What Is ARR vs MRR? Formula, When to Use Each, and 2026 Benchmarks

Annual recurring revenue (ARR) and monthly recurring revenue (MRR) measure the same subscription book on two clocks. MRR is the normalized recurring revenue your active subscriptions would generate in one month. ARR is that run-rate annualized — almost always ARR = MRR × 12. Neither is GAAP revenue. Both exclude one-time fees, professional services, and non-recurring usage spikes. The operator question is not which number is “correct.” It is which clock matches the decision in front of you.

Growth Centr publishes evergreen, research-backed analysis on SaaS unit economics for founders, marketers, and operators who need a clean recurring-revenue definition before they trust any retention, CAC, or Rule of 40 print built on top of it.

ChartMogul’s SaaS metrics reference treats MRR as normalized monthly subscription revenue and ARR as the annualized run rate (MRR × 12). Stripe’s operator guide makes the practical split: MRR shows momentum and recent decisions; ARR shows durable scale for boards, budgets, and peer comparisons. In 2026, that split sits next to a retention market that has loosened — Benchmarkit’s 2026 SaaS and AI-Native Metrics (CY-2025 actuals, 342 B2B SaaS and AI-native companies) puts median GRR at 84% (down from 88%) and median expansion contribution to net new ARR at 40%. If your ARR definition is dirty, every downstream metric is dirty.

Key Takeaways

• MRR = normalized recurring subscription revenue for one month. ARR = MRR × 12 (same run-rate, annual clock). They must reconcile; if they do not, your definitions or timestamps differ.
• Count only predictable recurring revenue. Exclude setup fees, professional services, taxes, and non-contracted usage spikes. Include base subscription, contracted seats, and committed add-ons.
• Decompose growth with the five ChartMogul movements: new, expansion, reactivation, contraction, churn. Net new = new + expansion + reactivation − contraction − churn.
• Lead with MRR for weekly ops, PLG, and monthly-billed SMB books. Lead with ARR for board packs, fundraising, annual contracts, and peer benchmarks (Rule of 40, NRR, CAC ratios).
• Benchmarkit 2026 context: software gross margin median 80%+, GRR 84%, expansion already 40% of net new ARR at the median. Clean ARR is the denominator for all of that.

What Is MRR?

Monthly recurring revenue (MRR) is the amortized, normalized value of active subscriptions for a single month. A customer on a $1,200/year plan contributes $100 MRR, not $1,200 in the month they pay. A customer on a $99/month plan contributes $99 MRR. The point of normalization is comparability: annual prepay and monthly billing land on the same monthly runway.

ChartMogul frames MRR as “arguably the most critical revenue metric in subscription businesses” because every other SaaS operating metric — churn, LTV, expansion rate, quick ratio — inherits its definition. If MRR includes a one-time implementation invoice, your churn rate, LTV, and growth rate all lie in the same direction.

What belongs in MRR

Worked example. Start of month: 80 customers × $200/month = $16,000 MRR. During the month you add 10 new logos at $200 ($2,000 new), expand 5 accounts by $100 each ($500 expansion), lose 2 logos at $200 ($400 churned), and see one downgrade of $50 (contraction). Ending MRR = 16,000 + 2,000 + 500 − 400 − 50 = $18,050. Net new MRR = $2,050.

That ending figure is what you annualize. It is not “what cash hit the bank this month” — annual prepaid customers may have paid twelve months of cash last quarter while contributing only one month of MRR now.

What Is ARR?

Annual recurring revenue (ARR) is the annualized value of that same recurring book. The standard operating formula is:

ARR = MRR × 12

So $18,050 MRR → $216,600 ARR. Stripe describes ARR as the total annualized value of all active recurring contracts, assuming today’s customers, pricing, and contracts stay the same. That last clause matters: ARR is a run-rate snapshot, not a forecast of recognized GAAP revenue over the next twelve months, and not a bookings total for deals signed this quarter.

Boards, VCs, and public comps speak ARR because annual contracts, multi-year ACVs, and peer tables are annual. Efficiency metrics inherit the same clock: Benchmarkit’s blended CAC ratio is $1.30 of S&M per $1 of ARR; CAC payback and Magic Number are usually discussed against ARR growth, not raw monthly cash.

