement, and ARPPU, the core metrics of app subscription monetization. (This post)
2) Explaining Free Trial, a metric that impacts the core metrics above. (Next post)
3) Explaining Trial to Paid, a metric that impacts the core metrics above.
3) Explaining Churn, a metric that impacts the core metrics above.

mail : stephan@delightroom.com
instagram : @seosinghan