Key Takeaways
- Gross revenue retention measures how much recurring revenue survives a year with no help from expansion — and the SaaS-wide median fell to 84% in 2025, down from 88% the year before.
- A healthy net revenue retention number can sit directly on top of a shrinking customer base, because expansion revenue from a handful of large accounts can cover for churn everywhere else.
- GRR isn't one metric with a single owner — it blends four distinct failure modes: logo churn, contraction, support-driven attrition, and self-serve gaps.
- Staffing more customer success headcount doesn't reliably move the number — research on 132 companies found no meaningful correlation between CS staffing ratio and retention.
- A four-point GRR decline on a $10M ARR base costs roughly $400,000 in lost recurring revenue a year — before a single new dollar gets booked.
- Separating logo-churn GRR from contraction GRR in your own reporting, before finance formalizes it, is the fastest way to route each fix to the right team.
The Number Every Board Slide Skips
Gross revenue retention measures how much of your starting recurring revenue survives a year with no help from upsells or expansion. Most board decks skip it entirely.
That's not an oversight. Net revenue retention counts expansion, so it almost always looks better. A company can show strong net revenue retention while its actual customer base is quietly shrinking underneath the accounts that grew. Gross revenue retention doesn't allow that flattery — the ceiling is fixed at what you started with, and there's nowhere to hide an expansion boost.
The reason to care right now: the median gross revenue retention across 342 B2B SaaS and AI-native companies fell to 84% in 2025, down four points from 88% the year before. That's not one struggling segment. That's the market median.
For a customer enablement or support operations leader, that shift changes the conversation. You're no longer defending against a slow leak. You're accounting for a faster one, and the fixes that worked when GRR sat at 88% may not be enough now.
What Gross Revenue Retention Actually Measures
Before you build anything on this number, it helps to know what it excludes on purpose — and where practitioners quietly disagree about what counts as a loss.
The formula, and the two things it deliberately throws away
Gross revenue retention starts with recurring revenue at the beginning of a period, subtracts what you lost to cancellations and downgrades, and divides by that starting number. Expansion revenue — upsells, cross-sells, seat growth — never enters the calculation. That's deliberate. GRR is built to answer one question only. Did you keep what you already had?
The ceiling is fixed by definition. You can hold onto every dollar you started with, but you can never gain more through this formula alone. A company that keeps every existing customer and adds nothing new still scores at the top of the range. A company that adds massive expansion revenue but loses much of its original customer base can still show a weak score here, even while its blended growth number looks strong.
Where practitioners disagree — dollars vs. logos, and what counts as "churn"
Two definitional splits matter more than most explainers admit. First, GRR is sometimes measured in customer count rather than dollars, which turns it into logo retention — a different number that answers a different question about how many accounts you kept, not how much revenue you kept.
Second, formula variants disagree about what belongs inside "churn." Some treat downgrades and cancellations as two separate subtracted terms. Others fold contraction directly into a single revenue-churn bucket. That gap matters because different published formulas can build the same 84% or 90% headline number from different underlying assumptions about what counts as loss. Before you compare your number to a benchmark, confirm the benchmark used your definition of churn — not just its label.
The 2025-2026 Reset: Why the Old Benchmarks No Longer Apply
If your target for gross revenue retention was set using last year's numbers, it's already out of date. The shift wasn't gradual. It landed across every percentile at once.
What moved, and why it moved everywhere at once
The 2025 Benchmarkit SaaS Performance Metrics report placed median GRR at 88%, with median NRR at 101% — the baseline most 2025 targets were built on. A year later, the median across 342 companies had fallen to 84%, with the 75th percentile sliding from 95% to 91% and the bottom quartile sitting at 76%.
| Metric | 2025 baseline | 2026 report (FY2025 data) |
|---|
| Median GRR | 88% | 84% |
| 75th percentile | 95% | 91% |
| Bottom quartile | Not separately reported | 76% |
Sources: Benchmarkit, 2025 SaaS Performance Metrics; Aleph/Benchmarkit, 2026 GRR report.
The detail that should worry a customer operations leader more than the median itself: the decline hit the top quartile, the median, and the bottom quartile together. When every percentile moves in the same direction at the same time, the report treats it as a market-level structural shift — not a cluster of individual companies executing poorly.
