The Ambiguity of ARR
ARR is one of the most commonly cited metrics in Startupland.
Annual Recurring Revenue (the OG meaning of ARR) is usually calculated by multiplying Monthly Recurring Revenue by 12. The logic is to give an extrapolation of the latest numbers, as fast-growing companies will have much higher ARR (latest month x 12) than LTM (“last twelve months”) revenue.
Originally it was mainly B2B startups who spoke about ARR. It was a way to strip out the noise of one-off income streams like implementation fees and professional services, and get to the nub of the core, high margin, recurring software income. With business customers typically less fickle than consumers, the middle R - “recurring” - was a relatively good bet as you could reasonably expect long term contracts, often with high renewal rates.
Recently, though, in the fight to “build in public” with impressive ARR charts showing up-and-to-the-right growth on LinkedIn posts, I am seeing the acronym increasingly misused in a few different ways.
First, I would question whether it’s right to call a lot of this revenue “recurring” - particularly in a pro/consumer context with minimal contractual commitment. Is it annual “recurring” revenue if churn is 10% each month? What about if you sell 1-year contracts and haven’t yet had a churn opportunity (i.e. you’re less than 12 months post-launch) to understand long-term behaviour? Is your recurring revenue actually just experimental transactional stuff that has no hope of sticking around?
Second, I think we need to dust off our understanding of “revenue”. Revenue is recognised over the period the product or service is delivered. If someone buys an annual subscription for £120 and pays up front in January, your cash income in January may be £120, but your revenue is only £120 / 12 = £10. This may seem like accounting semantics but it really matters if you’re annualising from monthly numbers. What we often see is the full £120 being included in a monthly figure that then gets multiplied by 12. So that customer signing up in January all of a sudden shows up as £120 x 12 = £1,440 of ARR, which is obviously wrong - by a mile.
Third, and linked to the above, is the trick of leaning into the ambiguity of acronyms to replace “annual recurring revenue” with “annualised run-rate revenue”, or some similar jiggerypokery. It’s important to point out that this is a completely different metric that is entirely incomparable to true ARR! The problem gets even worse when applying the calculation blindly. A few different variations lead to big errors and nonsensical numbers:
- Confusing cash income with revenue (covered above).
- Companies with seasonal purchasing patterns. For example, annualising a run-rate revenue number from November for a gift wrapping service is obviously stupid as there is massive seasonality in the revenue profile.
- Companies with transactional, rather than recurring, purchasing behaviour. Back in the day all the D2C mattress companies got excited about run-rate revenues, until they realised people only buy a mattress once every ten years…
The irony of all of this is that the founders often know the game they’re playing. Nobody could run a business this way because pretty quickly you’d get punched in the face by the lack of drop-through from ARR to actual revenue. But it’s all a song-and-dance to flex to investors, peers, competitors and other commentators online.
FWIW, the investors you actually want on your cap table see through this nonsense in an instant, so I’d advocate being more up front and transparent about where the cracks are showing. I’d rather have a self-aware founder with a nuanced understanding of their business, its problems and how to fix them, than a hype (wo)man focused on winning the LinkedIn pantomime.