The short version: GRR is a stickiness score that cannot lie about churn, and NRR is a growth score that includes the expansion motion. Reading them side by side is the only way to see whether a company is retaining or just upselling faster than it bleeds.
GRR (gross revenue retention) measures only how much recurring revenue you keep from existing customers, so it caps at 100% and exposes churn and downgrades honestly. NRR (net revenue retention) adds expansion revenue from upsells and cross-sells, so it can exceed 100% and often hides churn behind growth. Boards look at both because GRR shows product stickiness while NRR shows commercial momentum.
The five things to know about NRR and GRR before using either one in a board deck, a diligence conversation, or an executive review.
GRR measures what percentage of a cohort's starting ARR you still have one year later, counting only churn and downgrades. Expansion is excluded. The ceiling is 100% because the formula cannot credit growth from existing customers, which is exactly why it is a clean read on product stickiness.
NRR takes the same starting ARR and adds expansion revenue (upsells, cross-sells, price increases) to the ending number before dividing. The result can exceed 100%, and best-in-class SaaS routinely reports 120% or higher. The upside is real, but the number also masks churn when expansion is strong.
GRR = (starting ARR minus churn minus downgrades) divided by starting ARR. NRR = (starting ARR minus churn minus downgrades plus expansion) divided by starting ARR. The denominator is identical. The only difference is whether expansion is counted in the numerator.
A company with 85% GRR and 125% NRR is losing 15% of its base but adding 40% expansion to cover it. The headline NRR looks healthy, the GRR shows the leak. A company with 95% GRR and 102% NRR has sticky customers but a weak expansion motion. Reading only one metric misses half the story.
Investors use GRR to assess product-market fit and switching costs: can the business survive without expansion? They use NRR to value the growth engine: how much does each dollar of retained revenue multiply? Diligence conversations always ask for both, even when the pitch deck shows only one.
SMB SaaS: GRR 80-85%, NRR 95-105%. Mid-market SaaS: GRR 88-92%, NRR 105-115%. Enterprise SaaS: GRR 92-96%, NRR 115-130%. World-class usage-priced platforms report 130-150% NRR with GRR above 95%. Below 70% GRR almost always signals a product or ICP problem, not a sales problem.
Both NRR and GRR are measured over a defined period, almost always trailing 12 months, against a starting cohort of customers. The cohort is the group of accounts that existed at the start of the window. New customers added during the window do not count toward either number. That is important because it keeps retention honest. If new logos were allowed to pad the numerator, a fast-growing company could show 150% retention while losing every customer it signed last year.
The annualized recurring revenue of the cohort on day one of the measurement window. For a trailing twelve month calculation, that is the ARR of every customer on the books one year ago today. This number is fixed. Nothing that happens during the window changes it, which is why the denominator can be trusted.
Logo churn, counted as the ARR of accounts that cancelled during the window. If a $50,000 customer leaves, $50,000 is subtracted from the numerator. Both GRR and NRR include churn, which is why neither number flatters a company with high cancellation rates even when expansion is strong.
Contraction, counted as the ARR lost from customers who stayed but reduced spend. A customer that drops from $80,000 to $60,000 contributes a $20,000 downgrade. Downgrades and churn together give the gross erosion number. Both metrics subtract this fully.
Upsells, cross-sells, seat additions, usage overages, and price increases on existing customers. GRR excludes expansion entirely. NRR includes it. This single input is the whole difference between the two metrics, which is why a company's expansion motion dictates how far the two numbers diverge.
New customer ARR is tracked separately as new ARR or new logo ARR. Mixing it into retention would conflate acquisition with retention and make retention impossible to compare across companies. The discipline of excluding new logos is what makes both metrics useful for benchmarking.
Counting new business as expansion, resetting the cohort when ARR grows, and measuring over too short a window to catch annual contract churn. All three are flagged in diligence. Audit the definitions before signing a term sheet or writing them into a board deck.
The ratio between the two numbers is as useful as either number alone. A wide gap (NRR well above GRR) means the expansion motion is doing heavy lifting. A narrow gap means the business is retaining what it has but not growing within accounts. Both patterns are legitimate, but they imply different operating priorities and different valuations.
If GRR is 80% and NRR is 125%, the business loses 20% of its starting base every year but adds 45% expansion to compensate. The headline growth looks great, and the model works today, but the gross erosion will eventually outrun expansion if left alone. Fix the leak before scaling the expansion motion harder.
GRR at 93% and NRR at 100% means almost nobody leaves, but almost nobody grows either. The product is loved and used, but there is no second-dollar motion. This is where product-led companies often get stuck, and it is the pattern that triggers investment in pricing, packaging, and a dedicated expansion sales team.
GRR above 92% and NRR above 120% is the pattern that gets premium multiples. The product keeps customers, and the land-and-expand motion compounds every account. Snowflake, Datadog, and MongoDB all reported this shape at scale. It is rare, and it is why boards drill into both numbers rather than taking NRR at face value.
