NRR Meaning: Calculate the Cohort Change

Imagine a subscription business that finishes a quarter with more recurring revenue than it had at the start. The increase looks reassuring until someone points out that several customers canceled. Did existing customers become more valuable, or did new sales merely replace the revenue that disappeared? Net revenue retention, usually shortened to NRR, answers the first part of that question. It isolates the customers already present at the start of the period and follows what happened to their recurring revenue.

That distinction matters whenever a growing total hides movement inside the customer base. A company can add enough new accounts to raise overall recurring revenue while its original accounts spend less. It can also lose some original customers yet finish with more revenue from that same group because other accounts expanded. NRR describes the net result for the original group. To understand whether that result is durable, a reader also needs to see the losses and gains beneath the percentage.

NRR follows the same customers from start to finish#

Net revenue retention is the ending recurring revenue from a starting group of customers, divided by that group’s opening recurring revenue and expressed as a percentage. The group is called a cohort. The SaaS Metrics Standards Board’s definition measures the same customers at the beginning and end of a period; customers acquired after the opening date stay outside that comparison. If opening customers produced $100,000 in monthly recurring revenue and those same customers produced $105,000 at the close, NRR is 105%.

The word net signals that gains from the cohort are offset by its losses. An existing customer might buy additional capacity, increasing recurring revenue. Another might downgrade, reducing it. A third might leave, taking its recurring revenue to zero. These are expansion, contraction, and churn, respectively, in Stripe’s explanation of net dollar retention. The metric combines all three movements instead of counting only whether each customer remains.

The opening customer list must stay fixed. Suppose a company starts a quarter with 100 customers, adds 20 during the quarter, and loses 10 of the original 100. The 20 arrivals may lift company revenue, but their revenue does not belong in that quarter’s NRR numerator. The original 10 customers who left remain in the cohort with zero ending recurring revenue. Removing them from the calculation would turn a loss into an apparent improvement. The cohort method described by the standards board keeps those opening customers in view through the closing date.

NRR is a revenue measure, not a count of customers. A small account and a large account have different weights because their opening recurring revenue differs. Losing one large account can outweigh several small upgrades; one large expansion can conceal many cancellations. Neither possibility makes the percentage wrong. It means the percentage answers a narrower question than “How many customers stayed?” When a report shows a striking NRR figure, ask what happened to the customer count and to the separate revenue movements before drawing a conclusion about retention.

Calculate NRR from the opening revenue and its movements#

The calculation starts with recurring revenue from the opening cohort. Add expansion from those customers, subtract contraction from customers who still buy, and subtract revenue lost when cohort customers leave. Divide the result by the opening cohort revenue and multiply by 100 to express it as a percentage. Stripe sets out this opening revenue, expansion, contraction, and churn calculation. The same answer comes from dividing the cohort’s ending recurring revenue directly by its opening recurring revenue, provided both figures use the same customer list and revenue definition.

Consider an illustrative business with three customers and $100,000 in opening monthly recurring revenue. Customer A begins at $50,000, customer B at $30,000, and customer C at $20,000. By the end of the chosen period, A pays $80,000, B pays $25,000, and C has canceled and pays nothing. A contributed $30,000 of expansion; B contracted by $5,000; C churned $20,000. Ending monthly recurring revenue from the three original customers is $105,000, so NRR is $105,000 divided by $100,000, or 105%.

A second calculation makes the movements easier to see: begin with $100,000, add A’s $30,000 expansion, take away B’s $5,000 contraction, and take away C’s $20,000 churn. The result is still $105,000. The example is deliberately concentrated. One account’s $30,000 increase more than covers another account’s cancellation and a third account’s downgrade. The 105% figure is arithmetically strong, but it does not mean that all three customer relationships improved. Two did not.

The unit and period need to be explicit. In this example, every amount is monthly recurring revenue measured at two points in a chosen period; the calculation compares the monthly run rate at those points. If a report instead uses a different recurring revenue basis, its opening and closing amounts must be calculated consistently. A percentage labeled only “NRR” leaves a reader unable to tell which customers, dates, and recurring revenue definition produced it. Those details matter most when someone compares two periods or two companies.

The 100% line has a precise, limited meaning#

At 100% NRR, the opening cohort finishes with exactly as much recurring revenue as it began with. Expansion may have offset contraction and churn, or there may have been no change at all. Those situations carry different business implications even though the headline figure is identical. Below 100%, the cohort’s losses exceed its expansion. Above 100%, expansion exceeds its losses, so the cohort ends with more recurring revenue than it had at the start. Stripe explains this distinction when comparing net and gross retention.

Return to the three customer example. Its 105% NRR means the original group ends $5,000 per month above its $100,000 starting level. It says nothing by itself about revenue from customers acquired during the period, so it is not the company’s total recurring revenue growth rate. It also says nothing about whether the $30,000 expansion from customer A will repeat. NRR records the change between two chosen points; it does not promise the next period will behave the same way.

A 95% NRR has an equally concrete interpretation: the starting cohort ends with 95 cents of recurring revenue for each dollar it produced at the beginning. The missing five cents could reflect many modest downgrades, one large cancellation, or losses partially offset by upgrades. Before choosing a response, identify which movement caused the shortfall. Treating every below-100% result as the same problem can send attention to the wrong customers.

The 100% threshold is therefore a useful reading aid, not a complete quality grade. Whether 105% is reassuring depends partly on how much was lost and where the expansion occurred. Likewise, 95% tells a team that existing-customer revenue shrank; it does not tell the team whether product usage, a contract change, or a particular customer segment explains the shrinkage. Those questions require account-level information beyond the NRR percentage.

