Gross revenue retention is the share of recurring revenue you keep from an existing cohort of customers over twelve months, before a single dollar of expansion is counted. It subtracts churn and contraction and adds nothing back. That is the whole design. Net revenue retention can be flattered by a good upsell quarter. GRR cannot be flattered by anything, which is why boards see NRR on the slide and GRR nowhere.
The gap between the two is the diagnostic almost nobody runs. A company reporting 116% NRR on 74% GRR and a company reporting 114% NRR on 92% GRR look like twins on a board slide. One is compounding. The other is losing a quarter of its base every year and covering the hole with upsells to whoever is left. Those two profiles are illustrative arithmetic, not a company we audited.
Updated September 2026: the benchmark section now carries the 2026 Aleph x Benchmarkit reading of full-year 2025 data, where the median GRR is 84% rather than the 88% most decks still quote, alongside the published sources that disagree. New sections answer whether GRR can exceed 100%, whether downsells count, and what "gross ARR" means, plus what our own Search Console data shows about how people search this metric.
What gross revenue retention actually is
Gross revenue retention (GRR) measures the percentage of recurring revenue you keep from an existing group of customers over twelve months, before any expansion is counted. It captures one thing: how much of the revenue base you held on to.
The formula:
GRR = (Beginning ARR - Contraction ARR - Churned ARR) / Beginning ARR × 100
Where:
- Beginning ARR is the recurring revenue from a fixed cohort of customers at the start of the period
- Contraction ARR is revenue lost from customers who downgraded, dropped seats, or renegotiated to a lower price but stayed
- Churned ARR is revenue fully lost from customers who cancelled
Notice what is missing. There is no expansion term. Upsells, cross-sells, seat growth, and usage overages do not enter the calculation. GRR only subtracts. That single design choice is why the number is honest.
Can GRR be over 100%?
No. Expansion is excluded by design, so the arithmetic ceiling is 100%, and hitting it would mean the cohort lost nothing at all over the period. Almost nobody does. Every reference that defines the metric puts the cap in the same place, including Wall Street Prep and the Corporate Finance Institute, because upsells and add-ons never reach the numerator.
If you are looking at a gross retention number above 100%, one of three things is true.
- It is NRR wearing the wrong label. This is the common case, and it usually happens when a dashboard vendor or a spreadsheet inherits a definition nobody checked.
- Expansion has leaked into the cohort. An account that downgrades one product and upsizes another can net out to a positive movement if the CRM records a single blended delta instead of two separate events.
- New customers who signed during the window got counted in the beginning cohort, which mixes acquisition into a retention number.
The second and third are calculation faults, and both are covered below.
Does gross retention include downsells and downgrades?
Yes. Contraction is half of what GRR subtracts, and downsell is just another word for it. A seat reduction, a tier downgrade, a renegotiated lower price, a dropped module: each one is contraction, each one pulls GRR down, and none of them requires the customer to leave. Churnkey's breakdown of gross, net and logo retention states the same rule, and so does every formula that separates contraction from churn.
In practice this is where reported GRR quietly goes wrong, because churn is a loud event and contraction is a silent one. A cancellation creates a ticket, a save attempt, an exit conversation, a date somebody logs. A seat count dropping from 40 to 25 at renewal creates a new line on an order form and nothing else. Teams with clean churn data and no contraction data report a GRR that is too high, every quarter, in the same direction.
GRR vs ARR, and what people mean by "gross ARR"
These get typed into the same search box constantly, and they are not the same kind of thing. ARR is a level: the recurring revenue you hold at a point in time. GRR is a rate: how much of a starting level survived twelve months. You do not compare them, you compose them, because GRR is measured on ARR.
