Net revenue retention is the single most predictive signal of a SaaS business worth owning. It captures how much recurring revenue a cohort of customers kept, grew, and lost over a defined window, net of everything that happened inside the base. This guide walks the full calculation: define the cohort, pull starting ARR, roll in expansion, strip out contraction and churn, compute NRR, drop the expansion to compute GRR, segment the result, benchmark it, and operationalize a monthly refresh that finance and the board actually trust.
Time: half-day to design, monthly refresh
Start by defining who is in the base. The cohort is every customer that existed at the start of the measurement window and is held constant for the entire calculation. Do not add new logos closed inside the window; new business belongs in gross new ARR, not in retention. Lock the customer list in writing before pulling any numbers so finance and GTM are literally working from the same roster. For most SaaS companies the default window is trailing twelve months, which smooths out seasonality and renewal cycles. If the business has quarterly contracts and faster motion, a trailing six-month cohort can work, but keep the window stable quarter over quarter so trend analysis is comparable.
Pull the recurring revenue that cohort was contributing on day one of the window. This is the denominator for both NRR and GRR, and it must stay fixed for the full calculation. Use ARR, not billings, because billings reflect invoicing cadence and will distort the ratio for annual-paid-monthly customers. Strip out one-time services, implementation fees, and anything that is not recurring. If the business sells usage-based or hybrid contracts, pick the normalized ARR convention finance already uses in the P&L and apply it consistently. The starting ARR number is the baseline every subsequent movement is measured against, so it has to be defensible under audit.
Expansion is every recurring dollar the cohort added inside the window: seat upsell, tier upgrades, cross-sell of new products or modules, and usage-driven increases on committed plans. The rule is simple: the dollar has to be recurring, and the customer has to already have been in the cohort. New logos never count as expansion. Pull expansion from the subscription or billing system so each dollar is sourced from a specific change event, not an aggregate. Tagging matters here. If upsell and cross-sell are both lumped as expansion with no sub-category, the team loses the ability to answer whether growth is coming from existing products going deeper or from new products landing.
Contraction is every recurring dollar the cohort lost inside the window without the logo churning entirely: downgrades, seat reductions, module removals, and plan step-downs. Treat it as its own bucket, separate from churn, because the diagnostic is different. Contraction usually points to a product value or pricing-fit problem inside otherwise healthy accounts, whereas churn points to a relationship or outcome problem. Pull it from the same subscription change log as expansion so the numbers tie. Sign discipline matters; contraction is a negative movement against the starting ARR, so lock the convention early and keep it consistent across every dashboard that reads from this calculation.
Churn is every logo in the cohort that fully stopped paying inside the window. The dollars that come out are the full starting ARR of each churned account, not the amount at the churn date. This is where teams quietly inflate retention: if you measure churn at the point the account exits, a customer that was already contracting for six months looks smaller on the way out and the retention number looks healthier than reality. Hold churn at starting-cohort ARR. Then separately report the contraction those accounts experienced before churn as a leading indicator. Non-renewal, cancellation, and complete down-sell to zero all count as churn.
With the four buckets locked, the math is one line each. NRR equals starting ARR plus expansion minus contraction minus churn, divided by starting ARR, expressed as a percentage. GRR is the same math minus the expansion term: starting ARR minus contraction minus churn, divided by starting ARR. NRR can exceed 100 percent because expansion can outweigh contraction and churn; GRR is capped at 100 percent by definition because it ignores upside. The gap between NRR and GRR is the expansion engine. A healthy business shows a wide gap, because expansion is doing real work against the retention drag. A narrow gap flags a product or motion that is holding on but not growing inside the base.
A single company-level NRR number tells you nothing you can act on. Immediately cut it by segment, plan, and signup cohort so the pattern underneath becomes visible. SMB cohorts typically churn more and expand less; mid-market and enterprise cohorts typically churn less and expand more. If the SMB segment is dragging NRR down twenty points, the fix is a different motion, not a pricing tweak. Cut by product or plan to see whether a specific plan tier is retaining badly. Cut by signup cohort to see whether a particular launch quarter is aging better or worse than the ones around it. Segmentation is where NRR goes from a board slide to an operating lever.
A retention ratio in isolation is just a number. Benchmark it against external SaaS retention research before you draw conclusions. The commonly cited bands put NRR in the 100 to 110 percent range for healthy SaaS, 110 to 120 percent for strong, and 120 percent and up for world-class. GRR tends to run 85 to 90 percent for healthy mid-market and above 90 percent for enterprise. Rebuild the calculation every month on a trailing twelve-month window so the number refreshes as the cohort ages. Push both numbers into the finance review, the board deck, and the CS and sales dashboards from the same source so nobody is quoting a different value in a different meeting. One number, one source, every month.
NRR = (starting ARR + expansion - contraction - churn) / starting ARR, expressed as a percentage. Starting ARR is the recurring revenue of the cohort on day one of the window, and expansion, contraction, and churn are the recurring movements inside that cohort over the measurement period. New logos closed inside the window never enter the formula.
NRR includes expansion; GRR does not. GRR = (starting ARR - contraction - churn) / starting ARR and is capped at 100 percent by definition. NRR can exceed 100 percent when expansion more than offsets contraction and churn. The gap between the two numbers is the expansion engine, and a wide gap is a sign the land-and-expand motion is doing real work inside the base.
Commonly cited bands put NRR in the 100 to 110 percent range for healthy SaaS, 110 to 120 percent for strong, and 120 percent and above for world-class, with GRR typically 85 to 90 percent for healthy mid-market and above 90 percent for enterprise. The right benchmark depends on segment, deal size, and motion, so match the comparison population tightly before drawing conclusions.
Rebuild NRR and GRR monthly on a trailing twelve-month cohort so the number refreshes as the cohort ages. Monthly is frequent enough to catch trend breaks early, and the trailing-twelve-month window smooths out quarterly renewal spikes and billing cadence noise. Report both numbers in the same monthly finance and board cadence so there is only one authoritative version in circulation.
No. NRR is a retention metric and only measures what happened to the cohort of customers that existed at the start of the window. New logos closed inside the window belong in gross new ARR, not in expansion. Mixing new business into the retention ratio is the single most common way companies inflate their reported NRR, and it collapses the moment an investor asks for the cohort definition.
Contraction is every recurring dollar lost from a customer that is still paying something: downgrades, seat reductions, plan step-downs, and module removals. Churn is every logo that fully stopped paying inside the window. Hold churned accounts at their starting-cohort ARR, not their final ARR before exit, so months of pre-churn contraction do not quietly disappear into a smaller churn number.
Strkr ties customers, plans, expansion, contraction, and churn to one source of truth so finance and GTM can rebuild NRR and GRR every month from the same numbers without a spreadsheet reconciliation.
No sales call, no migration consultant, no four-month implementation. Enter your card, get 14 days of the full Pro tier, cancel any time before day 14 with zero charge. Spin up a workspace, import your CSV, and have something useful before lunch.
