Churn and Retention

Net Revenue Retention (NRR)

If a SaaS business had one metric that should never be ignored, it would be NRR. Not MRR, not ARR, and not Logo Churn. NRR shows what is happening to the existing revenue base all at once: loss, contraction, expansion, and the quality of the product's long-term value.

Language: Read this article in Russian

NRR

NRR (Net Revenue Retention): why it is the core SaaS health metric

Definition and Core Formula

MRR tells you where the business is now. ARR shows scale. Churn shows losses. NRR combines all of that into one number. A company with 120% NRR can grow without adding a single new customer. A company with 80% NRR is shrinking no matter how much it spends on acquisition.

Net Revenue Retention, also called Net Dollar Retention, measures how recurring revenue from the existing customer base changed over a period after churn, contraction, and expansion.

NRR = (Starting MRR − Churned MRR − Contraction MRR + Expansion MRR) / Starting MRR × 100

Where:

  • Starting MRR: recurring revenue from customers active at the start of the period.
  • Churned MRR: full recurring revenue lost from customers who left.
  • Contraction MRR: revenue lost from customers who stayed but pay less.
  • Expansion MRR: revenue gained from upsells, cross-sells, add-ons, or usage growth.

The critical rule is simple: only customers that already existed at the start of the period belong in NRR. New customers are excluded from both the denominator and the numerator.

NRR = (200,000 − 6,000 − 4,000 + 20,000) / 200,000 × 100 = 105%

That means the same customer base produces 5% more recurring revenue than one period earlier, without any new logos at all.

The threshold around 100% is qualitative, not just mathematical:

  • NRR > 100%: net negative churn, the installed base grows by itself.
  • NRR = 100%: neutral retention, expansion exactly offsets losses.
  • NRR < 100%: net positive churn, the base is shrinking.

Decomposing NRR into Its Three Forces

NRR is not one lever. It is the outcome of three forces that need to be managed separately.

GRR

GRR = (Starting MRR − Churned MRR − Contraction MRR) / Starting MRR × 100

Gross Revenue Retention is NRR without expansion. It shows how well the company holds revenue before upsell helps. GRR can never exceed 100%.

Expansion Rate

Expansion Rate = Expansion MRR / Starting MRR × 100

Expansion is the internal growth engine of the installed base.

Expansion MRR = Upsell MRR + Cross-sell MRR + Add-on MRR + Usage Expansion MRR

Revenue Churn Rate

Revenue Churn Rate = (Churned MRR + Contraction MRR) / Starting MRR × 100
GRR = 100% − Revenue Churn Rate

Each part of revenue churn deserves its own diagnosis: voluntary churn, involuntary churn, and contraction.

NRR = GRR + Expansion Rate

This decomposition is what makes NRR useful. Two companies can both report 105% NRR while having very different quality underneath.

Metric Company A Company B
GRR98%85%
Expansion Rate7%20%
NRR105%105%
DurabilityHighFragile

Company A is stable and healthy. Company B is compensating large losses through aggressive upsell. Investors do not value those two situations the same way.

Derived Metrics and Practical Variants

Net Revenue Churn

Net Revenue Churn = (Churned MRR + Contraction MRR − Expansion MRR) / Starting MRR × 100

This is the inverse side of NRR. When it turns negative, you have net negative churn.

Expansion Efficiency

Expansion Efficiency = Expansion MRR / (Churned MRR + Contraction MRR)
  • < 0.5: expansion is not meaningfully offsetting losses.
  • 0.5-1.0: partial compensation.
  • = 1.0: neutral retention, NRR around 100%.
  • > 1.0: net negative churn, NRR above 100%.
  • > 1.5: strong net negative churn.

Monthly vs Annual NRR

Monthly NRR is the standard operating metric, but it can be distorted by one large upsell or one large churn event. Annual NRR is more stable and is used more often in investor reporting.

Annual NRR = MRR(end of year) from starting-year customers / MRR(start of year) from those same customers × 100

Annual NRR is not Monthly NRR multiplied by 12. The rough theoretical relationship is closer to `Monthly NRR^12`, but in practice annual NRR is usually lower because churn continues through the year.

Trailing 12-Month NRR

TTM NRR = Current MRR from customers active 12 months ago / MRR from those same customers 12 months ago × 100

TTM NRR is often the most representative version for investors because it smooths seasonality and one-off enterprise events.