"Gross ARR" is a different term again, and a slippery one. In common finance usage it means recurring revenue before losses are applied, while net ARR is what is left after churn, contraction and expansion, as Drivetrain's ARR glossary sets out. The same glossary makes the point that matters more than the definition: organizations define gross ARR differently, so ask which version is in play before you compare two companies' numbers or accept one in a data room.
| Term | What it is | Can it exceed 100%? |
|---|---|---|
| Gross ARR | A level. Recurring revenue before losses are applied | Not a percentage |
| Net new ARR | A level. New plus expansion minus contraction minus churn, for a period | Not a percentage |
| GRR | A rate. Beginning ARR retained, expansion excluded | No, 100% is the ceiling |
| NRR | A rate. Beginning ARR retained, expansion included | Yes, and often does |
If the underlying definitions are what is unclear rather than the retention math, our guide to ARR vs bookings vs revenue covers the layer below this one.
GRR vs NRR: the gap is the whole story
Net revenue retention gets all the attention because it can break 100% and make a growth story look great. GRR gets ignored because it can only ever disappoint. That is exactly backwards.
If you want the full picture of how NRR is built, I wrote a separate breakdown on net revenue retention. The short version: NRR adds expansion on top of what you kept, so it answers "is this cohort growing?" GRR strips expansion out, so it answers "is this cohort leaking?" You need both, and the distance between them is a diagnostic most boards never run.
Both illustrative companies report a similar NRR. On a board slide they look like twins. In reality one is compounding and the other is running to stand still. A wide gap, say more than 25 points, means growth depends on squeezing more out of a shrinking set of accounts, and that is a strategy with an expiry date.
Buyers have worked this out. Advisers who sit in software diligence now describe GRR as the number audited first, because it is the one that cannot be dressed up, with NRR read afterwards as a measure of expansion quality. Livmo's write-up of GRR and NRR in SaaS valuation puts the arithmetic plainly: a company at 110% NRR on 78% GRR has to generate 32 points of expansion from its remaining customers just to reach that headline. Priced as risk, not as a premium.
What good gross revenue retention looks like in 2026
The benchmark moved, and it moved down. Full-year 2025 data across 342 SaaS and AI-native companies puts median GRR at 84%, four points below the 88% that most board decks and blog posts still quote, with every quartile falling rather than just the weak end.
Two caveats on those figures, both worth stating plainly. The report itself is a paid benchmark we could not open, so the numbers here are as reported by The SaaS CFO's summary of the 2026 benchmarks and Aleph's own GRR benchmark page. And other published reads still put the market at 88% to 91%, including Directive's 2026 B2B SaaS retention benchmarks. Sources disagree here, so treat a single median as a rough anchor rather than a pass mark.
Segment matters more than the market median anyway. Enterprise contracts are stickier, harder to rip out, and often multi-year. SMB churns fast because small companies go out of business, switch tools on a whim, and sign month to month. SaaS Capital's private-SaaS retention work makes ACV the primary lens for exactly this reason: gross retention rises as ACV rises, because higher-priced deals come with scoping, implementation and account management that make the product harder to leave. The published segment reads that follow that pattern land around 90% and up for enterprise, high 80s for mid-market, and 80% to 88% for SMB and self-serve.
As a rough read on your own number: if you sell to enterprise, 90% is the floor and best-in-class runs past 95%. Mid-market in the high 80s is fine. SMB in the low-to-mid 80s is healthy, and under 80% for SMB means churn is structural, not a rounding error. If you sell to small businesses and your GRR is 70%, you are not retaining a base, you are refilling a bathtub with the plug pulled.
One caution on benchmarks: your logo retention and your gross revenue retention can tell different stories. You can keep 95% of your logos and still have poor GRR if the 5% you lose are your biggest accounts. Always look at revenue retention, not just logo counts.
Why GRR matters more now than it did in 2021
For a few years, cheap capital and land-and-expand hype made NRR the only number that mattered. Sell a small foothold, expand relentlessly, and let a 130% NRR paper over a mediocre GRR. Investors rewarded the top-line growth and did not ask hard questions about the base.
