By Ron Davenport. Published 2026-05-23.
Cluster: Churn Analysis
120% NRR can mean retention is great. Or it can mean upgrades are masking defection.
The four retention metrics every operator should be tracking, where each one lies, and the dashboard worth keeping so the boardroom narrative doesnt drift from what the cohort actually did.
TL;DR
Net retention adds expansion to gross retention. Revenue churn is dollar-weighted. Logo churn is customer-count weighted. Each metric tells the truth under specific conditions and lies under others. NRR above 100% can hide a leaky base if expansion is paying for it. GRR is the cleanest read on whether your existing customers stay paying you the same amount. Tracking only one of them is how boards get blindsided.
Heres the operating point. 120% NRR can mean youre running an absolute machine. It can also mean a small share of your base is upgrading hard while a much larger share is quietly leaving, and the expansion dollars are papering over the churn until the board asks the wrong question and you have nothing to point at.
The fix is reading all four metrics together because each one shows you something the others hide. This is the metric breakdown I run on every retention diagnostic, including which one the boardroom actually wants and which one you should be running operationally. Sits under the churn analysis cornerstone alongside the curve archetypes and intervention matrix.
01 Definitions
The four metrics, cleanly defined.
Lot of metric confusion lives in this corner of SaaS analytics, partly because different boards use different definitions and partly because tools display them inconsistently. The operating definitions below are the ones I default to on every engagement.
| Metric | Formula | Ceiling | Best for |
|---|---|---|---|
| Gross retention (GRR) | (Start MRR − churn MRR − downgrade MRR) / Start MRR | 100% | Base durability read |
| Net retention (NRR) | (Start MRR − churn − downgrade + expansion) / Start MRR | Unbounded | Full revenue motion read |
| Revenue churn | (Churn MRR + downgrade MRR) / Start MRR | 0% (lower = better) | Dollar-weighted loss read |
| Logo churn | Customers lost / Customers at start | 0% (lower = better) | Breadth / concentration read |
Gross retention (GRR).
The percentage of starting recurring revenue you kept from your existing base, before expansion. Calculated as starting MRR minus churned MRR minus downgrade MRR, divided by starting MRR. The ceiling is 100% by definition because you cant retain more than what you started with at the gross level.
GRR is the cleanest read on whether your existing customers stay paying you the same amount theyre paying you today. It strips out expansion (which is its own growth lever) and shows you whether the base is leaky. A GRR of 92% means 8% of last years recurring revenue from existing customers walked away or downgraded this year.
Net retention (NRR).
GRR plus expansion. Calculated as starting MRR minus churned MRR minus downgrade MRR plus expansion MRR, divided by starting MRR. Can exceed 100% when expansion outpaces churn-plus-downgrade.
NRR is the metric investors and boards love because it captures the full revenue motion from the existing customer base. NRR above 110% generally means the existing base is growing without new logos, which is the dream growth model. Below 100% means the existing base is shrinking and new logo acquisition has to make up the difference just to stay flat.
Revenue churn.
The dollar-weighted churn rate. Calculated as MRR lost to cancellations and downgrades divided by starting MRR over the period. Inverse of GRR essentially (8% revenue churn equals 92% GRR), but framed as a loss metric rather than a retention metric.
Revenue churn is the operating metric I track most often because its the dollar reality. Two customers churning where one was a $50 plan and one was a $5,000 plan looks like 2% logo churn but might be 30% revenue churn at a small SaaS. The dollar weighting tells the truer story.
Logo churn.
The customer-count churn rate. Calculated as customers lost divided by starting customers. Doesnt account for revenue weighting at all, every customer counts the same.
Logo churn is useful for understanding the breadth of the problem. High logo churn with low revenue churn means youre losing a lot of small customers while keeping the big ones (which might be fine, might be a future expansion problem). Low logo churn with high revenue churn means a few big customers are leaving (which is usually a five-alarm fire).
02 Where each metric lies
What each one tells you and where it stops telling you the truth.
Every metric has a condition under which it lies. Knowing the conditions is how you avoid the boardroom narrative drifting from operational reality. Below are the most common ways each of these metrics has misled a team I worked with.
How NRR lies.
The classic NRR lie is hiding a leaky base behind expansion revenue. NRR of 115% sounds great until you look at GRR underneath and see its 85%, which means 15% of your base is churning every year and expansion is barely outpacing the bleed. That growth model breaks the moment expansion plateaus or competition tightens. Always look at NRR alongside GRR, never alone.
Second NRR trap: concentration. If 70% of your expansion revenue comes from 5% of your customers, NRR looks great until one of those whales churns. The metric tells you nothing about the breadth of expansion or the concentration risk. Cohort the NRR by customer segment to see whether the expansion is broad or concentrated.