GRR shows how much revenue the cohort kept before expansion#

Gross revenue retention, or GRR, starts with the same opening cohort but excludes expansion. It measures the share of opening recurring revenue left after contraction and churn. Stripe’s comparison of NRR and GRR describes that difference: NRR includes gains from existing customers, while GRR leaves those gains out. With the usual opening-revenue denominator, GRR cannot be lifted by a successful upgrade. That makes it particularly useful alongside NRR when expansions may be covering losses.

For the illustrative three customers, opening recurring revenue is $100,000. Customer B’s $5,000 downgrade and customer C’s $20,000 cancellation remove $25,000 of that opening revenue. GRR is therefore 75%. NRR is 105% because A’s $30,000 expansion more than restores the $25,000 loss. Put together, the figures say something more useful than either alone: the cohort grew in net dollars, but it lost one quarter of its opening revenue before expansion.

Now imagine another illustrative cohort that also begins at $100,000 and ends at $105,000. In this version no customer contracts or leaves, and existing customers add $5,000. NRR is again 105%, but GRR is 100%. The two cohorts have the same net outcome and very different paths to it. If the question is “Did existing customers spend more overall?”, either 105% answers yes. If the question is “How much of the original revenue survived without help from upgrades?”, the first cohort’s 75% GRR and the second cohort’s 100% GRR give sharply different answers.

NRR and GRR still do not tell a manager why a customer expanded or left. An upgrade could follow a customer’s growing needs; a downgrade could follow a change in those needs. The percentages classify the revenue movement. They cannot, on their own, establish the customer’s reason. I would look at the individual accounts behind a wide NRR–GRR gap before celebrating the NRR figure or prescribing a retention campaign.

Customer-count retention can point the other way#

The same three customer example also shows why customer retention by count is a separate measure. Two of the three opening customers remain at the close, so customer-count retention is two out of three. Yet recurring revenue from the opening customers rose from $100,000 to $105,000, giving 105% NRR. Customer C’s departure lowered the count; A’s larger contract lifted the revenue total. Stripe distinguishes revenue retention from customer-count retention, and the difference grows especially important when accounts vary widely in size.

A company could also keep nearly every customer but suffer lower NRR if many of those customers reduce their recurring spend. A retained customer is not necessarily a retained dollar. Conversely, an NRR above 100% does not establish that most customers expanded. The aggregate can be driven by a few large accounts. A reader who wants to know whether the customer base is broadening or narrowing needs the count of retained customers as well as the revenue weighted figure.

This distinction changes the practical conversation. If NRR rises while the number of retained customers falls, ask which accounts expanded and which customers left. If customer-count retention holds steady while NRR falls, inspect downgrades and the opening revenue attached to the lost accounts. The goal is to locate the movement that matters, not to force the two percentages to agree. They measure different things, so disagreement is information.

Keep the calculation comparable before interpreting a trend#

A reported NRR trend is only meaningful when its cohorts and revenue rules are clear. For each period, identify the customers present on its opening date and compare those customers’ opening and closing recurring revenue. Do not add a customer who first bought during the period simply because the customer later expanded. Do not drop a canceled customer from the closing list. The standards board’s fixed-cohort calculation makes these boundaries explicit.

The same care applies to what counts as recurring revenue. If one period’s opening figure includes a kind of recurring charge that the next period excludes, the percentages cease to be directly comparable. A changed classification can make NRR move even when customer behavior has not changed in the same way. Before interpreting a jump or drop, read the metric’s stated revenue definition and confirm that the closing revenue is measured on the same basis as the opening revenue. If the definition changed, explain that change beside the percentage.

The timing of a downgrade or cancellation can also change which period carries the loss. For that reason, a report should name the opening and closing dates, the recurring revenue basis, and the treatment of customer changes that occur near a boundary. These choices do not have one universal answer in the supplied definitions; what matters for interpretation is that the choice is stated and applied consistently. Otherwise, a neat sequence of percentages may invite a comparison the underlying data cannot support.

If a report uses the label net dollar retention instead of NRR, read its definition rather than assuming a new calculation. The SaaS Metrics Standards Board lists net dollar retention as an alternative name for net revenue retention. Labels alone cannot tell a reader which revenue basis, cohort, or period a particular company has chosen. Those details belong beside the figure whenever it is used to make a decision.

Let the components guide the next question#

For a first reading of an NRR figure, I would ask for four numbers: the starting cohort’s recurring revenue, its expansion, its contraction, and its churn. Those amounts recreate the reported percentage and show which direction deserves attention. If churn dominates, inspect the customers who left and the revenue attached to them. If contraction dominates, inspect continuing customers whose spend fell. If expansion dominates, check whether the increase came from many accounts or a single large one. These are different situations even when they produce the same NRR.

I would then put GRR and customer-count retention beside NRR. GRR shows how much opening revenue survived before expansion; customer-count retention shows how many opening relationships survived. Together, the three measures help prevent a strong net percentage from obscuring a fragile base. The trade-off is a little more reporting detail, but the alternative is a single number that can hide the very losses a team needs to address.

NRR’s meaning is straightforward once the boundary is fixed: it is the percentage of opening recurring revenue still generated by the same customers at the close, after their expansions, contractions, and cancellations. Its best use is equally specific. Read it as the net change in existing-customer revenue, then look at the movements beneath it before deciding what that change says about the business.