Skip to content
Churn & retentionMetric 10 of 35

NRRNet Revenue Retention

What % of last year's revenue you still have, including expansion.

Abderrahim Ben Jazia, PhDBy · Founder of FlowMRR, previously built and sold AIS (Avance Immédiate Services), French tax-credit softwareUpdated:

Put this metric to work: explore FlowMRR's MRR retention analytics to track revenue movements, retention and forecasts from your Stripe account, or see how FlowMRR compares as a Baremetrics alternative for Stripe.

The revenue kept and grown from one fixed group of customers over a year. FlowMRR takes every customer paying twelve months before the last closed month (one person per email, however many Stripe customer records they have) and compares what they pay now with what they paid then. Expansion counts; customers acquired since are left out, and a customer who left and came back counts at what they pay now. Above 100%, existing customers grow revenue even with zero new sales. With less than 13 months of history, there is no figure yet.

Formula

NRR = Today's MRR of the customers who were paying 12 months ago ÷ What those customers paid 12 months ago

Worked example

12 months ago the cohort paid $100K. Since then: +$12K expansion, −$2K contraction, −$4K from customers who left → the same customers now pay $106K → NRR = 106 ÷ 100 = 106%.

Other definitions you will see

Many tools compute NRR from one month's movements: (opening MRR + expansion − contraction − churn) ÷ opening MRR. That monthly figure sits close to 100% and is not comparable with an annual one; compounding it over twelve months, or adding up twelve months of movements, also lets customers acquired during the year into the later months' bases. Definitions also differ on customers who left and came back during the year: FlowMRR counts them at what they pay at the end.

Illustrative comparisons

B2B SaaS, annualised

These unsourced ranges are illustrative scenarios, not measured industry benchmarks, percentiles or recommended targets. Their qualitative labels are editorial interpretations. Compare your own cohorts using the same segment, period, currency and measurement basis.

Higher retention
> 100%
Healthy
90–100%
Leaking
< 90%

Why it matters

NRR above 100% is one of the most watched metrics in SaaS. It means your existing book is a growth engine: revenue grows before any new sale, and every new customer can become a compounding asset. Investors read it as the share of future growth your current customers already carry.

Common mistakes

  • Calculating on a single month's snapshot rather than year-over-year cohort comparison.
  • Including new logos in the numerator (NRR is about existing customers only).
  • Reporting NRR without GRR — they tell different stories.

Calculating it from Stripe data

Start from subscriptions, not invoices or payouts. For each subscription, take the recurring price × quantity, deduct permanent (forever) discounts and normalise it to a month: an annual price ÷ 12, a quarterly price ÷ 3. Leave out subscriptions that are trialing, incomplete, incomplete_expired or paused, because none of them is paying. Keep past_due subscriptions while Stripe is still retrying the card, and treat unpaid ones as churned at the end of the last period they actually paid for. A coupon that expires after a few months raises the bill without the customer buying anything, so counting only forever discounts keeps that from showing up as expansion.

Then fix the cohort. Take every customer who was paying in the starting month, twelve months before the month you measure, and record their MRR at that time. Twelve months later, sum what those same customers pay, whichever products they are on. Customers who first paid after the starting month stay out, however large they are. A customer who cancelled and came back within the year counts at whatever they pay at the end of the window.

Sum MRR per customer, not per subscription, and group Stripe customer records that belong to the same buyer: a second workspace or a re-created customer otherwise looks like churn plus a new sale, and NRR drops for no real reason. With several currencies, convert both months at the same exchange rate so currency moves do not read as expansion or contraction. A single 12-month reading needs at least 13 months of history; FlowMRR computes this cohort NRR from your Stripe history and shows nothing until a 12-month-old cohort exists.

Reading the result

NRR moves with three forces: accounts that leave, accounts that downgrade, and accounts that expand through seats, usage tiers or higher plans. Because expansion sits in the numerator, the same NRR can describe very different businesses: no churn and no expansion gives 100%, and so does losing 10% while gaining 10%. Read it next to GRR to know which one you have.

A cohort NRR covers a full year, so it reacts slowly: a pricing change made last month barely shows. When it drops, break the cohort down by plan, starting size and acquisition month, and check whether the fall comes from a few large accounts or many small ones. If one or two expanding customers hold NRR up, treat that as concentration risk rather than strength. Churn and contraction are retention problems (onboarding, payment recovery, product fit); thin expansion is usually a packaging problem, with no seat, usage or tier path for customers to grow into.

Frequently asked questions

What is a good NRR for SaaS?

There is no single good number. NRR depends on whether your pricing lets customers grow (seats, usage, tiers) and on customer size, since larger accounts tend to churn less and expand more. Above 100% means existing customers grow revenue on their own; compare with companies of similar size, pricing model and segment, measured on the same 12-month cohort basis.

How is NRR different from GRR?

Both start from the same cohort and the same starting MRR. NRR adds expansion, so it can exceed 100%; GRR ignores expansion and caps each customer at their starting revenue, so it never exceeds 100%. The gap between the two is how much expansion is offsetting losses.

Can NRR be calculated monthly?

Yes, but a monthly NRR is a different figure: it sits close to 100% because one month's losses are small, and it is not comparable with the annual figures people usually quote. A true 12-month cohort that follows the same customers through the year is the comparable measure.

Do new customers count in NRR?

No. Only customers already paying at the start of the window are in the cohort. Revenue from customers acquired during the window is excluded even if they expand quickly; it appears in the NRR measured a year later.

Contact us
FlowMRR

Let's talk

A question, a technical issue or an idea for a partnership? Write to the FlowMRR team.