Quick definition: Monthly recurring revenue (MRR) is the normalized monthly value of active recurring subscription commitments.
What is MRR?
MRR expresses recurring subscription value on a monthly basis so teams can monitor a changing customer base consistently. An annual contract worth $1,200 per year contributes $100 MRR; a month-to-month $100 subscription also contributes $100 MRR. It is useful for subscription planning because it separates recurring contractual value from one-time implementation fees, usage charges, tax, and non-recurring services.
MRR is not cash collected, recognized revenue, bookings, ARR, or profit. Cash can arrive annually in advance, revenue recognition follows accounting rules, bookings can include non-recurring commitments, and profit subtracts costs. Annual recurring revenue is commonly ARR = MRR × 12, but both measures need the same inclusion rules. Gross MRR counts active recurring value; net MRR calculations also reflect churn, contraction, expansion, and reactivation.
MRR formula and components
MRR = Σ normalized monthly recurring subscription value for active accounts. For a period bridge, ending MRR = starting MRR + new MRR + expansion MRR + reactivation MRR − contraction MRR − churned MRR. Define the snapshot date, contract status, proration, discounts, credits, trial treatment, foreign-exchange method, and whether usage fees count. A customer count is not a substitute: two accounts can contribute radically different MRR.
MRR measurement policy
An MRR policy should answer the operational questions that a formula leaves open. Select a daily or month-end snapshot convention; then apply it consistently. Many companies measure the committed monthly value active on the final calendar day, while others use a beginning-of-month snapshot for planning. Neither convention is universally correct, but changing between them can create a false trend. Specify whether a contract becomes active at signature, provisioning, payment, or service start, and whether a canceled plan leaves MRR at cancellation request or at the paid-through date.
Normalize non-monthly plans from their contractual recurring value. A $12,000 annual subscription generally contributes $1,000 MRR, while a $3,000 quarterly subscription contributes $1,000. Do not mechanically normalize a pre-paid one-time service fee. For discounted plans, measure the contracted recurring amount actually due during the discount period unless the organization separately reports list-price MRR. Credits, refunds, taxes, usage overages, professional services, and pass-through charges should be excluded or included by a written rule rather than case-by-case judgment.
Currency policy matters for international subscriptions. A team can report transactional-currency MRR converted at each period’s exchange rate, which reflects current reporting value, or at a fixed rate, which makes underlying subscription movement easier to compare. Both views can be useful; label them clearly and do not mistake foreign-exchange movement for expansion. Establish a late-data cutoff and an adjustment process so a retroactive cancellation does not silently revise a previously published bridge.
MRR in A/B testing
MRR is often too delayed and sparse for a short onboarding or pricing experiment, but it is a valuable longer-term outcome. A paywall test can use net revenue per assigned user as a near-term primary metric and retain cohorts to observe MRR, cancellation, and renewal. Analyze all eligible assigned accounts, including those that never subscribe; conditioning on payers answers a different question. Pair revenue with refund, support, margin, and retention guardrails as described in this guide.
Predefine the observation window and do not extrapolate first-month upgrades as durable MRR. A promotion can pull future purchases forward or produce low-retention subscribers. Plan sufficient maturity using A/B test duration, and report uncertainty around incremental recurring value.
Worked scenario
At month start, a SaaS product has $500,000 MRR. During the month it adds $40,000 new MRR, $18,000 expansion, and $5,000 reactivation; it loses $12,000 to contraction and $25,000 to churn.
ending MRR = 500,000 + 40,000 + 18,000 + 5,000 − 12,000 − 25,000 = $526,000
It also tests a revised annual-plan screen. Treatment has a promising first-month subscription lift, but the decision remains provisional until refunds and renewal behavior mature. Finance and product reconcile the subscription ledger, discounts, and assignment records before estimating incremental MRR.
Consider a second account-level case. A customer starts the month on a $500 monthly plan, upgrades to $800 on the 16th, and receives a $100 one-time credit after a support issue. Under a month-end snapshot policy, the account contributes $800 MRR and $300 expansion MRR. The credit affects cash and perhaps recognized revenue, but not MRR if the recurring commitment is unchanged. Under a prorated daily MRR reporting convention, the period’s recurring value may be represented differently, so the policy must name which convention powers the dashboard and which powers the period bridge.
Data-quality limitations
Subscription systems contain upgrades, downgrades, paused plans, backdated cancellations, refunds, multi-currency contracts, and account mergers. Define one source-of-truth ledger and version the transformation from contracts to MRR. Reconcile dashboards to billing and document late-arriving changes. In experiments, verify allocation and exposure; a treatment-specific billing integration error can look like monetization lift. Investigate unexpected splits with SRM diagnostics.
MRR is also vulnerable to timing bias. A test that ends just before a monthly billing date can appear to have no subscription effect even when a cohort will convert days later. Conversely, a time-limited discount can create a temporary MRR spike that disappears at renewal. Compare cohorts at equal age, preserve the original assignment date, and distinguish provisional booked subscription value from settled, refund-adjusted outcomes. For high-value enterprise contracts, report account counts and the underlying distribution because a single deal can dominate a small experiment.
Common mistakes
- Counting one-time fees as recurring: separate them from MRR.
- Mixing cash and MRR: annual prepayment does not create twelve times the monthly commitment.
- Ignoring discounts and status rules: publish the inclusion policy.
- Using payer-only analysis: it hides conversion effects.
- Assuming initial MRR persists: check churn and renewal.
Frequently asked questions
Is MRR the same as revenue?
No. MRR is normalized recurring contract value, while recognized revenue follows accounting treatment.
Should annual plans count in MRR?
Usually yes, normalized to one month according to the recurring subscription commitment.
Can MRR decline while customer count grows?
Yes. New accounts may be lower value or existing accounts may contract or churn.
Why use MRR in an experiment?
It captures durable subscription value, but needs a mature window and a clear assignment-based denominator.
Do free trials contribute to MRR?
Usually no, because they do not yet represent an active paid recurring commitment. Report trials separately and define the conversion event that moves an account into MRR.
How should upgrades during a month be classified?
Classify the increase as expansion MRR according to the snapshot or proration policy. Apply the same rule to all accounts and retain the underlying contract change for audit.
Should usage-based charges be included?
Include only the recurring committed component unless the organization has a documented recurring-revenue policy for predictable usage. Variable consumption is often better reported separately.
Summary
MRR is the monthly normalized value of active recurring subscriptions. Define contract, discount, currency, and status rules precisely; distinguish it from cash and revenue. In experiments, use it as a mature value outcome alongside near-term revenue, retention, and customer-protection metrics.