How GRR lies.
GRR can miss healthy expansion-led growth where the business model is built on land-and-expand. If your GRR is 88% but expansion is 40% and NRR is 128%, the business is great even though the GRR number alone looks mediocre. GRR in isolation undersells a usage-pricing or seat-expansion model.
GRR also misses timing distortion when a single big customer renews at a much higher price tier. The downgrade-to-churn ratio inside GRR can be misleading because some "downgrades" are actually pricing-tier corrections rather than dissatisfaction signals. Cut downgrade revenue by reason to separate dissatisfaction from price-tier reshuffling.
How revenue churn lies.
Revenue churn can mask logo concentration risk. If your revenue churn is 5% but its made up entirely of one customer leaving (because that customer was 5% of revenue), you have a concentration problem dressed as healthy retention. Always cross-check revenue churn against logo churn and look at the dollar distribution of churned customers.
Revenue churn also misses seasonality and contract timing distortion. A SaaS that does annual contracts with most customers will see revenue churn cluster around renewal anniversaries. A monthly snapshot misses the cohort reality. Always compute revenue churn on a trailing 12-month basis to smooth out the seasonality, not on a single-month snapshot.
How logo churn lies.
Logo churn can mask high-value customer loss when the SaaS has a wide range of customer sizes. Losing 5 customers from a base of 100 looks like 5% churn, fine. If those 5 were your top revenue accounts, your revenue churn is 40%. Logo churn alone is a vanity metric for any SaaS with meaningful customer-size variance, which is most of them.
Logo churn also misses voluntary versus involuntary distinction. A subscription product losing 20% of its monthly customers might look like a churn crisis when 12 of those points are involuntary (failed payment) and only 8 are voluntary. The fix for each is completely different. The dunning playbook handles the involuntary side specifically.
03 The dashboard
The operator dashboard worth keeping.
The actual metric set I keep on retention dashboards for the SaaS engagements I run. Six numbers, not one. Each one tells you something the other five hide. The point isnt comprehensiveness for its own sake, its making sure no single number is the only one anyone looks at.
| Metric | Cadence | What it reads |
|---|---|---|
| GRR (trailing 12mo) | Quarterly | Base durability without expansion mask |
| NRR (trailing 12mo) | Quarterly | Full revenue motion incl. expansion |
| Revenue churn (monthly) | Monthly | Dollar loss rate, includes downgrades |
| Logo churn (monthly) | Monthly | Breadth and concentration risk |
| Expansion MRR | Monthly | Source of NRR lift; cohort by segment |
| Contraction MRR | Monthly | Downgrade signal; separate dissatisfaction vs reshuffling |
The six metrics together give you both the dollar reality and the breadth reality. Cohort them by customer segment (size band, plan tier, vintage, acquisition channel) to spot the patterns underneath the headline numbers. Single-number reporting on retention is how boards and executives get blindsided.
04 The boardroom version
What the board actually wants to see.
Board-level retention reporting compresses the dashboard above into two or three numbers. The compression matters because boards have limited time and want a signal-rich summary. The trap is that compression hides nuance, so the operator version above should be the source of truth and the board version should be a derived view, not the other way around.
For growth-stage SaaS (Series A through Series C). NRR is the headline metric investors track. They want to see whether the expansion motion is real, because thats the leverage that justifies the growth-stage valuation. Pair NRR with GRR to show you arent hiding a leaky base behind expansion.
For mature SaaS (Series D+ or profitable independent). GRR becomes the headline because expansion is harder to keep growing at scale and the question shifts to whether the base is durable. Boards still want NRR for the growth story but theyre paying closer attention to GRR because base durability drives valuation at maturity.
For seed and pre-revenue SaaS. Cohort retention curves matter more than headline metrics because the data isnt yet at the volume where annual NRR or GRR is statistically meaningful. The cohort retention curves cluster covers what to look at instead.
05 Next step
Where this lands operationally.
The whole point of separating these metrics is making sure your retention story matches your retention reality. A SaaS with 120% NRR and 82% GRR has a completely different operating problem than a SaaS with 102% NRR and 96% GRR, and the interventions are completely different. The first is fighting a leaky base masked by expansion. The second is fighting incremental expansion ceiling.
The churn analysis cornerstone covers the diagnostic that determines which intervention class actually moves the metric, and the cohort retention curves cluster covers how to read the shape that underlies whichever number youre looking at.
If your NRR and GRR are telling different stories and you want a sharper read on which one to trust, send me a DM with the current numbers and Ill tell you where the gap is. Or book a full discovery call if you want a complete retention diagnostic.
Think your lifecycle is leaking?
Book a 30-minute call. One-page scope inside a week if there’s a fit. Clear no if there isn’t.