type: pattern tags: [nrr, saas, cohort-analysis, enterprise, expansion-motion, net-revenue-retention, smb] confidence: medium created: 2026-04-03 source: GTLB earnings-review Q4_FY26 persona: atlas provenance: legacy source_analysis_path: null source_paragraph_quote: null source_transcript_span: null source_loss_log_path: null

NRR Tier Bifurcation as Cohort Quality Discriminator

When headline NRR declines but top-tier customer count ($1M+ ARR cohort) accelerates simultaneously — and gross retention hits multi-year highs — the signal is cohort bifurcation, not broad-based expansion weakness. The enterprise segment is healthy; a price-sensitive SMB/mid-market cohort is contracting or flat. Accepting the blended NRR as the full story misdiagnoses the business. The correct read is: the enterprise growth engine is intact, and the drag is concentrated in a cohort the company can deliberately de-emphasize.

Evidence

• GTLB Q4 FY26: DBNR declined to 118% (from 123% four quarters prior, -1pp/quarter pace) while $1M+ customer count grew +26% YoY and $100K+ grew +18% YoY. Management confirmed gross retention hit 4-year best (>90%), proving churn is not the driver. CFO attributed NRR decline to a ~20% of ARR price-sensitive cohort not expanding. Enterprise adds accelerating while blended NRR declined created the appearance of a deteriorating expansion motion.

Implication

When evaluating SaaS NRR trends: always segment by tier before drawing conclusions.

1. Rising enterprise cohort count + declining NRR → cohort bifurcation; check gross retention to confirm.
2. Declining gross retention + declining NRR → actual churn issue; structurally different risk.
3. If the enterprise segment is the future revenue backbone (large deal sizes, high ACV), the SMB NRR drag may be acceptable and deliberate (letting price-sensitive customers churn while upselling enterprise).
4. Model NRR recovery relative to which cohort the new product (AI, usage-based, etc.) is targeted at — if enterprise, blended NRR improvement will lag the enterprise cohort metrics by 2-4 quarters.