ARR is not:

• Bookings. Bookings often include TCV of multi-year deals, services, and non-recurring commitments. Inflating ARR with bookings is the fastest way to fail diligence.
• GAAP revenue. ARR will not appear as a line on a 10-K. Public companies disclose subscription revenue under ASC 606; private operators still run ARR as the internal north-star for GTM.
• Committed / contracted ARR (CARR). Some teams track ARR that is contracted but not yet live, or subtract known churn. That is a different metric. Label it. Do not silently swap it into “ARR” on a board slide.

ARR vs MRR: Same Book, Different Clock

They are not alternative truths. A company that reports $2M ARR and $180K MRR has a definition bug — $2M / 12 = $166.7K. Force the reconciliation every month. Stripe’s guidance is the operator rule: use MRR to see what just happened; use ARR to decide what the business is becoming.

When to lead with MRR

• PLG or self-serve motions where weekly signup and expansion moves matter.
• Monthly-billed SMB books where annualizing hides the volatility you need to manage.
• Pricing tests, packaging changes, and campaign readouts inside a single quarter.

When to lead with ARR

• Series A+ board decks and investor updates.
• Hiring plans, capacity models, and annual budgets.
• Peer comparisons on Rule of 40, NRR, CAC ratio, and ARR/employee (Benchmarkit 2026 median ARR/employee $175K).

Most scaled teams track both: MRR waterfall for the operating review, ARR for the board summary. Emphasis shifts toward ARR as ACV and contract length rise, but the MRR movements never stop being the diagnostic.

The Five-Movement Waterfall

A single ending ARR or MRR number does not tell you whether growth is healthy. ChartMogul’s standard decomposition — also echoed in Stripe’s new / expansion / churn / contraction / net-new framing — breaks the change into five movements:

Net new MRR (or ARR) = New + Expansion + Reactivation − Contraction − Churn

Worked ARR example. Starting ARR $10.0M. New $1.2M. Expansion $0.8M. Reactivation $0.1M. Contraction $0.3M. Churned $0.5M.

Ending ARR = 10.0 + 1.2 + 0.8 + 0.1 − 0.3 − 0.5 = $11.3M.
Net new ARR = $1.3M.
Expansion share of gross adds (new + expansion + reactivation) = 0.8 / 2.1 ≈ 38% — in line with Benchmarkit’s median where expansion already supplies 40% of net new ARR.

If you only report “we grew ARR 13%,” you cannot tell whether you bought logos that will churn or compounded the installed base. That is why NRR and GRR / churn benchmarks sit next to ARR on every serious board pack: NRR asks what happened to starting customers; the waterfall asks what happened to the whole book.

How ARR and MRR Feed Every Other SaaS Metric

Clean recurring revenue is the shared denominator. Pollute it and the rest of the stack misleads:

• NRR / GRR. Both are ratios of ARR movements over starting ARR. Benchmarkit private median NRR 102%, top quartile 110%; GRR 84%. Usage-based books print higher NRR (108%) than seat-based (98%).
• LTV and LTV:CAC. SaaS LTV is usually gross-margin ARPA ÷ churn. Benchmarkit median CLTV:CAC is 4.1×. ARPA is average MRR (or ARR/12) per account — wrong MRR, wrong LTV.
• CAC and payback. Blended CAC ratio $1.30 S&M per $1 ARR; new-name $1.63; expansion $0.80; median payback 16 months. Expansion ARR is the cheap growth dollar.
• Magic Number. Annualized net new recurring revenue ÷ prior-period S&M. Benchmarkit median 1.37. The numerator is ARR change (or equivalent), not bookings.
• Rule of 40. YoY ARR (or MRR) growth + profit margin. Private median 25%, top quartile 43%. Feld preferred YoY MRR; most 2026 private books speak ARR once past ~$20M.

Software gross margin held at 80%+ median in the same Benchmarkit book — ARR quality still sits on a healthy contribution base at the industry median, even as retention loosened. That is the 2026 tension: scale metrics look workable; the installed-base leak is the silent tax on every ARR print.