What a four-point drop actually costs
On a $10M ARR base, a four-point GRR decline equals roughly $400,000 of lost recurring revenue in a single year — and at the 84% median, the typical company is now losing 16% of its existing ARR annually before new bookings even enter the picture.
That math changes how a customer lifecycle should be managed day to day. See our breakdown of customer lifecycle management in B2B SaaS for how each stage feeds into the number you eventually report to the board.
Why a Healthy NRR Can Sit on Top of a Broken GRR
Net revenue retention is the number most boards ask for first. It's also the number most likely to hide the exact problem gross revenue retention exists to catch.
The masking mechanism, with a worked example
Take the same $10M ARR base from the reset above, where a four-point GRR decline costs roughly $400,000 in lost recurring revenue. Now add a typical growth-stage expansion motion: a handful of large accounts each add new seats and modules over the year, and that expansion revenue more than covers the gap on the income statement. The blended net revenue retention number comes out looking healthy. The gross revenue retention number underneath it doesn't move — the company still lost the same customers and the same dollars. It just found new dollars fast enough to hide the loss on one line of the board deck.
Read only the growth number and the story is "we're expanding." Read the retention number underneath it and the story is "we lost part of our base, and a small number of large accounts covered for everyone else." Both numbers are accurate. Only one tells you where the company is actually vulnerable.
What this looks like from inside customer operations, not finance
This masking effect concentrates in the accounts your team touches first: onboarding, early support interactions, and the first renewal cycle. Retention economics differ meaningfully by contract size — companies with ACVs between $25,000 and $50,000 show a median net revenue retention of 102%, with the bottom quartile at 97%, a tighter spread than lower-ACV segments typically show, which SaaS Capital ties to longer sales cycles and deeper implementation creating a stickier product. Smaller, faster-onboarded accounts churn more easily and expand less — which is exactly the segment a support and enablement leader is accountable for.
If your dashboard only surfaces the blended NRR number, the accounts quietly leaving never show up until finance asks why bookings had to work so much harder this quarter. Our guide to net revenue retention in SaaS covers the expansion side of this equation in more depth. The point here is narrower: GRR is the number that tells you whether the foundation under that expansion is solid or eroding.
The Four-Component Breakdown: Decomposing GRR by Owner
Treating gross revenue retention as one metric to "improve" is why generic retention advice rarely moves it. The number blends four operationally distinct failure modes, and each one belongs to a different function.
1. Logo churn — an onboarding and time-to-value problem
A customer who cancels outright almost always decided early that the product wasn't worth the effort to adopt. This is fixable by whoever owns onboarding and time-to-value — not by finance, and not by a renewal negotiation months later. If your churn rate is higher than it should be, the root cause usually traces back to the weeks right after signature, not the weeks right before renewal.
2. Contraction and downgrade — a pricing and value-realization problem
A customer who stays but shrinks their contract is telling you something different from a customer who leaves entirely: the relationship is fine, but the value-to-price ratio isn't. This sits closer to pricing, packaging, and account management than to support — and it's the component customer operations leaders most often mistake for a support failure when it's actually a commercial one.
3. Support-driven attrition — a resolution-quality and effort problem
Some churn is neither an onboarding failure nor a pricing mismatch. It's the cumulative effect of every unresolved or slow ticket eroding confidence in the product. This component belongs squarely to support operations, and it's measurable independently of the other three if your ticketing and health-score data are structured to isolate it. A customer health score built on the right architecture should flag this pattern months before the renewal date, not on the call itself.
4. Self-serve and knowledge gaps — a deflection and enablement problem
The fourth component is the quietest. Customers who never file a ticket, never escalate, and never complain simply find the product too hard to use without help that isn't there, and they leave without telling anyone why. This is the component most directly tied to whether your knowledge base and self-service surfaces are structured well enough to answer real questions — and it's the one most support leaders underweight, because it never shows up as a support metric at all.
Split these four apart and a single GRR number turns into four separate diagnoses, each with an owner who can actually act on it — rather than one number everyone in the org can point at without anyone owning the fix.
