A customer cancels a $12,000 subscription on Tuesday. By Friday, another customer expands by $18,000. Did the business get healthier?
That is the real question behind NRR versus churn. One metric records the leak. The other tells you whether the whole revenue base is still filling the pool faster than it drains. Confusing them creates the kind of dashboard that looks fine from across the room, like a CRT with the brightness turned all the way up, but hides the signal you actually need.
For SaaS operators, founders, and Customer Success teams, these are not competing metrics. They are two views of the same machine. The trick is knowing what each view can and cannot tell you.
NRR versus churn: two different camera angles
Churn measures loss. Depending on the version, it can mean customers leaving, recurring revenue leaving, or a combination of cancellations and downgrades.
Net Revenue Retention, or NRR, starts with a cohort's recurring revenue and asks what remains after churn, contraction, and expansion. It ignores new-logo revenue on purpose. That makes it unusually useful: it shows whether the customers you already won are becoming a stronger asset or a more expensive treadmill.
The simplified formulas are straightforward:
Revenue churn rate = churned recurring revenue รท starting recurring revenue
NRR = (starting recurring revenue - churn - contraction + expansion) รท starting recurring revenue
If you begin the quarter with $1,000,000 in ARR, lose $80,000 to cancellations, lose another $40,000 to downgrades, and expand existing accounts by $170,000, your NRR is 105%.
You still had churn. You still need to understand why $120,000 moved backward. But the existing customer base grew by $50,000 overall.
That distinction matters. Saying, โOur churn is 8%,โ describes a problem. Saying, โOur NRR is 105%,โ describes the combined result of retention, adoption, pricing, packaging, account growth, and Customer Success execution.
Why a good NRR can hide a bad week
High NRR is powerful, but it can be a flattering wide-angle lens.
Imagine a company with 110% NRR. Great headline. Now look closer: its ten largest accounts expanded aggressively, while dozens of smaller customers quietly churned after failing to reach first value. The company may be building a strong enterprise motion while also carrying a product onboarding problem that will become painful when the enterprise pipeline slows down.
NRR does not tell you whether growth is broadly distributed. It does not distinguish a customer who doubled usage because the product is indispensable from one who accepted a price increase because procurement was busy. It also does not tell you how much expansion came from a handful of accounts that are now overexposed in the revenue base.
This is why I would not put one giant NRR number on a screen and call the revenue system understood. Segment it.
Look at NRR by acquisition channel, plan, industry, customer size, tenure, implementation path, and product usage pattern. A 108% blended result can contain a 125% enterprise cohort and a 78% self-serve cohort. Those are not variations of one story. They are two different products, two different customer journeys, or both.
The useful question is not, โIs NRR above 100%?โ It is, โWhich customers expand after realizing value, and which customers disappear before they have a chance?โ
Why churn alone is an incomplete alarm bell
Churn is easier to calculate and easier to react to. A cancellation arrives. An alert fires. Someone asks for a save playbook.
That response is necessary, but churn is a lagging event. By the time an account cancels, the useful signals may be months old: an implementation that stalled, a champion who left, a workflow nobody adopted, an ROI story that never became real, or a new stakeholder who saw the product as a line item instead of infrastructure.
Churn also lacks context without a denominator and a customer mix. Losing five customers can be catastrophic for a 40-account company and meaningless for a large self-serve business. Losing $50,000 in monthly recurring revenue means something different when it comes from fifty $1,000 accounts versus one strategic $50,000 account.
Track both logo churn and revenue churn. Logo churn tells you how many relationships ended. Revenue churn tells you how much recurring revenue left. When those move in opposite directions, pay attention.
Low logo churn with high revenue churn often means larger accounts are leaving or contracting. High logo churn with low revenue churn can mean small customers are failing while the core business remains stable. Neither pattern is automatically acceptable. They simply demand different work.
Build a retention model that does not lie to you
The cleanest retention model starts with fixed cohorts. Freeze a group of customers at the beginning of a month, quarter, or year. Track their starting recurring revenue, then separate the changes that happen afterward.
Use four buckets: retained revenue, contraction, churn, and expansion. Keep new-logo revenue outside the cohort calculation. Otherwise, strong sales can paint over a weak customer base and make the dashboard feel better than the business is.
Then add the operating details that make the number actionable:
- renewal date and contract term
- account owner and executive sponsor
- product adoption or usage milestones
- realized value, not just activity
- expansion trigger and expansion source
- churn reason, with evidence rather than a dropdown guess
A better churn review reads like a short case file. What job was the customer hiring the product to do? Did they get there? Who believed in it? What changed? Could anyone have seen the risk earlier? This takes longer than choosing a field in a CRM, but it produces patterns you can actually design against.
The trade-off inside expansion
Expansion is usually treated as pure good news. Mostly, it is. But not all expansion has the same quality.
Healthy expansion follows increased value. More teams use the product, more workflows depend on it, or the customer sees a measurable return and chooses to broaden the relationship. That type of expansion makes future renewal easier because the account has more roots in the organization.
Fragile expansion can come from a forced plan migration, a poorly timed price increase, or selling modules before the core product is adopted. It raises NRR now and can create a renewal cliff later. The spreadsheet will clap. The customer may not.
This is where Customer Success and revenue operations need the same definition of health. If Sales marks an upsell as complete while the customer has not reached value in the original product, the business has booked a larger version of the same risk.
Use NRR to choose where to build
NRR is most useful when it changes a decision.
If newer cohorts retain poorly, fix the first 30 days before investing in a fancier renewal forecast. If customers with multiple active users expand consistently, build onboarding around team adoption rather than administrator setup. If churn clusters after a champion leaves, create stakeholder maps and executive value reviews before the renewal becomes a rescue mission.
If a segment has weak NRR but healthy gross retention, expansion may be a packaging, product surface area, or account-planning opportunity. If gross retention is weak, do not try to sell your way around it. Find the moment value broke.
At Slam Dunks Studios, this is the kind of system worth building as a working tool rather than a quarterly slide. A cohort view, a simple account-risk model, and a place to record actual customer evidence can turn retention from a postmortem ritual into a design input.
The goal is not to worship NRR or panic at every churn event. It is to make the revenue base legible enough that the next useful move is obvious. Start with one cohort, one period, and one honest question: did customers receive enough value to stay, and enough more value to grow?