Cohort NRR

Cohort NRR(month t) = Cohort MRR(month t) / Cohort MRR(month 0) × 100

Aggregated NRR hides an important pattern: cohort NRR usually drops in the first few months because of early churn, then improves later as expansion begins to dominate.

NRR and LTV

LTV (NRR-adjusted) = ARPA × Gross Margin × NRR / Churn Rate

Flat-ARPA LTV underestimates the value of a base that expands over time.

NRR and CAC Payback

Effective Payback = CAC / (ARPA × Gross Margin × (NRR / 100))

Higher NRR reduces the effective payback period because retained customers keep becoming more valuable.

NRR by Segment

Total NRR can easily hide a weak SMB base under strong enterprise expansion. Segment-level NRR is not optional if you want a real operating view.

Common NRR Calculation Mistakes

  • Including new customers in Starting MRR or Expansion. NRR measures only the installed base.
  • Treating reactivation as expansion. Returning revenue was not present in Starting MRR.
  • Judging the business from one month. One large upsell can distort monthly NRR badly.
  • Mixing Monthly NRR with Annual NRR. These are different metrics and must be labeled explicitly.
  • Looking at NRR without splitting GRR and Expansion Rate. The same headline NRR can describe a robust business or a fragile one.
  • Using average MRR instead of Starting MRR in the denominator. That breaks comparability with standard SaaS benchmarks.
  • Calculating NRR by customer count instead of recurring revenue. Standard NRR is revenue-based.

Worked NRR Example with Diagnosis

Quarter data: Starting MRR from existing customers = $500,000

Component Amount % of Starting MRR
Churned MRR (voluntary)−$15,0003.0%
Churned MRR (involuntary)−$5,0001.0%
Contraction MRR−$10,0002.0%
Upsell MRR+$20,0004.0%
Add-on MRR+$12,0002.4%
Usage Expansion MRR+$8,0001.6%

Churned MRR total: $20,000

Contraction MRR: $10,000

Expansion MRR total: $40,000

GRR: (500,000 − 20,000 − 10,000) / 500,000 × 100 = 94%

Expansion Rate: 40,000 / 500,000 × 100 = 8%

NRR: (500,000 − 20,000 − 10,000 + 40,000) / 500,000 × 100 = 102%

Expansion Efficiency: 40,000 / 30,000 = 1.33

Net Revenue Churn: −2%

Diagnosis: NRR at 102% is positive, but only barely. GRR at 94% is below where many teams want it. Expansion is compensating for losses, but with a thin margin. If expansion slows even modestly, NRR falls below 100%.

The operating opportunity is clear: reduce involuntary churn with better dunning, lower contraction through stronger feature adoption, and keep expansion steady. Small operational improvements can move NRR materially.

How to Improve NRR

Reduce Churned MRR. This improves GRR, which is the most durable way to improve NRR.

  • Shorten time to value through better onboarding.
  • Monitor early-warning signals such as falling login frequency, declining feature usage, unresolved support issues, failed payments, and loss of the internal champion.
  • Improve dunning to reduce involuntary churn without changing the product at all.

Reduce Contraction MRR. Customers who stay but pay less are often ignored, but contraction can quietly destroy GRR.

  • Analyze why customers downgrade.
  • Improve adoption of premium-tier features.
  • Make pricing tiers create a real reason to stay at higher levels.

Increase Expansion MRR. Expansion is the most scalable NRR lever because it does not require CAC.

  • Use usage-based pricing where it fits.
  • Trigger seat-expansion plays when accounts approach purchased limits.
  • Use product-led expansion with clear upgrade paths and feature gates.
  • Give Customer Success explicit expansion ownership, not retention-only ownership.

Dnoise calculates NRR automatically from Stripe subscription history with full decomposition into Churned MRR, Contraction MRR, Expansion MRR, and GRR. TTM NRR updates daily, cohort NRR is available by monthly cohort, and segment-level NRR can be broken down by plan, acquisition source, geography, or other customer attributes.

See NRR in the demo

CTA

Dnoise calculates NRR from source data and shows which part of retention is truly healthy and which part is being covered by expansion.

MAP

Related Reading

Compare with:

CTA

Dnoise calculates NRR from source data and shows which part of retention is truly healthy and which part is being covered by expansion.