ARR waterfall: how to build one and what it's actually telling you.
A board member asks for the ARR waterfall. The number moved, and nobody in the room can say why.
A board member asks for the ARR waterfall. Or an investor update is due and "just show total ARR" isn't going to cut it this quarter, because the number moved and nobody in the room can say why. This is usually the moment a RevOps or finance lead realizes they don't have a clean answer, because getting one means stitching together billing exports, CRM notes on downgrades, and a spreadsheet that someone built eighteen months ago and half-updates each quarter.
Quick answer
An ARR waterfall breaks the change in annual recurring revenue between two periods into its component movements: new ARR, expansion ARR, contraction ARR, churned ARR, and reactivation ARR, reconciling a beginning balance to an ending balance. Instead of one growth percentage, it shows where the change actually came from, new sales, existing customers paying more, existing customers paying less, or existing customers leaving, which is the level of detail boards, investors, and finance teams actually need to evaluate the health of recurring revenue, not just its size.
What an ARR waterfall actually shows
Annual recurring revenue on its own is a single number. It tells you where you ended up, not how you got there. Two companies can both report $2M in ARR growth this year and be in completely different positions: one grew it almost entirely from new logos while quietly losing 15% of existing accounts to churn, the other grew it mostly from expansion inside an existing base that barely lost anyone. Same headline number, very different business.
The waterfall exists to separate those two stories. It takes beginning ARR, adds every reason ARR went up, subtracts every reason it went down, and reconciles to ending ARR. Because it uses ARR, which is simply monthly recurring revenue multiplied by twelve, the same movement framework applies whether a team reports in MRR or ARR terms; SaaS finance teams typically build the waterfall at the ARR level for board reporting because it's the annualized figure investors ask for.
The five movements, and where each one comes from
ChartMogul's breakdown of MRR movements, one of the more widely cited frameworks in SaaS finance, groups recurring revenue changes into five core categories, with a sixth ("neutral") for changes that don't affect the total. The same five apply directly to an ARR waterfall:
New ARR
Revenue from customers who weren't paying you before this period. This is the most visible line in the waterfall and the one most dashboards already track well on their own, since it maps cleanly to new deals closing.
Expansion ARR
Additional revenue from existing customers: a seat increase, an upgrade to a higher tier, a new module or add-on. ChartMogul's own research found that expansion accounts for roughly a third of all revenue gained once a company passes $1M in ARR, which is the reason mature SaaS businesses track this line separately instead of folding it into "new business."
Contraction ARR
The mirror image of expansion: a downgrade, a seat reduction, a discount applied to an existing account, or dropping a paid add-on. Contraction is a decrease in ARR from an account that is still active, which is what separates it from churn.
Churned ARR
Revenue lost because a customer canceled their last remaining subscription entirely. Churn and contraction get reported as one blended "loss" number more often than they should. They usually have different causes and different fixes: contraction is often a pricing or usage-fit problem, churn is often a product-fit or onboarding problem.
Reactivation ARR
Revenue from a customer who had previously churned and came back. It's typically the smallest line in the waterfall, but it's a real signal, a reactivated customer already knows the product and chose to return.
What actually counts as new ARR
One place waterfalls quietly break is disagreement over timing: is a multi-year deal counted as new ARR the moment it's signed, or spread across the periods it's actually billed and delivered? The same question applies to a large annual contract booked in month one but recognized monthly. Getting this wrong doesn't just misstate one quarter, it distorts every quarter after it, because the waterfall's beginning balance for period two is the ending balance from period one. This is the same underlying tension covered in the difference between bookings, billings, and revenue: a waterfall built on bookings data will show different movements, in different periods, than one built on recognized recurring revenue, and a board deck that mixes the two without saying so will show numbers that don't reconcile from one quarter's deck to the next.
Where mid-cycle changes actually land
Upgrades, downgrades, and discounts that happen mid-billing-cycle are where waterfalls most often disagree with each other, because the accounting treatment depends on how proration is handled. An upgrade with an instant prorated charge lands as expansion immediately; the same upgrade billed at the next cycle instead shows up a month later. Getting this consistent matters more than getting it "right" in some abstract sense, since a waterfall that changes its own rules quarter to quarter isn't reconciling to anything. This is one of the reasons proration and mid-cycle plan changes need a single, consistently applied rule rather than a judgment call made differently by whoever happens to be closing the books that month.
