Net Revenue Retention (NRR)

How to calculate net revenue retention

NRR answers a narrow question: if you had signed no new customers, how much recurring revenue would your existing base be worth now? Above 100%, the base grew on its own because upgrades and add-ons outweighed cancellations and downgrades. Below 100%, it shrank.

The measurement is a cohort. Take every customer who was paying at the start (for example 12 months ago), add up what they paid then and what those same customers pay now, and divide. You can run it on MRR or ARR as long as start and end use the same one.

Worked example

A software company had $500,000 of ARR from its existing customers twelve months ago. Since then those customers have added $70,000 of upgrades and extra seats (expansion), downgraded $15,000 (contraction), and cancelled $40,000 (churn).

NRR is $515,000 ÷ $500,000 = 103%. Gross revenue retention (GRR) leaves out expansion: ($500,000 − $15,000 − $40,000) ÷ $500,000 = 89%. GRR can never exceed 100%, and NRR can.

The gap between the two, 14 points here, is the expansion engine. It tells you how much of your retention depends on selling more to customers who stay and how much on simply keeping them. A company at 103% NRR and 89% GRR is losing 11% of its base each year and covering it with upsells. Fixing churn would help more than any new upsell campaign.

Why a blended NRR can hide the real story

The 103% figure is an average across very different customers. Split the same $500,000 starting cohort by segment and the picture changes.

Small-business customers shrink by 10% a year while enterprise customers grow by 20%. The blended 103% suggests a steady business, but it is really one segment leaking and another compounding. If the company keeps adding small-business customers, it will keep refilling a bucket with a hole in it. Shifting sales effort toward mid-market and enterprise accounts, or fixing the small-business product experience, would lift the blended figure faster than any pricing change.

When you report NRR to investors or a buyer, state the period (trailing twelve months is the usual choice), whether it is based on ARR or MRR, and how you treat price increases, usage overage and reactivated accounts. Two companies quoting the same percentage under different definitions are not comparable.

What is a good NRR?

Benchmarks move with pricing model, company size and contract value, so quote the one that matches you. Aleph, summarizing its 2026 report with Benchmarkit on 2025 data from 342 SaaS companies, gives a median NRR of about 102% and a median GRR of about 84%. The top quartile of the whole sample is at about 110% and the bottom quartile at about 92%.

Read these as directional. A small company at 94% is average for its size, not failing. But below 90% is a sign that the business is leaking customers faster than it can grow them, and above 110% is strong in most segments.

Why NRR compounds

NRR applies every year to the base you already have, so small differences add up. Start with a cohort worth $100,000 and sign no one new:

After three years, the 110% cohort is worth $133,100 and the 90% cohort is worth $72,900, which is 82% more. That is why investors weigh NRR heavily. A company above 100% needs less new-customer spending to reach the same growth, and it is less exposed to a slow sales quarter. It also improves the return on customer acquisition cost, since each customer becomes worth more over time.

Monthly and annual figures need care. A monthly NRR of 101% compounds to 1.01^12 = 112.7% a year. If you quote a monthly figure next to someone's annual one, the comparison is wrong.

Reading NRR correctly

• Check GRR alongside it. An NRR of 105% with 75% GRR is a very different business from 105% with 92% GRR. The first depends on heavy upselling to cover heavy churn.
• Look for concentration. One large customer expanding by $100,000 can lift NRR for a company with 20 customers. Recompute with the top account removed.
• Separate price increases from real expansion. A 5% list-price increase raises NRR with no new value delivered, and it can raise churn later.
• Segment it. Small customers often retain worse than large ones. A blended number can hide one segment that is leaking.
• Consider the pricing model. Usage-based products can post high NRR when customers grow, and fall sharply when they cut back. Seat-based products usually move more slowly.

For the churn side in more detail, see churn rate and the churn rate calculator.

How to improve NRR

• Reduce involuntary churn. Failed card payments cancel accounts by accident. Retry logic and card-update reminders recover revenue at almost no cost.
• Onboard for the outcome. Customers who reach value in the first weeks renew. Track activation, not just sign-ups.
• Build expansion into the product. Seats, usage tiers and add-ons give customers a natural way to buy more as they grow.
• Focus on the right customers. If one segment churns at twice the rate of the rest, changing who you sell to can lift NRR faster than changing the product.
• Watch contraction. Downgrades often come before cancellations. Contact accounts whose usage falls.

Common mistakes

• Including revenue from new customers. That turns NRR into a growth rate. Only customers present at the start of the period belong in the numerator.
• Counting reactivated customers as expansion. A customer who left and returned belongs in new business unless they never left the cohort.
• Mixing MRR and ARR. Start with one and finish with the other and the ratio is off by a factor of twelve.
• Using a period that is too short. One good month proves little. Look at a trailing twelve months, or a cohort followed across the same number of months.
• Ignoring definitions when comparing. Ask whether a competitor's NRR includes usage overage, one-time fees or price increases.

What NRR can't tell you

NRR says nothing about how fast you win new customers, how much it costs, or whether the business is profitable. A company can post 115% NRR and still burn cash on acquisition. Combine it with the burn multiple and the Rule of 40 for a fuller picture.

Frequently asked questions