Net Revenue Retention (NRR) measures how much revenue a company retains and expands from its existing customer base over a given period, excluding new customer acquisition. An NRR of 130% means existing customers are spending 30% more than they did a year ago, even after accounting for churn and downgrades. The formula is: (beginning revenue + expansion − contraction − churn) / beginning revenue × 100.
NRR above 100% indicates a company can grow revenue even without acquiring any new customers — existing customers expand their usage faster than others leave. The best SaaS companies historically maintained NRR of 120–150%: Snowflake, Datadog, and CrowdStrike regularly reported figures in this range. NRR below 100% signals structural problems — customers are shrinking or leaving faster than remaining ones expand. In the agentic AI context, NRR is a leading indicator of disruption: if AI agents reduce the number of seats a customer needs, per-seat SaaS pricing models see NRR compress even when the customer remains. This is the mechanism behind the SaaSpocalypse thesis.
Net Revenue Retention is the single best metric for SaaS business quality. NRR above 130% means the company grows revenue from existing customers by 30% annually before acquiring any new customers. This compounds powerfully — a cohort of customers paying $1M in year one generates $1.3M in year two, $1.7M in year three. Snowflake, Datadog, and CrowdStrike have built extraordinary businesses on this dynamic.
However, NRR is also the canary in the coal mine for the SaaSpocalypse. When AI agents begin automating workflows that SaaS tools currently serve, seat expansion slows, usage-based pricing faces compression, and NRR declines. A drop from 130% to 110% NRR can cut a company's implied growth rate in half — and its valuation multiple with it. Closelook tracks NRR trends across the software sector as an early warning system.
NRR is a key input for the EBITDA Multiple framework, signals moat strength (see Economic Moat), and is central to the SaaSpocalypse and Software-Credit Nexus theses.
