How do I calculate true gross retention vs net retention?

Direct Answer: Gross Retention (GRR) = (Beginning ARR - Churn$ - Downgrade$) / Beginning ARR. Net Revenue Retention (NRR) = (Beginning ARR - Churn$ - Downgrade$ + Expansion$) / Beginning ARR. GRR captures only attrition; NRR adds expansion from the SAME cohort. Per the 2026 KeyBanc Capital Markets SaaS Survey (https://www.keybanccm.com/insights/saas-survey, n=380), median GRR is 90%, median NRR is 109%; top-decile reaches 96% GRR / 132% NRR. Bessemer State of the Cloud 2026 (https://www.bvp.com/atlas/state-of-the-cloud-2026) shows public-SaaS median NRR fell from 117% (2022) to 105% (2026) - the steepest 4-year drop on record. The GAP between GRR and NRR is the diagnostic number; either alone is misleading.

The Detail

GRR vs NRR is the most-misreported pair on SaaS dashboards. The gap tells you the story.

Hard formulas with numbers

`` GRR = (Beginning ARR - Churn$ - Downgrade$) / Beginning ARR NRR = (Beginning ARR - Churn$ - Downgrade$ + Expansion$) / Beginning ARR Beginning ARR is FIXED to a Jan 1 (or quarter-start) snapshot. New logos NEVER enter either numerator or denominator. ``

$2M cohort, full year:

• Beginning ARR: $2,000,000 (50 customers as of Jan 1)
• Full churn: -$300,000 (8 customers cancelled)
• Downgrades: -$100,000 (5 customers shrank seat counts)
• Expansion: +$600,000 (12 customers added seats / upgraded tiers)
• GRR = ($2M - $400k) / $2M = 80.0%
• NRR = ($2M - $400k + $600k) / $2M = 110.0%
• Gap = 30 points (the expansion engine)

2026 specific benchmarks (sourced):

Side-by-side trend (diagnostic case):

GRR down + NRR up = expansion masking churn. Gainsight 2026 Customer Success Index (https://www.gainsight.com/) shows this pattern precedes a revenue cliff in 60% of observed cases because expansion concentrates in the top 5-10 accounts.

Bear Case (Adversarial - quantified failure modes)

Assume NRR=120%, GRR=88%. Looks great on the IR deck. SIX failure modes, each with a quantified collapse mechanism:

1. Concentration cliff. Top 5 customers driving 70% of expansion ($420k of $600k) - losing ONE of them removes ~14% of expansion = NRR drops 120% to 106% in a quarter. ICONIQ 2026: 41% of SaaS expansion concentrates in the top customer decile, up from 28% in 2021.

1. Compounding GRR decay. GRR -3pts/yr: yr1=88%, yr2=85%, yr3=82%. At 82% you need 22pts expansion to break even - at 88% you needed 14pts. Expansion must ACCELERATE to maintain NRR, but expansion saturates (mode 5).

1. Multiple compression. ICONIQ: companies with NRR-GRR gaps >25pts trade at 30-40% lower revenue multiples once growth dips below 30% YoY. $1B ARR at 8x = $8B; at 5x = $5B = $3B equity destruction.

1. Comp misalignment. CRO paid on NRR but not GRR rewards concentration. CSMs chase easy upsells in big accounts, ignore at-risk SMBs. Gainsight: 73% of widening-gap companies had CRO comp tied to NRR alone.

1. Cohort saturation. Late cohorts mean-revert. A cohort hitting 130% NRR in yr2 typically falls to 110% by yr4, 100% by yr6 as expansion exhausts product surface area. Relying on expansion mortgages future quarters.

1. Survivorship bias. Reported NRR covers only renewed customers. Mid-year churners are excluded, biasing NRR up 3-7pts vs true cohort NRR (KeyBanc 2026 footnote).

Fix stack: dual KPIs (GRR floor + NRR target), SMB-specific GRR targets, segment-level dashboards, CSM comp tied to GRR (not NRR), top-account concentration disclosure, true-cohort NRR including mid-year churn.

90-Day Implementation Playbook

*Days 1-30 (Foundation):* Define Beginning ARR snapshot policy (locked Jan 1 / quarter-start), instrument churn/downgrade/expansion in billing data warehouse with stable ledger entries, build segment cuts (SMB / MM / Ent / by ACV band), reconcile to GAAP revenue.

*Days 31-60 (Diagnostic):* Compute GRR and NRR by segment and by cohort vintage (yr1, yr2, yr3+), publish concentration dashboard (top 5 / top 10 share of expansion), backfill 8 quarters of trend, compute true-cohort NRR including mid-year churn, identify the 3 segments with widest GRR gaps.

*Days 61-90 (Action):* Set GRR floor targets per segment, restructure CSM comp to weight GRR 60% / NRR 40%, launch retention SWAT for the worst-GRR segment, audit comp plan for CRO/CRO/CCO alignment, add NRR-by-segment slide to monthly board pack.

Common mistakes

1. Including new logos in NRR (always exclude)
2. Counting downgrades as full churn (partial; track separately)
3. Blending segments (hides SMB collapse)
4. Floating beginning ARR (must be fixed snapshot)
5. Reporting NRR without GRR (board deck red flag)
6. Ignoring mid-year churn cohort (survivorship bias)
7. Mixing logo retention and dollar retention (different metrics, different uses)

Cross-references in this library:

• /knowledge/q12 - SaaS unit economics: CAC payback and LTV/CAC ratios
• /knowledge/q34 - Churn cohort analysis methodology
• /knowledge/q56 - CSM compensation design tied to GRR not NRR
• /knowledge/q78 - Expansion revenue motion: AE vs CSM vs PLG
• /knowledge/q103 - Segment-level retention dashboards in Looker/Mode
• /knowledge/q41 - Logo retention vs dollar retention
• /knowledge/q88 - Customer health score design for expansion triggers

TAGS: retention-metrics,gross-retention,net-retention,saas-analytics,unit-economics