Key takeaways
- ARR is MRR × 12, and MRR moves through five components every month: New, Expansion, Contraction, Churned, Reactivation. The waterfall is only as reliable as those five reconciling.
- The three most common causes of a deck-vs-books ARR gap: churned accounts never formally removed, non-recurring revenue counted as recurring, and annual contracts counted before renewal is confirmed.
- Investors verify ARR through three checkpoints: ARR schedule to billing system, billing system to bank deposits, and GAAP revenue to ARR-implied revenue.
- An illustrative $2.4M deck ARR reconciling to $1.87M in the books is a $530K gap — explainable, but explaining it costs time and investor conviction that a clean reconciliation would have kept.
Your ARR figure is the first number a new investor encounters — it’s in the pitch, the update, the board deck. The question that gets asked in diligence isn’t whether you know the number. It’s whether the number is real. And “real” means something specific: it means the figure can be traced from the deck back to a billing system, and from the billing system back to your financial statements.
When it can’t, the ARR number becomes the story — not in a good way.
What the MRR waterfall is
The MRR waterfall is the structured breakdown of how monthly recurring revenue moves from one period to the next. It answers: where did revenue come from, where did it go, and how did we end up where we ended?
The five components:
- New MRR — Revenue from customers who didn’t exist in the prior month
- Expansion MRR — Additional revenue from existing customers: upgrades, seat additions, price increases
- Contraction MRR — Reduced revenue from customers who downgraded or negotiated discounts
- Churned MRR — Revenue from customers who canceled entirely
- Reactivation MRR — Revenue from previously churned customers who re-subscribed
The arithmetic: Starting MRR + New + Expansion − Contraction − Churn + Reactivation = Ending MRR
Financial Edge describes the waterfall’s analytical value clearly: it reveals whether a company’s growth is coming from acquiring new customers or from expanding existing ones — “a distinction that matters enormously for both growth durability and investor valuation.” Expansion-led growth implies a product that deepens customer relationships over time. New logo-only growth implies a product that customers don’t buy more of — which raises retention questions investors will probe.
ARR is simply MRR × 12. A company with $200K ending MRR is showing $2.4M ARR. An investor verifying that number will pull the waterfall and reconcile each component to the billing system and, from there, to bank deposits and GAAP revenue. If the waterfall doesn’t hold up at each step, the ARR figure doesn’t either.
Where the deck-to-books gap lives
Companies track MRR/ARR in separate systems from their GAAP books — billing platforms (Stripe, Chargebee), dedicated metrics tools (Baremetrics, ChartMogul), or spreadsheets. The GAAP books live in QuickBooks, Xero, or NetSuite. These systems were never designed to sync automatically, and in practice, they drift.
Baremetrics has documented this divergence directly: even between Stripe and Baremetrics — two platforms designed to track the same subscriptions — MRR numbers differ because Stripe includes free trial users in its active calculations (with only 25–50% of free trials converting to paid customers), uses subscription creation or renewal dates while Baremetrics uses paid invoice payment dates, and doesn’t adjust for mid-month signups or cancellations the way Baremetrics does. If two purpose-built systems diverge, the gap between a metrics spreadsheet and a GAAP general ledger is almost always larger.
The most common sources of the discrepancy:
1. Churned customers still in ARR
An account stopped paying three months ago. The founder mentally flagged it as “churning” but hasn’t formally removed it from the ARR schedule. It’s still in the $2.4M number. The analyst reconciles the ARR schedule to cash deposits and finds no deposits from that customer in Q3. That $80K in ARR disappears from the count — and the analyst now wonders what else is in the schedule that hasn’t been collected.
2. Non-recurring revenue counted as recurring
Professional services, implementation fees, one-time migration work, and annual support packages sometimes end up in the ARR calculation because they appear on the same invoice or in the same P&L line. Andreessen Horowitz identifies this as the single most common mistake in their 16 Startup Metrics guide: “The most common error is including one-time fees — hardware, setup, installation, professional services — when annualizing monthly bookings.”
A company with $2.1M ARR that includes $200K of annual professional services and $80K of one-time implementation work has $1.82M of actual recurring ARR. That’s a 13% haircut. Investors apply it regardless.
3. Annual contract value counted before renewal is confirmed
A 12-month contract signed in April is counted at $96K ARR. The renewal conversation starts in February. If the customer doesn’t renew, the full $96K churns — and the founder will have been showing $96K of ARR that was at risk for months without any signal in the metrics.