That era is over, and the benchmark drop is the evidence. The commentary published alongside the 2026 numbers attributes the four-point fall to longer sales cycles, harder ROI scrutiny at renewal, and customers testing AI-native alternatives to tools they already own, and notes that the top quartile fell too, which makes it a market condition rather than a cohort of weak operators. When expansion slows, GRR is what is left holding the business up. A company at 92% GRR survives a flat expansion year. A company at 74% GRR that loses its expansion engine watches revenue shrink with a full sales team hitting quota.
Median gross revenue retention across 342 SaaS and AI-native companies in full-year 2025, down from 88% a year earlier, with every quartile falling. Figures as reported from the 2026 Aleph x Benchmarkit benchmarks.
There is also a valuation angle. When investors underwrite a SaaS business now, GRR is the durability test. High NRR shows upside. High GRR shows the revenue will still be there if the upside does not materialize. In a tougher funding market, durability is what gets priced in.
How people actually search for this number
Here is something we can see that most writing on this metric cannot, because it comes from our own Search Console rather than a keyword tool. Over the 90 days to 10 September 2026, this page collected 562 impressions at an average position of 7.3. Google discloses the query for only 15 of those rows, about 30 impressions, so roughly 95% of the demand arrives anonymized. What the disclosed slice shows is still instructive.
- Seven of the 15 rows are abbreviations or ARR-shaped phrasings: "grr number", "grr vs arr", "arr vs grr", "what is gross arr", "gross arr". Those sit between positions 7 and 11.
- The spelled-out phrase "gross revenue retention" appears exactly once, at position 25.
- The single largest row is not a definition at all. It is the edge case: "can grr be over 100", nine impressions at position 7.6. Right behind it is "does gross retention include downsell", at position 30.
The practical read for anyone writing a metrics definitions document or a board appendix: people arrive at this number through the abbreviation and through the edge cases, not through the textbook phrase. Define GRR both ways, spell out whether downgrades count, and say what the ceiling is, because those are the questions being typed. That is also why this page now answers all three above the calculation section rather than burying them in an FAQ. One site's search data is one site's search data, and the sample here is small, so treat it as a signal rather than a study.
How to calculate GRR without fooling yourself
The formula is simple. The mistakes are where teams quietly inflate the number.
First, fix your cohort. GRR measures a specific group of customers who existed at the start of the period, tracked forward twelve months. New customers who signed during the window do not belong in the calculation. If you let them in, you are mixing acquisition with retention and the number means nothing.
Second, never let expansion sneak in. This sounds obvious, but I have seen CRMs where an account that downgraded one product and upsized another nets out to zero contraction, hiding the churn. GRR needs contraction and expansion tracked as separate movements on the same account, not a blended delta.
Third, decide how you handle downgrades that happen mid-term versus at renewal, and apply it consistently. A seat reduction in month three and a price cut at renewal are both contraction. Pick a rule and stick to it every quarter, or your trend line is noise.
Fourth, watch the annualization. If you report GRR monthly and multiply, small measurement errors compound into a very wrong yearly figure. Measure the cohort across a real twelve-month window where you can.
Most teams get this wrong not because the math is hard but because the CRM data underneath it is a mess. Contraction is rarely logged cleanly, downgrades get recorded as edits rather than events, and churn dates drift. If your customer success operations are not capturing these movements as structured data, your GRR is a guess dressed up as a metric.
Where gross revenue retention actually breaks
Low GRR is a symptom. The disease is almost always one of a few specific failures, and they show up in the data if you know where to look.
The most common is onboarding. When a customer never reaches first value, they churn at renewal no matter how good the sales pitch was. A weak start is the single biggest predictor of a lost renewal, which is why I treat customer onboarding as a retention lever, not an admin task.
The second is the renewal itself being run as a fire drill. If your team finds out an account is at risk 30 days before the renewal date, it is already gone. Renewals are won 120 days out, in the health data, not in the panic window. I broke down that whole motion in the renewal management playbook.
The third is no early warning system. Teams that retain well can see churn coming because they track usage, support load, and sentiment as a live customer health score. Teams with bad GRR find out an account is unhappy when the cancellation email arrives.