A simplified example
To make this concrete, here's a simplified illustration, not a real company's figures, of how the math works for one quarter:
| Movement | Amount |
|---|---|
| Beginning ARR | $2,000,000 |
| + New ARR | $250,000 |
| + Expansion ARR | $180,000 |
| − Contraction ARR | $(60,000) |
| − Churned ARR | $(140,000) |
| + Reactivation ARR | $20,000 |
| Ending ARR | $2,250,000 |
Net ARR growth for the quarter is $250,000, a 12.5% quarterly increase. But reading only the ending number would miss that gross new bookings ($250,000 + $180,000 + $20,000 = $450,000) were nearly offset by $200,000 in losses. A board looking only at "ARR grew 12.5%" sees a healthy quarter. A board looking at the waterfall sees a company that is still growing, but is losing close to half of what it brings in, which is a very different conversation about where to focus.
Why most ARR waterfalls don't reconcile
The most common reason a waterfall doesn't tie out cleanly isn't a math error, it's that the underlying data lives in three or four disconnected systems: a billing tool that knows what was charged, a CRM that knows what was sold and when a downgrade was requested, and a spreadsheet someone uses to manually reconcile the two once a quarter. Each system has its own definition of "when" something happened, and each round of manual reconciliation is a new opportunity to misclassify a contraction as churn, or a delayed invoice as new business landing in the wrong month.
The practical fix isn't a better spreadsheet template, it's having the billing event that actually triggers a movement (an upgrade, a cancellation, a plan change) generate that movement automatically, in the system that also tracks the customer relationship, rather than reconstructing it after the fact from exports. Revvly's ARR and MRR tracking dashboards classify these movements from live billing and subscription data as they happen, and the underlying breakdown can be exported to CSV or Excel for board reporting rather than rebuilt by hand each quarter.
What to watch for once the waterfall is visible
A few patterns are worth checking every time the waterfall is built, not just when something looks wrong:
- Is contraction ARR rising as a share of losses even if churned ARR looks stable? Contraction often precedes churn by a quarter or two, an account that just downgraded is a flight risk before it becomes a cancellation.
- Is new ARR growing while expansion ARR stays flat or shrinks? That can mean an existing customer base isn't finding enough reason to buy more, which is worth investigating before it becomes a retention problem.
- Does net new ARR (new + expansion + reactivation, minus contraction and churn) track differently than headline ARR growth reported elsewhere in the business? A gap between the two usually means a timing or classification issue somewhere in how movements are being logged.
- How does gross retention (ending ARR from existing customers only, divided by beginning ARR) compare to benchmarks? A waterfall whose gross retention sits well below the 91% median (top performers at 100%) reported by SaaS Capital's 2026 survey of more than 1,000 private B2B SaaS companies is worth a closer look at the contraction and churn lines specifically, not just the net number.
Managing revenue visibility without a unified system
None of this is really about the waterfall as a chart. It's about whether a team can answer "why did the number move" on demand, in a board meeting, without a scramble beforehand. That answer depends entirely on whether new ARR, expansion, contraction, churn, and reactivation are each tracked as they happen or reconstructed afterward from exports that don't always agree with each other. The same underlying breakdown also feeds directly into net revenue retention, since NRR is just the expansion, contraction, and churn lines of the waterfall expressed as a single retention percentage instead of a set of movements.
Frequently asked questions
What is an ARR waterfall?
An ARR waterfall is a breakdown of the change in annual recurring revenue between two periods into its component movements, new, expansion, contraction, churn, and reactivation, reconciling a beginning ARR balance to an ending balance.
How do you calculate net new ARR?
Net new ARR equals new ARR plus expansion ARR plus reactivation ARR, minus contraction ARR and churned ARR, for the period being measured.
What's the difference between contraction and churn?
Contraction is a decrease in ARR from a customer who is still active, such as a downgrade or seat reduction. Churn is the loss of ARR from a customer who cancels their last remaining subscription entirely and is no longer an active customer.
Why do boards and investors ask for an ARR waterfall instead of just the ARR growth rate?
A single growth rate can hide the underlying mix, a company can show healthy net growth while losing a large share of existing accounts to churn or contraction, offset by strong new sales. The waterfall separates those movements so the health of the existing customer base can be evaluated independently from new customer acquisition.
Can Stripe alone produce an ARR waterfall?
Stripe records billing and payment events but doesn't natively classify them into new, expansion, contraction, churn, and reactivation categories the way a subscription analytics or revenue platform does; most teams using Stripe alone build the waterfall manually by exporting billing data and reconciling it against CRM records.
How often should an ARR waterfall be built?
Most SaaS finance teams build it monthly for internal tracking and quarterly for board and investor reporting, since a monthly cadence catches classification issues before they compound across several quarters.
Does a multi-year contract count as new ARR all at once?
Practice varies, but the accounting treatment should be applied consistently: if new ARR is recognized at signing for one deal, the same rule needs to apply to every comparable deal, otherwise the waterfall's beginning balance won't reconcile cleanly from one period to the next.
Alain spent a decade building and scaling subscription businesses before founding Revvly. He writes about the infrastructure gap between what SaaS companies report and what they actually earn.