Clean ARR practices remove the contract from ARR either 30 days after expiration or when churn is confirmed, whichever comes first. Not at the moment it’s most convenient.
4. Collections falling below MRR
As SaaStr founder Jason Lemkin warns, when cash collections fall below Monthly Recurring Revenue, it “almost always leads to a restatement of revenue… downwards.” The MRR may look right on paper but the cash isn’t following it — a sign that some of what’s being called recurring revenue isn’t actually being collected.
Deck vs. books — illustrative waterfall
| Metric | Deck (investor update) | Books / billing reconciliation | Gap | Root cause |
|---|---|---|---|---|
| ARR | $2.40M | $1.87M | $530K | 3 churned accounts still in ARR ($110K); professional services included ($200K); one account double-counted ($80K); unconfirmed renewals ($140K, offset by partial) |
| Churn (MRR %) | 3.8% | 6.1% | 2.3 pts | Slow-to-churn policy; manual removal only when customer explicitly requests |
| Expansion MRR | $22K/month | $9K/month | $13K | Expansion counted at upsell signing, not at billing/delivery |
| Implied NRR | 114% | 97% | 17 pts | Drives directly from churn and expansion gaps above |
How investors actually verify ARR
GSquared CFO describes the verification process directly: “Buyers verify ARR through monthly ARR bridges showing new bookings, expansions, contractions, and churn at the customer level — then reconcile the seller’s ARR schedule against payment processors and bank deposits, stripping out one-time and services revenue.”
That reconciliation has three checkpoints. First: ARR schedule → billing system. Every customer in the ARR schedule should have an active subscription or contract in the billing platform. Any customer in ARR with no corresponding active account in Stripe or Chargebee is a red flag.
Second: Billing system → bank deposits. Cash deposited should match invoices issued (with timing adjustments for payment terms). Customers invoiced but not paid in 90+ days don’t belong in ARR.
Third: GAAP revenue → ARR-implied revenue. The relationship between recognized GAAP revenue and ARR is expected to show a gap (ARR is forward-looking; GAAP is backward-looking). But the gap should be explainable — growth rate times average contract value times recognition timing. If the ARR-implied GAAP revenue is significantly higher than actual recognized GAAP revenue, something is wrong with either the ARR count or the revenue recognition.
SaaS metrics focus on subscription-based recurring revenue and forward-looking indicators, while GAAP metrics rely on sales that have already been made — and a company that can produce the bridge between them fluently signals control. A company that discovers the gap for the first time during diligence signals something else.
What “board-ready” metrics look like
Bessemer Venture Partners, in their framework for scaling SaaS companies to $100M, notes that investors are not uncomfortable with ARR diverging from trailing GAAP revenue — the divergence is expected and appropriate for growing companies. What they are uncomfortable with is ARR that can’t be supported by retention data, billing records, and financial statements when they go to verify it.
The standard for a clean ARR/MRR waterfall:
- One system owns MRR. Not a spreadsheet that gets updated quarterly and a billing system that’s the “real” source and a metrics tool that shows something different. One system, updated continuously.
- Definitions are documented and haven’t changed. What counts as active? What triggers a churn entry? What qualifies as expansion vs. a new contract? Write it down. If anything changed, document when and why.
- Non-recurring revenue is separated. Services revenue, implementation, one-time fees — these belong on a separate line in the revenue schedule and never in the ARR calculation.
- Churned accounts are removed promptly. Best practice: within 30 days of churn confirmation, not when it’s convenient or when the reconciliation process surfaces it.
The waterfall reconciles monthly. Before any investor conversation, someone on the finance team should be able to reconcile ending MRR this month to ending MRR last month through the five components, and that ending MRR should tie to total cash collected from active subscriptions.
The practical test
Before your next raise, pull your ARR schedule and try to reconcile it yourself: take the customer list, match each customer to the billing system, check for deposits against each account over the last 90 days, strip out any non-recurring revenue, and see what number you end up with.
If that number matches what your deck says, you’re in good shape. If it doesn’t, you’ve identified the gap before an investor does — and that gap is almost always smaller and cheaper to address on your own timeline than on theirs.
Revenue recognition can be right and the waterfall can reconcile, and there’s still one layer where clean books routinely fail: the expense side. The next part of this guide covers accruals and cutoff — the accounting discipline that determines whether the margin in your P&L is real.