Team Structures That Actually Move the Number (and Ones That Don't)
Most leadership teams respond to a falling GRR number the same way: add customer success managers. The staffing data doesn't support that instinct. A 132-company survey of how customer success teams are staffed, paid, and structured found no meaningful correlation between CS staffing ratio and revenue retention or renewal rate. Most teams are still staffed by customer count rather than revenue, even as they scale. Adding headcount without changing who owns which piece of the four-component breakdown just spreads the same undifferentiated workload across more people.
What the staffing-ratio data really shows
The same survey found org charts, P&L placement, and comp plans for customer success vary widely from company to company, which makes ratio comparisons harder to use as a benchmark than most operators assume. A CSM ratio that looks thin on paper can still produce strong retention if logo churn, contraction, and support-driven attrition are split across distinct owners. A ratio that looks generous can still underperform if every CSM is expected to catch all four failure modes alone.
Real ratios operators are running in 2025-2026
When ratios are measured by revenue rather than logo count, the range is wide. Most CSMs carry between $2 million and $5 million in ARR, with a median closer to $1.4 million and a top quartile around $4.2 million. Enterprise accounts sit at the low end of that range with a handful of accounts per CSM. SMB and tech-touch tiers push into the hundreds of pooled accounts per person. A commonly cited scaling rule among CS leaders puts the starting ratio near two million dollars in ARR per CSM. Heavier early-stage investment runs to 10% of revenue into customer success, closer to 5% at scale.
None of those ratios tell you whether the team is structured to catch logo churn early, price contraction correctly, resolve support issues before they compound, or close self-serve gaps before a customer quietly leaves. Ratio math answers "how many people," not "who owns what." Getting the team structure right means assigning each of the four components to a specific role before deciding how many of that role you need.
Building a GRR Diagnostic for Your Team
Before gross revenue retention shows up on your dashboard as a single line, it needs to be built from data that already separates the four components. Bolt a GRR calculation onto a system that only tracks blended churn, and you'll get a number nobody can act on.
The questions to ask before you touch a dashboard
Four questions determine whether your data can support the breakdown:
- Can you isolate accounts that canceled outright from accounts that downgraded but stayed?
- Does your health-score model flag declining support sentiment separately from declining product usage?
- Do you know which lost accounts never filed a single support ticket before they churned?
- Can you trace a churned account back to its onboarding timeline and time-to-first-value?
If the answer to any of these is no, the fix isn't a new dashboard. It's restructuring the underlying records so each account carries the fields that make the split possible.
What to instrument in support, onboarding, and knowledge tooling
The instrumentation work sits closer to your knowledge and support stack than to finance. Onboarding milestones need timestamps that connect to time-to-value, not just completion checkboxes. Support tickets need resolution-quality tags, not just close status. Self-service search logs need to capture failed queries — the questions customers asked and never got answered — because that's the leading indicator for the fourth, quietest component. A health score built to separate these signals turns a lagging GRR number into a leading one your team can act on months before a renewal date. A documented renewal management process gives contraction and logo-churn risk two different intervention paths instead of one generic save play.
What to Do This Quarter
You don't need finance to redefine gross revenue retention before you start acting on it. The move that changes the most, and costs the least, is simple: report logo-churn GRR and contraction GRR as two separate numbers internally, even informally. Do it before the board slide forces the conversation. The two failures have different owners, different timelines, and different fixes. Collapsing them into one metric is why "improve GRR" initiatives so often stall after a quarter.
That split only works if the underlying knowledge, support, and onboarding data actually separates the signals. It's the same structural problem that keeps AI agents from resolving real questions and keeps health scores lagging instead of leading. MatrixFlows connects onboarding milestones, support resolutions, and self-service search gaps in one foundation. A churned account's full history — what it searched for, what it asked support, when it stalled in onboarding — surfaces as one record instead of three disconnected exports. That's what turns a diagnostic exercise into a system you can run every quarter instead of rebuilding it every time the board asks.
Start by pulling last quarter's churned and downgraded accounts and sorting them into the four buckets by hand if you have to. The pattern will tell you which owner needs to move first. Create a Free Workspace → and start building the diagnostic on data that's already structured to answer the question your board is about to ask.
.svg)
.svg)
.svg)
