On this page
Net revenue retention is a cohort definition before it is a growth metric. The familiar arithmetic is straightforward. The starting population, revenue boundary and observation window are where the meaning lives.
A simplified practical expression is:
NRR = (starting recurring revenue minus churn minus contraction plus expansion) divided by starting recurring revenue.
The formula describes a fixed base. If new customers enter the numerator, or if the revenue definition changes between periods, the result is no longer a clean read of what happened to the starting cohort.
Why is Net Revenue Retention fundamentally a cohort definition?
The first line in an NRR calculation should name the customers who were eligible at the start of the window. “Customers at the beginning of the year” is not enough if the company changes its definition of customer, excludes small accounts, moves accounts between products or treats an acquisition as organic expansion.
The starting revenue also needs a definition. Contracted recurring revenue, recognised revenue, subscription and support revenue, usage revenue and services revenue can each produce a different metric. None is automatically wrong. They answer different questions.
| Term | Include | Decide before calculating |
|---|---|---|
| Starting cohort | Customers eligible at the beginning of the window | Customer status, product scope and acquisition treatment |
| Starting revenue | Recurring revenue attributed to that cohort | Contracted, recognised, subscription, usage or total recurring basis |
| Churn | Revenue removed because the customer leaves | Full account, product, region and timing treatment |
| Contraction | Revenue lost while the customer remains | Downgrade, seat reduction, usage decline and price change |
| Expansion | Additional revenue from the starting cohort | Cross-sell, upsell, price increase and usage growth |
| Exclusions | Items intentionally left out | Acquisitions, currency, credits, services and one-time items |
Table 1The NRR formula with its boundaries restored
The percentage is the last step. The boundaries are the metric.
Source: Author's operating worksheet. Definitions must be checked against the issuer's own filing.
A stable number can still change its meaning
The definition has a shorter memory than the number shows the practical problem. A company can continue publishing NRR while changing how it defines the metric. The historical series remains numerically tidy. The comparison is not.
The right response is not to distrust every disclosure. It is to keep the definition beside the number. When the company changes the cohort, the revenue basis, the period or the treatment of currency, record the change as part of the data. A recalculated prior period can improve comparability. An unannounced change should not be silently treated as a continuation.
There is also a difference between a metric disappearing and a cohort disappearing. The metric did not die. The cohort did tracks the public disclosure question. The commercial reader still needs the private version: which customers are in the cohort, what they bought, what expanded and which definition was used.
NRR is not a customer-quality score
An NRR above 100% means that the starting cohort produced more revenue on the chosen basis by the end of the window. It does not mean every customer expanded. A few large expansions can offset many small contractions. The metric is therefore a portfolio result, not a quality score for an individual account.
The same issue appears across segments. Enterprise NRR can be high because a small number of large accounts expanded. Self-serve NRR can be lower because the cohort has more small accounts and more early churn. Comparing the percentages without the starting distribution can turn a mix difference into a product conclusion.
The metric also does not show who owns the relationship. If expansion relies on a salesperson or partner, the future value may not be transferable even when the historical cohort looks strong. That is why a retention review belongs beside the relationship ownership questions in what the customer relationship costs while it stays.
| Check | Yes means | No means |
|---|---|---|
| Same cohort rule | The starting populations are constructed the same way | The percentages are not directly comparable |
| Same revenue basis | Both measures use the same recurring revenue concept | Explain the boundary before comparing |
| Same window | The periods have the same length and timing | Seasonality may be part of the difference |
| Same currency treatment | Foreign exchange is handled consistently | Separate price, usage and currency effects |
| Same definition over time | The series has a stable meaning | Mark a break or restate the history |
| Same segment mix | The populations have a comparable distribution | Interpret the result as portfolio composition |
Table 2The NRR comparability audit
A percentage earns comparison only after its construction survives the same six questions.
Source: Author's worksheet for reading public retention disclosures and internal cohort tables.
Put the metric into the forecast carefully
NRR is useful in a forecast when the forecast carries the conditions. Start with the cohort, split churn, contraction and expansion, and show the distribution behind the aggregate. Then test what happens if a large expansion account is removed, if the currency effect is isolated, or if one segment is reweighted.
Do not use NRR to conceal acquisition economics. New customers do not repair a weak starting cohort inside the metric. Do not use it to stand in for customer lifetime value. LTV also needs contribution margin, cost to serve, acquisition cost and a horizon.
The final rule is simple: keep the definition on the same page as the number. If the definition cannot be reconstructed from the data, the percentage can still be reported as management’s chosen metric. It is not yet a comparable measure of the customer base.
Evidence base. The analytical frame also draws on these additional sources: U.S. Securities and Exchange Commission 2020; Ordway Labs 2024; Gustafson 2024. The links identify the exact works; they support the mechanisms and boundary conditions discussed here, not every claim in isolation.
References
- U.S. Securities and Exchange Commission. (2020). Commission guidance on management's discussion and analysis of financial condition and results of operations (Release No. 33-10751). https://www.sec.gov/rules/interp/2020/33-10751.pdf
- Ordway Labs. (2024). Dollar-based net retention rate: How public companies report. https://ordwaylabs.com/resources/research/how-public-companies-calculate-net-revenue-retention/
- Gustafson, C. J. (2024). NDR is slowly being euthanized. Mostly Metrics. https://www.mostlymetrics.com/p/ndr-is-slowly-being-euthanized-3cf4
