A founder stands up at a board meeting and says net revenue retention is 118%. The room relaxes, the customer base is growing on its own. What nobody asks is the second number: gross revenue retention is 84%. The base is also bleeding badly, and a handful of expanding whales are papering over it. Both numbers are true. Only together do they tell you what is actually happening.

The quick version

  • Gross revenue retention (GRR) measures pure leakage: how much of last year's recurring revenue you kept after churn and downgrades. It caps at 100%, expansion is excluded on purpose.
  • Net revenue retention (NRR) adds expansion back in. Above 100% means your existing customers grow faster than they shrink, you can grow even if you never sign another logo.
  • You need both. High NRR with low GRR means you are expanding fast enough to mask serious churn, a leak hidden by a few big upsells.
  • NRR is one of the strongest single predictors of how the market values a subscription business, which is exactly why it gets gamed.

The idea in depth

Both metrics start from the same place, the recurring revenue of a cohort of customers you had at the start of a period, and ask what it is worth a year later. The difference is one line in the formula, and that line changes everything.

Gross revenue retention = (starting recurring revenue − downgrades − churn) ÷ starting recurring revenue. It strips out every upgrade. By construction it can never exceed 100%, because the best you can do is lose nothing. A GRR of 90% means you leaked a tenth of the base to cancellations and shrinkage before you sold anyone a single extra seat.

Net revenue retention = (starting recurring revenue − downgrades − churn + expansion) ÷ starting recurring revenue. Adding expansion back in lets the number cross 100%. An NRR of 118% means last year's customers are paying 18% more this year, upsells and cross-sells outran every cancellation. Both numbers look only at existing customers; new logos are excluded from both, which is the point, they isolate the health of the base from the noise of acquisition.

flowchart TD
    A(["Starting recurring revenue
(last year's cohort)"]) --> B(["− Churn (cancellations)"])
    B --> C(["− Downgrades (contraction)"])
    C --> G(["= Gross revenue retention
caps at 100%"])
    G --> D(["+ Expansion
(upsell / cross-sell / usage)"])
    D --> N(["= Net revenue retention
can exceed 100%"])
The same waterfall, stopped at two different points. GRR is the leak; NRR is the leak plus the growth on top. Leaders Loop

Why NRR became a headline metric, and what the evidence really says

NRR moved from back-office trivia to boardroom headline for one reason: it predicts how a subscription business grows and how the market prices it. Investors and operators mostly agree here. Bessemer Venture Partners' much-quoted benchmarks in State of the Cloud put NRR at 100% (good), 110% (better) and 120%+ (best), with the obvious caveat that those targets move by stage, and by whether you sell to enterprise, mid-market or SMB.

The reason it matters financially is the arithmetic of compounding. As Jason Lemkin of SaaStr argues in "What's More Important, New Customers or Retention? It's a Trick Question", at scale most growth comes from the base you already have: he notes PagerDuty entering its IPO at 139% NRR while growing 48% year-on-year, and Salesforce reporting roughly 73% of new bookings from existing accounts. A company above 100% NRR rides an escalator that climbs on its own; one below 100% runs up a downward escalator, where every new logo first has to refill the bucket before it adds anything.

So make NRR a number the whole commercial org owns, not a slide finance produces once a quarter. Set a target appropriate to your segment (enterprise can aim past 120%; SMB-heavy books realistically sit lower), forecast it like a pipeline, and decompose it into its drivers, expansion, contraction, churn, so a bad month points to a fixable cause.

Gross retention is the truth serum

NRR's great weakness is that expansion can hide a leaking base. This is where gross retention earns its keep. Because GRR ignores expansion entirely, it cannot be flattered by a few large upsells. A business can post a comfortable 115% NRR while its GRR sits at 82%, meaning nearly a fifth of customers are leaving or shrinking each year, and the headline only survives because a small set of accounts is expanding hard. That is a fragile model: it depends on a handful of whales continuing to grow, and it usually signals that the product delivers for a narrow segment and disappoints everyone else.

The benchmarks bear this out: the two numbers move differently. Industry surveys, Benchmarkit's 2025 SaaS metrics report and KeyBanc's annual private-SaaS survey among them, have put median private B2B GRR in the high 80s while median NRR sat around 101–104% in recent years. Both swing sharply with deal size. Enterprise books retain and expand far better than SMB ones, which churn more and expand less.

NRR tells you how fast the base grows. GRR tells you how fast it leaks. A leader who watches only the first is one bad quarter from an unpleasant surprise.

So read them as a pair, always. Put GRR and NRR side by side. A healthy business has a high GRR (a base that holds) and NRR comfortably above it (expansion on top of a base that holds). A wide gap between a strong NRR and a weak GRR is not a triumph, it is a warning that your growth rests on a few accounts and a leaking floor.

