Net Revenue Retention

What NRR measures, precisely

Take the customers you had on day one. Ignore everyone acquired since. Ask what that fixed group is worth now compared to then, counting upgrades, downgrades and cancellations. That ratio is NRR, and it is the cleanest single measure of whether a product deepens or decays inside an account.

Above 100% describes a business with a genuinely unusual property: revenue grows even if acquisition stops entirely. That is why investors weight it so heavily — it separates a product that becomes more valuable as it is used from one that must be resold every month to stand still.

What above 100% does not prove

NRR is a net figure, and netting hides its own components. A business can post 120% NRR while losing a fifth of its customers, if the survivors expand hard enough. Three specific readings go wrong:

• It does not mean customers are staying. Pair it with gross revenue retention, which caps at 100% and cannot be rescued by expansion, and with logo churn.
• It is often driven by very few accounts. If most expansion comes from a handful of customers, NRR is a concentration statistic wearing a retention label, and it will fall the month one of them plateaus. Report expansion concentration next to it.
• Seat-based expansion tracks your customers' headcount, not your product. A per-seat business selling into fast-hiring companies inherits their growth — and inherits the reversal when hiring stops. That is real revenue, but it is not evidence of product value in the way an NRR chart implies.

The cohort must be fixed

The definition depends entirely on the denominator being a closed group. Two errors reopen it. Including customers acquired during the period inflates NRR with new business and turns it into a growth metric. Excluding the cohort's churned customers from the ending measurement — quietly dropping accounts that left — produces a figure that only measures survivors and is systematically too high.

Whether reactivations belong is a real judgement call: a customer who churned in January and returned in June may rejoin their original cohort or count as new. Both conventions exist; treating a reactivation as cohort expansion is the more flattering one, and it should be disclosed if used.

Window length changes the number

NRR over a month and NRR over twelve months are different metrics with the same name. Annual NRR compounds monthly movement and gives expansion far more time to accumulate, so it is almost always the higher figure — a business at 101% monthly is around 113% annually, which is a very different headline from the same underlying behaviour.

Trailing-twelve-month NRR is the convention in investor reporting and is the more stable measure. Monthly NRR is more useful operationally because it responds faster. Just never place them on the same chart, and always state the window when quoting a figure.

How NRR flows into everything else

NRR above 100% means the standard LTV formula understates lifetime value, because it holds ARPA flat while your real accounts grow — which makes every expansion investment look worthless in the model. It also shortens CAC payback, though only if computed properly by accumulating the cohort's actual monthly gross profit rather than by applying a blanket multiplier. And it pairs with the SaaS quick ratio to separate a retention problem from an acquisition one: strong NRR with a weak quick ratio means the base is healthy and new business has stalled.