Net Dollar Retention (NDR)
A revenue retention metric identical to NRR, measuring how much recurring revenue existing customers generate over time including expansion and churn.
If you've read about net revenue retention (NRR) and then stumbled across NDR in an earnings call, you might wonder if they're different metrics. They're not. Same formula, same inputs, same benchmarks. The only difference is where each term shows up. NDR tends to appear in earnings calls, S-1 filings, and investor decks. NRR is the more common label in SaaS operations and analytics platforms like Stripe, Baremetrics, and ChurnWard.
The formula
NDR = (Starting MRR + Expansion − Contraction − Churn) / Starting MRR × 100
If your existing customers generated $108,000 in MRR this month from a starting base of $100,000, your NDR is 108%. That $8,000 net gain comes from upgrades outpacing downgrades and cancellations.
The label doesn't matter. The trend does.
Why companies report NDR
Public SaaS companies favour NDR because it signals expansion efficiency to investors. Snowflake, Datadog, and Twilio have all highlighted NDR above 120% in their filings. It tells Wall Street that existing customers are spending more over time, which means the revenue base compounds without proportional increases in sales and marketing spend.
For earlier-stage companies, we'd recommend tracking NDR (or NRR) monthly. It surfaces problems quickly. A declining trend means either churn is accelerating, expansion is slowing, or both. Catching that early gives you time to act.
NDR vs GRR
NDR includes upsells and cross-sells. Gross revenue retention (GRR) strips them out. A company with 88% GRR and 120% NDR has strong expansion but a leaky bucket. A company with 97% GRR and 108% NDR has a stickier product but less expansion headroom. Investors want to see both because NDR alone can mask retention issues. For the full breakdown, see NRR vs GRR.
Reduce your churn, protect your revenue
ChurnWard recovers failed payments automatically for $29/month. No percentage fees, no complexity.