Net Revenue Retention (NRR) measures how much revenue from an existing customer cohort grew or shrank over a defined period — typically twelve months. The formula is (starting cohort revenue + expansion − contraction − churn) / starting cohort revenue, expressed as a percentage. Above 100% means the cohort grew without any new acquisition; below 100% means the brand is on a treadmill where new acquisition has to refill the base before any net growth shows up on the topline.
The three components inside the calculation each describe a distinct behavior on the starting cohort. Expansion is the additional revenue retained customers brought in over the period — plan upgrades, larger basket sizes, higher purchase frequency, add-on SKUs. Contraction is the revenue lost on customers who stayed but spent less — downgrades to cheaper subscription tiers, smaller orders, reduced frequency. Churn is the revenue lost to cancellations or failure to renew. A subscription apparel brand might start a cohort at $100K monthly revenue, see $18K in expansion from upsells and frequency lifts, $6K in contraction from plan downgrades, and $9K lost to churn — landing at 103% NRR.
NRR differs from Gross Revenue Retention (GRR) in one structural way: GRR excludes expansion and caps at 100%, so it shows only the leak; NRR adds expansion back and can exceed 100%, showing whether refills outpace the leak. The two numbers are read together — a brand at 105% NRR on 88% GRR is masking 12 points of base churn with 17 points of expansion. NRR alone makes the cohort look healthy; the GRR pairing reveals the underlying erosion.
A second calculation choice operators face: revenue-NRR versus customer-NRR. Revenue-NRR (the standard, and what the formula above describes) tracks dollars retained from the starting revenue base. Customer-NRR — sometimes called logo-NRR — tracks how many customers retained from the starting headcount, regardless of how much they spent. Subscription-commerce diligence almost always anchors on revenue-NRR; customer-NRR answers a different question and is reported less often.
NRR migrated into DTC vocabulary from B2B SaaS, but it applies cleanly to subscription-commerce, replenishment, and any DTC model where existing-customer revenue is a meaningful share of the business. Investors anchor on it in subscription diligence because it is the cleanest proxy for whether the existing base is an asset or a leak. Read NRR by acquisition cohort rather than blended — a 115% vintage averaged with a 92% vintage tells the operator nothing useful about either. NRR is realized period behavior on a closed cohort; LTV projects the same cohort forward. The two metrics answer different questions, and an LTV forecast built on a blended NRR number is a forecast built on a number that doesn’t exist.