If the churn showing up in your GRR is broad rather than concentrated, the SaaS churn rate benchmarks guide is the better starting point, because it works the problem from the customer count side.
How to fix low GRR: the RevOps build
Fixing GRR is not a customer success pep talk. It is a system that catches revenue risk early and routes it to the right person while there is still time to act.
The point of the loop is that GRR stops being a number you report after the damage is done and becomes a number you can move. Most of the work is data plumbing: getting contraction and churn logged cleanly, wiring a health score off real signals, and making sure a falling account lands on someone's desk with enough runway to save it. That is a RevOps job, and it is exactly the kind of thing we build in our CRM and RevOps engagements.
Retention and expansion are two builds, not one. Once the base holds, the land and expand playbook is where the other half of the NRR number gets made.
Not sure what your real GRR is?
Book a free 30-minute audit and we will walk your cohort data with you and show you the retention number your board deck is hiding.
Book an audit →The number to put next to NRR
If your board deck shows NRR and not GRR, you are telling half the story, and it is the flattering half. Put both numbers on the same slide. Show the gap. A 114% NRR built on 92% GRR is a business that compounds. A 116% NRR built on 74% GRR is a business betting it can keep upselling faster than the base leaks, which is a bet that gets harder every quarter, and harder still in a year when the median gross retention in the market fell four points.
Gross revenue retention is not the exciting number. It is the honest one. And in a market where expansion is no longer a given, honest is what keeps you alive.
FAQ
What is a good gross revenue retention rate?
Judge it by segment, not against a single market number. Enterprise SaaS should sit at 90% or higher, with best-in-class above 95%. Mid-market lands in the high 80s. For SMB, low-to-mid 80s is healthy and anything under 80% signals a structural churn problem. For context, the 2026 Aleph x Benchmarkit benchmarks put the median across 342 SaaS and AI-native companies at 84% for full-year 2025, with the top quartile at 91% and the bottom at 76%. Other published reads still put the market median at 88% to 91%, so anchor on your own segment and your own trend.
Can gross revenue retention be over 100%?
No. GRR excludes all expansion revenue by design and only accounts for losses from churn and downgrades, so 100% is the ceiling and it would mean you lost nothing from the existing base over the period. A gross retention number above 100% is almost always NRR mislabeled, expansion leaking in through blended account deltas, or new customers wrongly counted in the beginning cohort.
Does GRR include downgrades and downsells?
Yes. Downgrades, seat reductions and renegotiated lower prices are contraction, and contraction is subtracted from GRR alongside full churn. The customer staying does not matter to the calculation, only the revenue movement does. This is the most common source of an overstated GRR, because contraction is usually logged as an edit to a record rather than as an event.
How is gross revenue retention different from net revenue retention?
GRR only subtracts churn and contraction from a cohort, so it can never exceed 100%. It measures how much of your base you kept. NRR adds expansion on top, so it can go above 100% and measures whether the cohort grew. GRR is the floor, NRR is the floor plus expansion.
What is the difference between GRR and gross ARR?
They are different kinds of measure. GRR is a rate, the percentage of a starting cohort's ARR that survived twelve months. Gross ARR is a level, generally meaning recurring revenue before losses are applied, against net ARR as the figure left after churn, contraction and expansion. Definitions of gross ARR vary between companies, so confirm which one is being used before comparing numbers.
Why does the gap between NRR and GRR matter?
The gap shows how much of your growth depends on expansion versus retention. A small gap means your base is solid and expansion is a bonus. A wide gap, more than 25 points, means you are covering heavy churn with aggressive upsells to the customers who remain. That works until expansion capacity runs out, then revenue drops fast. It is also the first thing a buyer or investor tests in diligence, because gross retention is the harder number to dress up.
How do I improve gross revenue retention?
Fix the three things that drive churn: weak onboarding that delays first value, renewals run as last-minute fire drills, and no early warning system for at-risk accounts. Build a health score off real usage and engagement data, flag risk 120 days before renewal, and route every at-risk account to a named owner with a play. Measure GRR by segment so you know where the leak actually is.