Most growth metrics mix together new customers and existing ones. Net revenue retention separates them. It asks: if we had won nobody new this year, would revenue be up or down?
What goes into NRR
Start with the recurring revenue from a fixed group of customers at the beginning of a period. Over the period, some of them upgrade (expansion), some downgrade (contraction) and some cancel (churn). NRR compares what that same group pays at the end with what it paid at the start.
Same customers, one year later
$100,000 + $12,000 − $3,000 − $5,000equals$104,000
Net revenue retention
$104,000 ÷ $100,000equals104%
Gross revenue retention
($100,000 − $3,000 − $5,000) ÷ $100,000equals92%
Revenue from customers who signed up during the year is deliberately left out. Including it would turn NRR into ordinary revenue growth.
NRR and GRR together
Gross revenue retention (GRR) ignores expansion, so it only shows how much revenue you kept. It can never go above 100%. Looking at both tells you where your growth or losses come from.
In the example, the company lost 8% of its starting revenue to downgrades and cancellations, but upgrades more than made up for it. If cancellations had been $15,000 instead of $5,000, NRR would fall to 94% and GRR to 82%.
Why investors watch it
A business with NRR above 100% grows every year even before sales and marketing do anything. That makes each new customer more valuable over time, and it makes growth cheaper, because part of it comes from customers you already paid to acquire.
How to raise it
- Pricing that grows with use, such as per seat or per usage, so customers pay more as they get more value.
- Add-ons or higher tiers that existing customers actually want.
- Reducing cancellations, which lifts both NRR and GRR.
What is a good NRR in 2026?
Net revenue retention has come down across the industry since 2021, so older rules of thumb such as "120% is normal" are out of date for most companies. Current benchmarks:
- Private B2B SaaS: a median of 102% and a top quartile of 110%, in the 2026 Benchmarkit and Aleph benchmarks based on full-year 2025 data.
- Bootstrapped companies with $3 million to $20 million of ARR: a median of 103%, with the top 10% at about 118%, in SaaS Capital's 2026 survey of more than 1,000 companies.
- Public software companies that disclose it: a median of about 109% as of late August 2026, from Meritech's public comparables.
Pricing model makes a large difference. In the Benchmarkit and Aleph data, usage-based companies had a median NRR of 108% against 98% for seat-based ones, because usage-based revenue grows automatically as customers use more. Customer size matters too: products sold to small businesses tend to retain less than those sold to large companies.
Gross revenue retention is under more pressure. The same benchmarks put median GRR at 84%, down from 88% a year earlier, with the top quartile at 91%. A falling GRR means more revenue is leaking out, and expansion has to work harder just to keep NRR above 100%.
Why a few points matter so much
NRR compounds. A group of customers retained at 110% a year is worth 1.1 × 1.1 × 1.1 = 1.331 times its starting revenue after three years, without any new sales. At 95%, the same group is worth about 0.86 times after three years. The difference between those two companies is not a few percent: after three years, one has 55% more revenue from the same starting customers.
This is also why investors value high-NRR businesses more highly. Each customer acquired keeps growing in value, so the money spent on acquisition earns a better return.
Mistakes that make NRR look better than it is
- Including revenue from customers who joined during the period. NRR must follow a fixed group from the start date.
- Counting customers who cancelled and came back as expansion.
- Using contracted ARR that has not started billing yet.
- Measuring over a month that happened to include large annual renewals or price increases, and presenting it as typical.
Tip: Measure NRR over a full year where you can. Monthly figures move around a lot, especially with annual contracts renewing in some months and not others.
Questions people ask
- Can NRR be above 100% with high churn?
- Yes, if expansion from remaining customers outweighs the losses. But gross revenue retention will be low, which is a risk if expansion slows.
- How often should I report NRR?
- Quarterly on a trailing twelve-month basis is common. Monthly figures swing too much with individual renewals.
- How do you calculate net revenue retention?
- Recurring revenue from a fixed group of customers at the end of a period ÷ what the same group paid at the start. New customers are excluded.
- What is the difference between NRR and GRR?
- Gross revenue retention ignores expansion, so it can never exceed 100%. NRR includes upgrades and can go above 100%.
- What is a good NRR?
- Above 100% means the existing customer base grows on its own, which investors treat as a strong sign of product value.
- What is a good net revenue retention rate in 2026?
- The median for private B2B SaaS is about 102%, with 110% top quartile (Benchmarkit and Aleph, 2025 data). Public software companies median about 109%.
- Why is NRR higher for usage-based pricing?
- Revenue rises automatically as customers use more. In the 2026 Benchmarkit and Aleph data, usage-based companies had a median NRR of 108% against 98% for seat-based.