Annual Recurring Revenue

Annual Recurring Revenue, or ARR for short, is the revenue a company reliably expects from ongoing subscriptions per year. The metric shows how much money comes in dependably if no customer cancels and no new one is added.

Some companies sell their product once, others rent it out permanently for a monthly or yearly fee. Such ongoing payments are called subscriptions. Annual Recurring Revenue, usually abbreviated as ARR, measures exactly these subscription payments: it is the amount a company expects in a year from regular fees. Only what repeats on a planned basis is counted. One-time revenues, such as for a consulting engagement or a one-off setup fee, are excluded. ARR is therefore not a statement about how much money has actually already flowed in, but about the reliable part of the business.

Why investors look at ARR first

At young software and AI companies, classic annual revenue fluctuates a lot. A large one-time order makes it spike, and the next year it's missing again. ARR filters out these outliers and shows the underlying baseline of the business. Anyone wanting to know whether a company is sustainable gets a more stable answer here.

ARR also allows a rough estimate of the future. A company with 50 million euros in ARR effectively starts the next year already with those 50 million euros behind it. Everything that sales and marketing additionally win comes on top of that. It is precisely this stacking effect that makes subscription businesses so attractive to investors.

That is why investors often value software companies as a multiple of their ARR. In good market phases, companies have been valued at as much as twenty times their ARR, in weaker phases at five times. The metric therefore helps determine how much money a start-up can raise in a funding round. This is also the reason why some companies deliberately calculate their ARR figure favorably.

The math behind the number

In the simplest case, all ongoing annual contracts are added together. Monthly subscriptions are extrapolated: a customer paying 100 euros per month contributes 1,200 euros to ARR. A two-year contract for 2,000 euros, on the other hand, counts only as 1,000 euros per year. The cutoff date is usually the end of a quarter or fiscal year.

Over the course of the year, ARR changes from four directions. New customers bring additional ARR. Existing customers who upgrade to a larger package bring so-called expansion ARR. Anyone who downgrades their package reduces it. And anyone who cancels entirely causes churn, in industry jargon. The ARR at year-end is the starting value plus additions minus losses.

A common misconception: ARR is not the same as revenue or even profit. Revenue on the balance sheet only captures what was actually earned during the period, including one-time deals. ARR, by contrast, is an extrapolated snapshot and is not defined by any accounting standard. Each company decides for itself what it counts, for example whether trial customers or uncertain contracts are included. That's why it's always worth reading the fine print when comparing two companies.

ARR in reports about AI start-ups

The metric appears most often in news about software companies that rent out their product over the internet. This model is called Software as a Service, or SaaS for short. Reports then read something like: an AI start-up reached 100 million dollars in ARR within twelve months. Such statements have become the benchmark for how fast a company is growing.

Outside the tech industry, too, the principle is embedded in many products. A music streaming service, cloud storage, or a gym membership contract generate recurring revenue. The monthly version of the metric is called MRR, Monthly Recurring Revenue, and is simply one-twelfth of the value.

Caution is warranted with very young AI companies. Some extrapolate a single strong month to twelve months and call the result ARR. If customers are only trialing the product and drop out after three months, the number was never reliable. When reading such reports, it is therefore always worth asking the follow-up question of how many customers are still paying after a year.

Subscribe free. Unsubscribe the second it sucks.

High-signal news across AI, business, UX, and tech. Every morning.