Annual Recurring Revenue

Run rate, not trailing revenue

ARR is a forward-looking statement made from a single day of data. It takes the contracted book as it stands right now and asks what a year of it would be worth if nothing moved. For a company growing 8% a month, ARR will always sit far above trailing twelve-month revenue — not because anyone is being dishonest, but because the two questions are different. Trouble starts only when a deck reports ARR under a heading that implies history.

The corollary is that ARR is as volatile as the day it was measured on. Land one large annual contract on 30 June and ARR jumps by twelve times its monthly value on 1 July. That is arithmetically correct and analytically fragile, which is why ARR should be read next to a trend of MRR rather than on its own.

Why ARR = MRR × 12 rather than the reverse

Some businesses, particularly enterprise ones, contract annually and think in annual terms, and it is tempting to compute ARR directly from contracts and derive MRR by dividing. The result is the same when every contract is a clean twelve months. It stops being the same the moment you have multi-year contracts, mid-term upgrades, quarterly plans or ramped pricing — all of which normalise cleanly to a monthly figure and messily to an annual one. Computing MRR first and annualising keeps one definition, one movement ledger and one comparable series. It is also what annual contract value is for: contract-level annual economics belong in ACV, not in a redefined ARR.

Only recurring revenue is recurring

ARR inherits every inclusion decision from MRR, and inherits it multiplied by twelve. A one-off implementation fee wrongly included in MRR costs you a small distortion; the same error in ARR misstates the headline by twelve times the fee. The categories that must stay out are the familiar ones: setup and implementation, professional services, hardware, and variable usage. Services revenue in particular is often significant and often lumpy, and rolling it into ARR is how a business ends up reporting a run rate that it would have to repeat a sales cycle to sustain.

Where ARR is genuinely the better unit

Annual-contract businesses have a real case for leading with ARR. When customers commit for twelve months, the contracted book is the year, monthly movement is mostly an artefact of renewal timing, and the annual figure is the one both sides of the table negotiate in. Investors in that segment benchmark in ARR, so reporting it is a communication decision as much as an analytical one.

Self-serve and month-to-month businesses have the opposite situation. Their customers can leave in thirty days, so multiplying by twelve asserts a commitment nobody made. The figure is still usable as shorthand, but MRR with its movement breakdown is the honest primary and ARR is the summary.

ARR and the growth rate people actually mean

When a company says it "grew 3x last year", it is almost always comparing ARR at two points in time, not comparing two years of billed revenue. Those produce different numbers, and the ARR version is the larger one for any growing company. It is a legitimate measure — period-end run rate against period-end run rate is exactly how growth rate is defined — but state which one you mean, because the difference between run-rate growth and trailing revenue growth is the difference most commonly exploited in fundraising materials.