Net revenue retention (NRR)
How much revenue today's customers still carry twelve months from now — after churn, downgrades and expansion, excluding new customers.
NRR = (beginning ARR + expansion − downgrades − churn) ÷ beginning ARR
It is the only common software metric that separates bought growth from growth that carries itself. Revenue growth can be manufactured with a sales budget. Net revenue retention above 100% cannot: it arises only when existing customers consume more of their own accord.
Why above 100% is possible at all
Because in software businesses billed by usage or by seat, a customer can grow without anyone selling them anything new: more users, more data, an additional module. At 115% NRR the company grows 15% even if it wins no new customers at all.
That is the economic heart of the metric. Sales then only has to generate the growth on top — not replace the base load.
Calibration — but only within the segment
The thresholds in circulation come mostly from listed cloud providers serving large enterprises. Applied to a small software house they measure the wrong company. The dependable comparators come from the annual surveys of private SaaS providers; SaaS Capital, KeyBanc/Sapphire and Benchmarkit arrive independently at the same order of magnitude.
| Distribution in private SaaS | NRR |
|---|---|
| Upper quartile | 111% |
| Median | 101–102% |
| Lower quartile | 97% |
The median, then, sits just above 100%, not at 115%. And it splits further as soon as one separates by customer size:
| Customer segment | Typical NRR | Why |
|---|---|---|
| Enterprise | around 118% | Expansion through seats, sites and add-on modules; contracts run multi-year |
| Mid-market | around 108% | Growth within the base exists, but is bounded |
| SMB | around 97% | Small customers disappear from the market more often, switch faster and expand little |
That is the point of this page. Dismiss everything below 95% as problematic and you declare a large part of the small-customer software business sick — when it is sitting precisely on its own comparator. An SMB provider at 97% is average. An enterprise provider at 97% has a problem. The same number, two opposite findings.
Alongside this sits the framework from Bessemer Venture Partners: 100% good, 110% better, 120% and above best. It is quotable and widely used, but it remains a target picture for venture portfolios, not a market average. As a bar to clear it works; as a benchmark it does not.
Where the metric misleads
There is no binding definition. NRR is defined in no accounting standard — neither IFRS nor US GAAP. It is not even a "non-GAAP financial measure" under the SEC rules, for which a reconciliation to a reported line item would be prescribed. Every issuer defines it itself, every issuer may change the definition, and nothing is audited. Many companies say exactly that in the risk section of their annual report: that their metric is not comparable with the identically named metric of other providers.
Three levers move the result by double-digit percentage points without anything in the business changing:
- Which cohort? All customers as at the date twelve months ago — or only those above a certain revenue size? "NRR of our customers above €100k of annual revenue" is a quite different number, and it is often quoted without that qualifier.
- Is expansion capped? Some companies limit the contribution of individual customers, others do not. Without a cap, one large customer can carry the quarter.
- Do price increases count as expansion? Formally yes. Economically, an inflation adjustment is something other than a customer using more product.
It is revenue-weighted and hides customer churn. Ten small customers can leave while one large one expands — and NRR still rises. Which is why logo churn belongs next to it. When the two diverge, that is an early signal: the product carries at the top but not across the base.
It is backward-looking. It measures contracts largely signed twelve months ago. A deterioration in new business shows up only a year later.
And it says nothing about profitability. A company at 130% NRR can be burning cash. Which is why it sits alongside the Rule of 40 and gross margin — not in their place.
At smaller companies it often does not appear at all
Small caps report NRR less often than large software providers, and when they do, without a definitional footnote. If the disclosure goes missing for several quarters at a company that used to give it, that in itself is information.
How we use the number
We take no NRR without the accompanying definition from the annual report. We assign it to the customer segment the company actually sells into — a threshold without a segment is meaningless — set it against logo churn, and look at the trajectory across at least four quarters. A single reading is far too easy to pick to suit.
The full framework is set out under How we value technology companies.
Sources
- SaaS Capital: What Is a Good Retention Rate for a Private SaaS Company? — median, quartiles and breakdown by customer segment for private software providers. saas-capital.com
- Bessemer Venture Partners: State of the Cloud — origin of the "good 100%, better 110%, best 120%+" framework. A target picture, not a market average. bvp.com
- The order of magnitude of the median is independently confirmed by the annual surveys from KeyBanc/Sapphire and Benchmarkit.
As of: 11 August 2026 · Version 2