NRR Is Lying to You

Net Revenue Retention became the north star health metric of the SaaS decade for a good reason: in a seat expansion world, it correlated tightly with the thing boards actually cared about. That correlation breaks down once revenue is driven by consumption rather than seats, and it breaks down in a way that is easy to miss because the number itself keeps looking fine.

NRR is a blended average, which means a shrinking base of struggling customers can hide comfortably behind a handful of expanding accounts, and the topline figure never reveals the split. It also can't distinguish between two very different behaviors that look identical on a usage chart: a customer optimizing their usage to get more value for less spend, and a customer quietly disengaging on their way out the door. And because it's typically reported on a trailing twelve month basis, it tells a board what happened roughly a year ago in a market that now moves in weeks.

The sharpest version of the problem shows up when the product actually gets better. When a model becomes more efficient at solving the same problem, customers can get an identical outcome using meaningfully less usage: real value creation, delivered more cheaply. Under a standard NRR calculation, that improvement registers as revenue contraction. The metric penalizes the company for the thing it should be rewarded for.

The fix is not to abandon retention measurement but to redefine what it measures: outcomes delivered rather than units consumed, weighted by actual contribution margin rather than raw revenue, tracked with usage depth signals that move in near real time rather than on a twelve month lag. Call it Value Realised Retention. Boards celebrating a strong NRR print today may be standing on a number that no longer means what they think it means.