The honest limitation: the metric is real, the number is negotiable

Here is where the playbook breaks down. NRR has no single enforced definition, so two companies quoting "120%" may be measuring different things, the cohort window, whether churned customers sit in the denominator, how usage-based and multi-year contracts get recognised. Reviews of public-company filings have found that a large share of firms adjust or customise their net-retention formula, which makes cross-company comparison looser than it looks. So treat the benchmarks above as directional. They are not a scoreboard to read literally against your own number.

There is a deeper limitation too. NRR is a lagging, aggregate measure, it tells you what already happened to the whole base, not which customers are about to leave or why. It can stay healthy right up until a concentration risk (a few large accounts) unwinds. So treat NRR and GRR as the destination metrics, and instrument the leading indicators underneath them, product adoption, support health, and cohort curves, so you see trouble in the signal before it shows up in the number. (We cover those upstream signals in cohort and retention-curve analysis and the wider recurring-revenue metrics picture.)

A worked example

The figures below are illustrative, chosen to show the mechanics rather than to report a real company.

Picture a SaaS business that started the year with £10,000,000 in recurring revenue across its existing customers. Over the year, three things happen to that cohort. Some customers cancel outright, £1,200,000 of churn. Some shrink, dropping seats or tiers, £600,000 of contraction. And some grow, buying more, £2,500,000 of expansion.

Work the two formulas. GRR ignores the expansion: (10,000,000 − 1,200,000 − 600,000) ÷ 10,000,000 = 82%. NRR adds it back: (10,000,000 − 1,200,000 − 600,000 + 2,500,000) ÷ 10,000,000 = 107%.

flowchart LR
    S(["Start: £10.0m"]) --> L(["− £1.2m churn
− £0.6m contraction"])
    L --> GR(["GRR = £8.2m / £10.0m
= 82%"])
    GR --> EX(["+ £2.5m expansion"])
    EX --> NR(["NRR = £10.7m / £10.0m
= 107%"])
Same cohort, two readings. The 107% headline looks healthy; the 82% floor is the part the board should worry about. Illustrative figures. Leaders Loop

Read alone, the 107% NRR is a good-news slide: the base grew 7% with no new customers. But the 82% GRR says nearly a fifth of the starting revenue walked out or shrank, and the net number is only positive because of £2.5m of expansion likely concentrated in a few accounts. The move is now obvious, and would have been invisible from NRR alone: before chasing more expansion, find out why so much is leaking. Lift GRR from 82% to 90%, closing the leak, not adding upsell, and NRR jumps to 115% with no extra expansion effort. Fixing the floor is often the cheapest growth in the building.

Frequently asked questions

What's the simple difference between NRR and GRR?

GRR measures only the bad news, churn and downgrades, so it can never exceed 100%; it tells you how leaky your base is. NRR adds expansion (upsells, cross-sells, usage growth) back in, so it can exceed 100%; it tells you whether your existing customers are net growing. Same cohort, same period, one excludes expansion, one includes it.

Is net revenue retention the same as net dollar retention (NDR)?

Yes, in practice. "Net dollar retention" (NDR) and "net revenue retention" (NRR) are used interchangeably for the same calculation; NDR is the more common term in US filings, NRR the more global one. Watch for definitional differences between companies rather than between the two labels, the formula choices (cohort window, denominator, contract treatment) matter far more than the name.

What's a good NRR, and is 100% the line?

For NRR, 100% is the meaningful threshold: above it, your base grows on its own. Bessemer's rule of thumb is 100% good, 110% better, 120%+ best, but that was framed for growth-stage enterprise SaaS. SMB-heavy businesses run structurally lower because small customers churn more and expand less, so a 105% NRR can be excellent in one model and mediocre in another. Benchmark against your own segment, not a headline figure.

Why do investors care so much about NRR?

Because it compounds and it predicts efficient growth. A business with high NRR gets cheaper growth each year, more of next year's revenue comes from customers it already serves rather than from ever-rising acquisition spend, and analyses of public software companies have repeatedly tied higher net retention to higher revenue multiples. It is, in effect, a single number that captures product value, pricing power and stickiness at once. That is also why it gets gamed, which is why a serious investor asks for GRR too.

Can NRR hide a failing business?

Yes, that is the classic trap. A strong NRR driven by a few rapidly-expanding accounts can sit on top of a badly leaking base. If those accounts ever stop expanding (or churn), the underlying decay surfaces all at once. The defence is to always read GRR alongside NRR, and to check how concentrated your expansion is across customers.

Related in the Toolkit

Where to go next