Net revenue retention when you have 40 customers, not 4,000

Net revenue retention (NRR) is the share of last year's recurring revenue that the same customers pay you today, upgrades and downgrades included and new customers left out. Over 100% means the customers you already had now pay you more than they did. With 40 customers, though, one upgrade moves it six points. So read it once a year and with its parts beside it, never as a monthly score.

Here's the formula, a worked example on a 40-customer business, and why the benchmarks you'll find weren't measured on businesses your size.

The net revenue retention formula

ChartMogul, which makes subscription analytics, defines it as a cohort:

NRR = MRR today from that customer cohort ÷ MRR from that same cohort a year ago

The same thing, written as movements, is the form most people use:

NRR = (starting MRR + expansion − contraction − churned MRR) ÷ starting MRR

• Starting MRR is what your customers paid a month, twelve months ago.
• Expansion is what those same customers added: upgrades, extra seats, price rises.
• Contraction is what they dropped by downgrading.
• Churned MRR is what the ones who left used to pay.

Customers who arrived during the year aren't in it. That's the point: NRR asks what happens to revenue you already had, so it can't be flattered by a good month of sign-ups.

ChartMogul's own example uses four customers: $770 of MRR a year ago, $800 from the same four today, so NRR is 103.9%. The arithmetic doesn't care how many customers there are. What you can conclude from it does.

Gross revenue retention, its partner

Gross revenue retention (GRR) is the same calculation without expansion. It counts only what you lost:

GRR = (starting MRR − contraction − churned MRR) ÷ starting MRR

GRR can't go above 100%. NRR can, because upgrades can outweigh losses. ChartMogul names the common mistake directly: "comparing your GRR to someone else's NRR."

The two together tell you more than either alone. A high NRR with a low GRR means a few customers growing a lot while many leave, which is a different business from one where almost nobody leaves and nobody grows.

A worked example with 40 customers

This is an illustrative business, not a real one. Forty customers, all on a $29 monthly plan:

$1,160

NRR = $1,184 ÷ $1,160 = 102.1%. GRR = ($1,160 − $116) ÷ $1,160 = 90.0%.

Both look healthy. Now look at how little it takes to move them.

One customer, six points

Two upgrades instead of three is the difference between 102% and 108%. Three cancellations instead of five is five points. None of that tells you whether the product got better or worse; it tells you that 40 is a small number.

It gets sharper when prices differ. Put one customer on a $299 plan beside 39 on $29, and that one customer is 21% of your MRR. If they leave, NRR drops 21 points, more than the ten smallest customers leaving together.

Why monthly NRR misleads on a small base

You can calculate NRR monthly, and Stripe's explainer says it can be monthly, quarterly or annual. On a big base, monthly NRR moves smoothly. On 40 customers, most months nothing happens: no upgrade, no cancellation, and NRR reads exactly 100%. Then one month someone cancels and it reads 97.5%, and the next someone upgrades and it reads 106%.

Each of those moves is one person's decision. A monthly chart of it at 40 customers is flat lines and spikes, and each spike is one person.

Software Equity Group, writing about public companies far bigger than this, makes the same point in general: a smaller customer base "means each customer has a larger impact on overall retention", and a few lost contracts "can disproportionately impact NRR". At 40 customers, each one is 2.5% of the base.

The benchmarks weren't measured on you

You'll see 100% called "good" and 120% called "great". Before comparing, check who was measured.

They measured different businesses, so they don't agree. More useful is why ChartMogul set its floor where it did: "historical retention data is less reliable for very early-stage startups with few customers coming up for renewal."

At $29 a month, 40 customers is about $14,000 a year: under a tenth of ChartMogul's floor and well under 1% of the bottom of SaaS Capital's band. No study I found reports retention for businesses this size, and I'd be wary of any number that claims to.

What to look at instead, at 40 customers

• The events, by name. Who cancelled this quarter, who upgraded, and what each said. At this size a list of people is more informative than a percentage.
• NRR once a year, with its parts. Starting MRR, expansion, contraction and churned MRR, side by side. A 102% made of 4 cancellations and 2 upgrades is a different story from a 102% made of 0 and 0.
• GRR beside NRR. If GRR is falling while NRR holds, upgrades are covering for people leaving.
• Revenue concentration. If one customer is a fifth of your MRR, that's the number to know before any retention rate.

TapRev, which I make, keeps those parts in view: its dashboard's details show the month's MRR movements, churn and your customers, for each platform on its own or all of them together (the details are part of Pro).

By the arithmetic above, single decisions stop moving NRR by whole points only once you have hundreds of customers. Until then, it's a summary of a short list you can read directly.

One more thing before you compare your NRR with anyone else's: the MRR underneath it may be counted differently on each platform you sell on. Why Stripe, RevenueCat and App Store Connect disagree about your MRR walks through the reasons.