Net revenue retention explained with a worked example

Net revenue retention (NRR) answers one question: of the recurring revenue you had at the start of a period, how much do you have at the end, from the same customers? It counts upgrades as gains and downgrades and cancellations as losses. It ignores customers who joined during the period.

The four pieces

Take the monthly recurring revenue (MRR) you had at the start of the period, from the customers you had then. During the period, four things can happen to that revenue:

  • Expansion: customers upgrade, add seats or buy more.
  • Contraction: customers downgrade or cut seats but stay.
  • Churn: customers cancel completely.
  • No change: everyone else.

The formula

NRR = (starting MRR + expansion − contraction − churn) ÷ starting MRR × 100

An NRR of 100% means the customers you started with are paying you exactly as much as before. Above 100%, they pay more in total. Below 100%, they pay less.

A worked example

At the start of the month you have 10,000 of MRR. During the month:

  • Expansion: 2,000
  • Contraction: 500
  • Churn: 1,000

NRR = (10,000 + 2,000 − 500 − 1,000) ÷ 10,000 × 100 = 105%

The customers you had at the start now pay 10,500 in total.

Some teams count downgrades inside expansion, as a “net expansion” of 1,500, and write the formula as (10,000 + 1,500 − 1,000) ÷ 10,000. The answer is the same, 105%. The point is to use one method every time.

New customers are left out

NRR is about the customers you already had. Revenue from customers who joined during the period does not count, however large.

Here is the same month, customer by customer. Customers A to E were there at the start. F and G joined during the month.

CustomerStart → endWhat happened
A4,000 → 5,500expansion 1,500
B3,000 → 3,000no change
C1,500 → 1,000contraction 500
D1,000 → 0churn 1,000
E500 → 1,000expansion 500
F0 → 2,000new
G0 → 1,000new
Total10,000 → 13,5003,000 of it new

Total MRR went from 10,000 to 13,500. But 3,000 of that comes from F and G. The customers who were there at the start (A to E) now pay 10,500. That is why NRR is 105%, not 135%.

Three scenarios

Same starting MRR of 10,000 each time:

ScenarioChangesNRR
Only lossesChurn 1,00090%
The example aboveExpansion 2,000, contraction 500, churn 1,000105%
Strong expansionExpansion 3,000, contraction 500, churn 500120%

The middle row shows why NRR can pass 100% even while some customers leave: upgrades from the rest of your customers more than make up for the loss.

The period matters

The example above covers one month. A monthly NRR of 105% is not the same as an annual NRR of 105%. If you quote NRR, say what period it covers, and compare like with like.

One way to see what a monthly figure would mean over a year: if the same 105% repeated for 12 months, the starting customers would pay 179.59% of what they pay today (1.05 to the power of 12). This is our own calculation, not a published formula. It assumes the monthly rate never changes, which is rarely true, so treat it as an illustration and not a forecast. At 90% a month, the same calculation gives 28.24%.

What this does not tell you

  • Whether your NRR is good. We give no benchmark. The right comparison is your own NRR over time.
  • Where growth comes from. NRR ignores new customers. Combine it with how much you are adding.
  • Customer-level detail. One large customer upgrading can lift NRR while many small ones leave.
  • Different definitions. Teams differ on how they treat downgrades, one-off charges and customers who leave and come back. Write down your rule and keep to it.

Try it with your own numbers

The net revenue retention calculator takes your starting MRR, expansion, contraction and churn, and shows NRR. To look at the cancellation side on its own, see the churn rate calculator. To see how total MRR grows, try the MRR growth calculator.

Sources

  1. Orb: How to calculate NRR (opens in a new tab) (formula, the 10,000 example, new customers excluded)
  2. The 12-months-at-the-same-rate illustration is Webtilly’s own.

W

Written by the Webtilly team

Formula and sources last checked: 26th Oct 2026 · How we check our work