Common Definition Traps

1. Booking TCV into ARR. A three-year $300K contract is not $300K of ARR. Recurring annualized value is $100K ARR (or $8.33K MRR), unless you have a documented multi-year recognition policy you disclose every time.
2. Services and implementation. Recurring means the customer expects to pay again for the same access. Onboarding projects do not qualify because they feel “attached.”
3. Annual prepay timing. Cash lands day one; MRR still spreads across twelve months. Confusing cash with MRR overstates growth in prepaid months and understates it later.
4. Usage spikes. Contracted minimums or reliably recurring usage bands can sit in MRR. One-off overages should not, or your churn math becomes fiction the month usage normalizes.
5. Mixing GAAP subscription revenue with ARR in peer tables. Public comps often report GAAP; private decks report ARR. Label the basis or you are not comparing peers.
6. Silent CARR swaps. Contracted-but-not-live deals and known churn adjustments belong in a labeled CARR or CMRR view, not in the ARR line you trend for five years.

A 30-Day Operator Test

Days 1–7 — Lock the definition. Write one page: what counts as recurring, how annual plans amortize, how usage is treated, and the exact ARR = MRR × 12 rule. Pull ending MRR and ARR for the last 12 months. If finance and the CEO cannot produce the same pair in one sitting, stop reporting growth and reconcile.

Days 8–14 — Build the waterfall. Split each month into new, expansion, reactivation, contraction, and churned (MRR and ARR). Flag any month where “other” exceeds 5% of gross movement — that is usually a classification bug.

Days 15–21 — Attach retention and efficiency. Compute GRR and NRR on the same ARR definition. Pull CAC ratio / payback and Magic Number. Compare expansion’s share of net new ARR to Benchmarkit’s 40% median. If expansion is carrying the print while GRR is sliding, you have a substitution problem, not a growth story.

Days 22–30 — Set the reporting rule. Operating review: MRR waterfall + net new. Board pack: ARR, YoY ARR growth, NRR, GRR, CAC payback, Magic Number, Rule of 40. One sentence in the deck: “ARR and MRR reconcile at ×12; services and bookings are excluded.” Revisit after the next closed month — not after every pipeline call.

FAQ

What is the difference between ARR and MRR?

MRR is normalized recurring revenue for one month. ARR is the same recurring book on an annual clock, calculated as MRR × 12. They describe one subscription base; they answer different planning questions.

Is ARR just MRR times 12?

Yes, in standard SaaS operating practice (ChartMogul, Stripe, and most private board packs). If your ARR is not twelve times your MRR, you are either using different cut dates, mixing bookings into ARR, or applying a non-standard contracted-ARR adjustment you have not labeled.

What should not be included in ARR or MRR?

One-time setup fees, professional services, taxes, unpaid trials, and non-recurring usage spikes. Include base subscriptions, contracted seats, and committed add-ons at the recurring rate.

Should early-stage startups track ARR or MRR?

Track both, but operate on MRR while billing is monthly and cycles are short. Introduce ARR as the lead external number when you raise, sell annual contracts, or need peer benchmarks. Do not wait until Series B to make the two reconcile.

How does ARR relate to NRR and churn?

NRR and GRR are ratios built on starting ARR and the expansion / contraction / churn movements of that cohort. Logo churn counts customers; revenue retention counts dollars. You need clean ARR before any of those ratios mean anything — see our NRR guide and churn benchmarks.

What is a good ARR growth rate in 2026?

There is no single “good” rate — it compresses with scale. Use YoY ARR growth next to margin (Rule of 40: private median 25%, top quartile 43%) and GTM efficiency (Magic Number median 1.37, CAC payback 16 months). A high ARR growth print with GRR at 84% and expansion substituting for new logos is a different business than the same growth with top-quartile retention.

Methodology

This is a 2026 planning brief, not a survey we ran. MRR and ARR definitions, the MRR × 12 relationship, and the five-movement waterfall (new, expansion, reactivation, contraction, churn) follow ChartMogul’s SaaS metrics reference and cheat sheet, cross-checked against Stripe’s recurring-revenue operator guide. Private-company 2026 context — GRR 84% (from 88%), NRR median 102% / top quartile 110%, expansion 40% of net new ARR, software gross margin 80%+, ARR/employee $175K, blended CAC ratio $1.30, Magic Number 1.37, Rule of 40 median 25% / top quartile 43%, CAC payback 16 months — is from Benchmarkit’s 2026 B2B SaaS and AI-Native Performance Benchmarks (CY-2025 actuals, 342 companies). No statistic appears here unless it was on a page we fetched. ARR/MRR are non-GAAP operating metrics; they are not substitutes for ASC 606 revenue.