N
Glossary
Net Dollar Retention
Net dollar retention is the percentage of recurring revenue an existing customer cohort still produces twelve months later, counting expansion and excluding customers won during the period. Public companies report it under names like dollar-based net retention rate, and each filer writes its own cohort rule, revenue base, and averaging method.
Key Takeaways
Net dollar retention and net revenue retention are one measurement under two names, and public filers use this one.
Datadog reports a range, "in the low-120%'s" as of June 30, 2026, averaged from twelve monthly point-in-time rates.
Okta runs the calculation on annual contract value, not ARR, so its 107% for the twelve months ended July 31, 2026 isn't built like an ARR-based 107%.
Braze publishes two rates for the same trailing year, 110% across all customers and 112% above $500,000 in ARR; monday.com splits the same way at a $50,000 cut, at 110% and 116%.
Confluent annualizes three months of actual cloud consumption, so its 114% carries usage volatility a seat-based filer's number never picks up.
No accounting standard governs NDR, so the filing's methodology paragraph is what makes the percentage readable.
How do public companies define net dollar retention?
Each company writes its own definition into its filings, and the five below disagree on revenue base, cohort, and averaging. I pulled these figures from SEC EDGAR on September 11, 2026.
Company | Filing and period | Name used | Disclosed figure | Base and cohort |
|---|---|---|---|---|
10-Q, quarter ended June 30, 2026 | dollar-based net retention rate | "in the low-120%'s" | ARR, all customers, weighted average of twelve monthly point-in-time rates | |
10-Q, quarter ended July 31, 2026 | Dollar-Based Net Retention Rate | 107%, against 106% a year earlier | ACV, all customers, self-service-only accounts excluded | |
10-Q, trailing year ended July 31, 2026 | dollar-based net retention rate | 110% all customers, 112% above $500,000 ARR | ARR, two cohorts published side by side | |
10-K, year ended December 31, 2025 | dollar-based NRR | 114% | ARR with cloud consumption annualized from three months, point-in-time, adjusted for acquisitions | |
20-F, year ended December 31, 2025 | Net Dollar Retention Rate | 110% overall, 116% above $50,000 ARR | ARR, quarterly point-in-time, size-gated second cohort |
The mechanics live on the net revenue retention page. The problem starts once five filers apply those mechanics five different ways.
Why do two companies report different NDR on the same book?
Methodology choices move the percentage before any customer behavior does, so an identical book produces different rates.
Revenue base. Okta measures annual contract value; the other four measure ARR. Contracted value and run-rate diverge the moment a customer consumes above or below commitment.
Averaging. Datadog and Braze average twelve monthly point-in-time rates, which flattens a single strong month. Confluent and monday.com publish a point-in-time rate, which doesn't.
Cohort gate. Braze's 112% and monday.com's 116% cover larger customers only. Both also publish an all-customer figure two to six points lower, and quoting the gated one without the gate is how comparisons go wrong.
Consumption treatment. Confluent annualizes actual cloud usage from three months into ARR, so a quiet quarter drags the rate down without a cancellation. A filer measuring contracted seats has no equivalent exposure.
Corporate actions. Confluent adjusts its rate for acquisitions, consolidations and spin-offs. Filers silent on this are making a choice too, just not a disclosed one.
What NDR should a private company hold itself to?
Publish the methodology alongside the number and keep it fixed, because nobody outside the company can audit a rate whose definition moves.
Write the cohort rule, revenue base, window and averaging method onto the same slide as the percentage.
Pick one base and stay on it. Switching mid-year produces a trend line that measures your accounting, not your customers.
Report the all-customer rate first. Show a size-gated rate next to it, never instead of it.
Pair NDR with gross revenue retention. A book losing a fifth of its revenue while two accounts expand reads fine on NDR alone and bad on both together.
Restate prior periods whenever you change the method, and say so.
Related terms
Six neighbours from the same investor conversation:
Net revenue retention carries the formula and a worked example.
Gross revenue retention removes expansion and shows what the book holds without upsell.
Annual contract value is the base Okta runs its rate on.
Contracted ARR annualizes signed contracts, a close relative of the ARR base.
Expansion MRR is the component that pushes any of these rates past 100%.
Logo churn counts departures by account, which a dollar-weighted rate hides.
FAQ
Is NDR a GAAP measure?
No. Net dollar retention sits outside GAAP, so no auditor signs off on it and no regulator prescribes how to calculate it. Companies define it themselves in the key business metrics section of their filings, which is why two filers can both be honest and still not be comparable.
What counts as a good net dollar retention rate?
It depends on the definition behind it. Among the filings above, a 2026 infrastructure and application software cohort lands between 107% and the low 120s. A rate gated to large customers runs several points higher than the same company's all-customer rate.
Why do investors prefer the name NDR over NRR?
Public filers standardized on the dollar-based phrasing because it says what's being weighted: dollars, not accounts. "Dollar-based net expansion rate" and "dollar-based net retention rate" are both in active use, and all of them describe one calculation.
Can NDR be compared across two companies?
Only after you read both methodology paragraphs. Check the revenue base, the cohort gate, the window, and whether the rate is point-in-time or averaged. Two of those differing makes the comparison meaningless.
Back to glossary
