Usage-based revenue is an increasingly popular pricing mechanism. But it poses new challenges compared to the more traditional recurring revenue. Let’s start with an everyday world example: A man walks into a barber shop and gets a $35 cut. What is the resulting ARR for that customers? And can we call it ARR at all?

1. If the man lives in another city and just happened to walk by, it’s probably not ARR. It’s just one-off revenue
2. If the man just moved to the neighborhood and signaled he will come back every month for a cut, then we could call that $35 * 12 = $420 ARR
3. If the man has been coming to the barber shop every month for the last two years, we could also call that $35 * 12 = $420 ARR
4. If the man signed a contract with the barber, in which he promised to pay $420 every year to get his monthly cut, ARR is again $420

In the fourth example, we feel very confident about that $420 because there is a contractual agreement in place (and maybe some of the amount, or its entirety, was prepaid). The situation is a lot less clear in the second and third examples, because the fact that the customer came before, or said he would come again, does not constitute a firm commitment to future revenue. We will explore this dynamic in the rest of the post.

What Is Recurring Revenue?

In the traditional definition of ARR, the meaning of “recurring” was quite clear. A customer would purchase a subscription to a service for a certain period of time, and at the end of that period, they could (and often would) renew their subscription. Thus, the revenue would recur.

The subscription period typically varied. For example, a $50/month SMB product might have a month-to-month contract with monthly billing, while a $50K/month Enterprise product might have a three-year contract with annual billing.

It’s important to note that “recurring” did not mean that a company would receive $1K per month from a $12K annual contract. Instead, “recurring” referred to the fact that the $12K contract had a fixed duration and was up for renewal periodically.

However, I propose an updated view: that $420 of ARR firmly committed through an annual contract (example four) is more valuable than $420 of ARR implied by the customer’s intent (example 2). That’s because the former is fairly “safe”, or “fixed”, whereas the latter is “at risk”, or “variable.” This is where usage-based becomes nuanced.

What Is Usage-Based Revenue?

Let’s first continue with the barber shop example to understand the nuances. Let’s say there is no contract in place between the barber and the customer. For the first six months, the customer comes once per month and pays $35 each time. In the next six months, the customer decides to come every two weeks, alternating a beard trim (costing $20) and a full service of hair cut + beard trim (costing $35). What is the ARR?

The temptation is to say:

$35 per monthly cut * 6 months = $210 for the first six months + 
($20 + $35) for 2 monthly cuts * 6 months = $330  for the other six months 

= $540 ARR

I argue this is wrong because we are trying to apply the traditional notion of ARR to usage-based pricing.

A much better way to estimate ARR is to use trailing averages adjusted for a period of time that covers potential seasonality, because they reflect the variable nature of usage-based pricing:

In the last three months, the customer paid ($20 + $35) * 3 = $165, so we annualize the amount to $165 * 4 = $660 ARR

Forecasting & Metrics Implications

These nuances make forecasting slightly more complicated. For example, when calculating the Payback Period, should we do it off of the “fixed” portion of ARR, or do we consider the “variable” portion too? The simple answer is, we should consider both. But a more sophisticated way is to split their impact.

For example, we could calculate 1) Payback based solely on fixed ARR and 2) Payback based on the combined fixed ARR and variable ARR:

Payback with Fixed ARR only = CAC / (Fixed ARR * Recurring Gross Margin %)

Payback with Combined ARR = CAC / ((Fixed ARR + Variable ARR) * Recurring Gross Margin %)

This gives a range of confidence with the worst-case Payback (fixed) and a base-case Payback (fixed + variable.) The same logic can be applied to all other metrics, particularly to “risk” metrics where splitting “fixed” revenue and “variable” revenue can help create ranges of confidence.

Usage-based pricing is an increasingly popular pricing strategy, for several good reasons we will explore in later posts. However, it requires an update of traditional views on SaaS metrics. This is to ensure metrics reflect the true dynamics of the business and give us the confidence to steer the strategy in the right direction.