What counts as ARR
A common question in SaaS—consider frequency, source, timing, and adjustments.
Does this count as ARR? We get this question constantly when helping companies set up their ARR reporting.
To the uninitiated, counting Annual Recurring Revenue probably looks something like this:
Rule #1
Is it recurring? If yes, count it.
Rule #2
Refer back to Rule #1.
However, those of us who have dealt with the edge cases surrounding different pricing models and products in the market, and know exactly when to start and stop counting ARR, understand the situation is more nuanced.
Below are some guiding principles for what we think counts as ARR, and what doesn’t so you can develop the right methodology for your business.
If it walks like ARR, talks like ARR
Of all the variations to consider, you can boil things down into four categories:
Frequency
How often the customer will be billed (e.g. recurring vs. non-recurring).
Source
What the customer is being billed for (e.g. product or service).
Timing
When to start and stop the clock (e.g. trial periods, cancellation requested vs. end of the billing period).
Adjustments
Exceptions and concessions (e.g. discounts, coupons).
Let’s walk through each with some of the most common examples below.
Frequency
The basic question to answer here: is the charge recurring or non-recurring?
For traditional SaaS companies that charge a fixed fee per license, this is an easy yes but things are less clear for usage-based pricing.
The latter has become more popular with the rise of AI and as traditional SaaS companies shift to hybrid pricing models that charge both a flat fee per seat, along with variable pricing based on a product usage metric (e.g. tokens used, API calls, tickets resolved, contacts synced to CRM).
I think the short answer to the usage-based pricing question is yes, this revenue is recurring. Take the example of an AI company that charges its customers based on tokens consumed. In any given month, a customer’s usage on the platform might increase (more or growing workloads) or decrease (deprecating workloads) but the customer must continue to consume tokens each month to get utility.
The harder, but different, question is knowing how much a customer’s usage varies from period to period. I think some conflate the challenge of predictability with the question of whether the revenue is recurring or not. The predictability question feels similar to the challenges PLG businesses face where the majority of customers are on month-to-month plans. However, few folks debate the validity of counting revenue from month-to-month customers as ARR.
This distinction is worth reflecting on. It’s less about should this count as ARR vs. increasing the predictability of the inputs that drive customer usage and, ultimately, revenue.
For more resources on improving predictability, have a look at our posts on building a better ARR reporting system and Snowflake’s forecast transformation.
Source
Revenue source can either be a product or service offering.
Based on what we discussed regarding frequency above, we know that one-time charges of any kind would NOT count as ARR. This applies to both product and service-based charges.
While one-time service charges like onboarding and implementation are more common, some companies offer continuous services for things like premium support.
You can certainly make a case for including these in ARR but we would advise against doing so. The reason is these revenue streams will have very different margin structures which translate to different valuation premiums.
If you do include recurring service-based charges in ARR, just make sure you have a method for breaking out product vs. services ARR and watch out for mix-shift changes. Doing so is important because when it comes time to fundraise, investors will value these revenue streams differently.
Timing
When to count a customer
Knowing when to start counting a customer in your ARR calculation varies based on the purchase experience and go-to-market motion.
On the purchase experience, the biggest question comes down to whether trials are offered. In PLG businesses, two-week free trials are common while enterprise motions may offer a paid trial or “proof of concept.”
In these evaluation periods, we recommend erring on the side of conservatism and not counting the customer until they’re successfully through the trial phase.
For enterprise businesses, this means waiting until the longer-term deal is complete which typically aligns with comp plans design and quota retirement for reps.
As an aside, many companies have a “grace period” for enterprise deals that starts the clock on ARR based on the contract signature date, rather than the contract start date. This is typically limited to deals where the contract start is within ~two weeks of the signature date. Doing so better aligns Sales bookings, or contracted ARR, with Finance’s view of ARR since most deals sign on the last week or day of the month.
For PLG companies, we often recommend adding one more requirement which is not counting the customer in ARR until they’ve made their first payment. This will help reduce the noise from customers who sign up using fake credit cards that never intended to pay.
When to stop counting a customer
The second component of timing is aligning on when to stop counting a customer: when they request to cancel, or at the end of their contract/billing period.
Normally, we recommend erring on the side of conservatism but in this case, either approach will work. Our advice is to align the methodology here to how compensation is designed for your sales or growth team.
If those plans don’t yet exist, we’d recommend waiting until the end of the billing cycle to better align with GAAP revenue.
Adjustments
Because nothing’s ever as simple as setting a price and charging customers for it, it’s important to have a policy for handling adjustments like coupons, discounts, and credits.
This may sound straightforward but drawing the line on items that are one-time or have a finite life is more complicated. Support could issue an invoice credit for a poor customer experience, marketing incentivizes new signups with a three-month coupon code, or sales offers a 12-month discount for a multi-year contract.
Our advice is to keep things simple and adjust ARR for everything in the beginning. This is a great way to see the frequency and impact of concessions to list pricing. At a minimum, we advise adjusting for anything that’s applied for greater than one month.
Bringing it all together
Once you’ve aligned on your methodology, share it with your go-to-market partners.
This will help clarify how their actions will influence ARR recognition and any potential discrepancies to metrics they’re goaled against. We’ll cover how to bridge between those in a future post.
Until then, we hope you found this helpful. Please reach out with any questions!
P.S. For a comprehensive guide on ARR, including how to think about counting customers and making adjustments, see Chapter 4: Defining Your ARR in our book, The Ultimate Guide to ARR. It’s free.