The number that flatters you in the board deck
Picture the slide. Net Revenue Retention: 105%. It's green. Someone on the board nods, because 105% is "above water," and the conversation moves on to pipeline. Nobody asks the follow-up question that actually matters: how many customers did you have to lose to get there?
That's the whole problem with NRR as a headline number. It blends two opposite forces — accounts expanding and accounts leaving — into a single figure that can look healthy while the underlying logo base is on fire. A SaaS company can post 105% net retention while shedding 15% of its customers a year, because a handful of big accounts upsized enough to paper over the exodus. The board sees one number. The cap table feels the other one, eventually.
This matters more at Series B and C than at any other stage, and it's worth being precise about why. Pre-product-market-fit, new logos can outrun almost any churn rate, so retention is noise. Post-Series-B, your new-logo growth rate is mathematically slowing — you're a bigger base, the easy market is taken — and net retention quietly becomes the dominant term in your growth equation. The retention you ignored at $5M ARR is the ceiling you hit at $30M. The investor who quotes you a 4x multiple instead of an 8x is usually staring at exactly this, and the link runs in one direction: as ChartMogul's growth and retention data shows, the companies that compound do it on the strength of their installed base, not their hunting.
So before you celebrate a green number, you need to know whether it's green because customers love you, or green because three accounts got big enough to hide the bodies.
The formula, and the one input most teams quietly fudge
Net Revenue Retention measures one thing: what happened to the revenue from the customers you already had, ignoring everyone you've signed since. The formula:
NRR = (Starting ARR + Expansion − Contraction − Churn) / Starting ARR
The single most common error is letting new-logo revenue sneak into the numerator. The moment a new customer's ARR touches this calculation, the number stops measuring retention and starts measuring sales — and it'll read artificially high right up until the day you need it to be honest. Lock the cohort first: take the exact set of customers you had on day one of the period, then track only what that frozen set does. New logos go in a different report.
Run GRR alongside it, every time
Here's the pairing that exposes the masking. Gross Revenue Retention strips out expansion entirely — it only counts what you kept and what shrank or left. The gap between the two numbers is the diagnosis.
Say you start the year with 100 customers at $10k each — $1M ARR. Over twelve months, 15 of them leave. Your gross retention just fell to 85%. But your customer success team ran a hard upsell on the 85 survivors and pulled in $200k of expansion. Ending ARR: $1.05M. NRR: 105%.
That 105% is real. It's also a warning. You're net-positive only because you successfully extracted more from a shrinking pool of accounts. That works until it doesn't — every year you have fewer customers to upsell, the expansion engine gets less to work with, and the churn keeps grinding. A wide NRR-to-GRR spread isn't strength; it's a dependency on a few accounts agreeing to pay you more, forever. The practical rule: if GRR sits below 90%, treat the NRR above it as a number you have to defend, not one you get to celebrate. Pavilion's 2025 B2B SaaS benchmarks consistently show the durable companies keeping that gross floor intact, then adding expansion on top — not using expansion to plug a hole.
Benchmarks that actually mean something — and the cohort cut that finds the leak
"Industry average churn" is a useless number because it averages across businesses that aren't comparable. A 5% annual churn rate is excellent if you sell $3k tools to small businesses and a five-alarm fire if you sell $200k enterprise platforms. Segment by what you charge before you compare yourself to anyone. Drawing on SaaS Capital's 2025 retention benchmarks for private B2B companies:
- Enterprise (ACV above $100k): high switching costs should buy you loyalty. The bar is roughly 6–8% annual gross churn. Push past 10% and you don't have a churn problem, you have an implementation or product problem dressed up as one.
- Mid-market (ACV $10k–$100k): expect 11–15% annually. This is the budget-review kill zone, where a tool that's merely "nice to have" gets cut the first quarter someone tightens spend.
- SMB (ACV under $5k): 25–30% annual churn is normal and survivable — but only if your top-of-funnel volume is large enough to keep refilling the tank.
On the net side, the 2025 median for B2B SaaS lands around 106%, meaning the typical company grows modestly off its base. The top decile — the ones commanding the premium exit — live at 120%+. Below 100% and you're fighting a mathematical headwind that strengthens as you scale: even with zero new sales effort, the base shrinks on its own. If you're staring at a number under 100, our breakdown of what a sub-100% NRR actually signals walks through why those firms so rarely clear the Rule of 40.
The Monday move: cut NRR by acquisition vintage
Stop staring at one global number — it's an average of averages and it hides everything. Instead, split your base by the quarter each cohort signed: "customers acquired Q1 2023," "Q3 2024," and so on. Then compute NRR per vintage.
This is where the real story usually surfaces. Older cohorts often stabilize north of 110% — the bad fits already churned, the survivors expand. But a recent vintage, the one you signed during a quarter when sales was chasing a number and onboarding was overwhelmed, might be sitting at 80%. That single cut tells you the exact period your acquisition quality or onboarding broke, which team owned it, and where to point the fix — none of which the blended 105% would ever have admitted. Run it this week, and the next board deck has a number you can actually stand behind.