GRR below 80% and NRR below 95% means the business is churning faster than it can expand. Growth, if any, is being bought with new logo acquisition. This is a product-market fit question, not a sales execution question, and it usually means the ICP, the pricing, or the core offer needs to be rethought before pouring more into the top of the funnel.
Both metrics should be measured on a trailing twelve month basis, not month over month or quarter over quarter. Annual contracts churn in waves, and shorter windows can create wild swings that misread as improvement or regression. The trailing twelve month view smooths the renewal cycle and reflects how customers actually behave.
Enterprise customers typically retain and expand far better than SMB customers. A blended NRR of 110% can hide an enterprise segment at 135% and an SMB segment at 85%. Always ask for retention by segment, by cohort year, and by product. The blended number is the headline, the segmented numbers are the diagnosis.
The hard part of calculating these metrics is not the formula, it is the data. Starting ARR, churn, downgrades, and expansion all live in different systems: the CRM holds the contract, the billing tool holds the invoices, and the finance team holds the general ledger. Teams that calculate retention monthly without pain do it by pushing every subscription change through the CRM first and letting a renewal pipeline do the bookkeeping.
A separate pipeline in the CRM, parallel to new business, where each upcoming renewal is a deal with an amount, close date, and stage. Renewed, expanded, downgraded, and churned become stage outcomes. The pipeline makes retention visible before it happens, not after the quarter closes.
A rolled-up ARR field on the account record, updated when a renewal or expansion deal closes. Historical values are kept in a snapshot table or audit log, so last year's starting ARR can be queried any time. This single field is the backbone of every retention calculation downstream.
A required picklist on every lost renewal deal: price, product gap, consolidation, acquired, out of business, competitive. Freeform notes capture the detail, but the picklist is what makes churn reportable. Three quarters of churn-reason data gets wasted because it was captured as a text blob nobody can filter.
Expansion deals need the same source and owner fields as new business so the forecast can distinguish rep-driven expansion from product-led expansion. Without the attribution, there is no way to coach the expansion motion or compensate it correctly.
A dashboard that shows GRR and NRR trailing twelve months, broken out by segment, cohort year, and product. The gap chart (NRR minus GRR) is the most useful single visual: it shows at a glance whether expansion is doing the heavy lifting and where retention is actually healthy.
Automated alerts on usage drops, support ticket spikes, exec sponsor turnover, and health score decay. The customer success team gets a flag ninety days before renewal, not thirty. The leading indicator of a churn deal is the behavior that preceded it, and the CRM is where that behavior gets logged.
Strkr runs the whole revenue motion in one tool: new business pipeline, renewal pipeline, account ARR, churn-reason capture, and the retention dashboards that read off all of it. One login, one data model, one trusted number per account.
For SMB SaaS, 95% to 105% is healthy and 110% is strong. For mid-market, 105% to 115% is healthy and 120% is strong. For enterprise, 115% to 130% is the common band and anything above 130% is world-class. Usage-priced platforms sometimes report 140% to 150%. Below 100% means the business is contracting on its existing base, which is a drag that new business has to overcome every quarter.
GRR cannot exceed 100%, so good is measured by how close to 100% a business can stay. SMB SaaS: 80% to 85%. Mid-market: 88% to 92%. Enterprise: 92% to 96%. Public SaaS leaders typically report 95% or higher. Below 80% almost always signals a product-market fit problem, not a sales execution problem, because sales cannot talk customers into staying if the product is not delivering value.
Yes, and it is a common pattern that boards watch carefully. A business with 80% GRR and 125% NRR is losing 20% of its starting ARR to churn and downgrades every year, then adding 45% expansion to cover the gap. The business grows, but the growth is being bought by an aggressive expansion motion on top of a leaky base. The leak is the thing that needs fixing, not the expansion.
Because the GRR formula excludes expansion from the numerator. The best possible outcome is that no customer churns and no customer downgrades, which leaves the ending ARR equal to the starting ARR, which is 100%. By design, GRR cannot credit a business for growing within existing accounts. That is what makes it a clean read on stickiness: the number only moves down, and only in response to real loss.
Neither one alone tells the full story, which is why boards and investors ask for both. GRR is the better signal for product stickiness and switching costs. NRR is the better signal for commercial momentum and the quality of the land-and-expand motion. Reading them together is the only honest read. A company that leads with NRR but will not share GRR is almost always hiding a churn problem.
NRR and NDR are the same metric with two different names. Some companies, especially those that IPO'd in the 2019-2021 cohort, prefer NDR for consistency with their financial filings. The formula is identical: starting ARR minus churn minus downgrades plus expansion, divided by starting ARR. Treat the two acronyms as interchangeable unless a specific company defines them differently in its own disclosures.
Monthly, on a trailing twelve month basis. The trailing twelve month window smooths out the lumpy renewal calendar of annual contracts so the metric does not swing wildly from one quarter to the next. Reporting the number monthly gives operators a steady view of the trend without creating a false signal from a single bad or good month.
No. Both metrics are measured against a cohort that was already on the books at the start of the window. Revenue from customers acquired during the window is tracked separately as new ARR or new logo ARR. Mixing new business into the numerator is one of the most common ways teams fudge the math, and diligence always catches it.
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.
