Gross Dollar Retention (MoM)

What it measures

The fraction of the month's opening recurring revenue that survives intact, counting only losses from churn and downgrades. Expansion — upsells, cross-sells, seat additions, price increases — is excluded by design, so GDR can never exceed 100%. That makes it a pure read on how much of the base you keep before any growth motion adds revenue back on top.

Why it matters

GDR isolates the destructive forces of churn and contraction without the flattery of expansion. It answers one brutal question: if we sold nothing new and upsold no one, how much of our revenue would we still have next month? Because it caps at 100%, GDR can't be inflated by a strong expansion engine the way NRR can — a company with heavy churn but stellar upsell can post healthy NRR while GDR exposes the leak underneath. Investors scrutinize GDR to gauge customer stickiness independent of how much you can expand, because a base you can't hold is a base you have to keep re-buying through new sales.

How to read it

Read GDR month-over-month and by customer segment — size, industry, geography, product line. A company-wide GDR of 88% can mask 95% retention in Enterprise and only 70% in SMB, and those two segments need entirely different playbooks. The single most useful read is the gap to NRR: if NRR is 120% but GDR is 80%, you're growing through expansion, not retention, and that only works while expansion stays strong. If both run high, you have a genuinely sticky, growing base; if both run low, you have a crisis. Always decompose GDR into its churn and downsell halves before you act — losing a marquee logo is a different problem from broad, low-level downgrading.

What good looks like

Watch-outs

• Letting expansion sneak into the numerator. Upsells, cross-sells, seat adds, and price increases must be excluded — GDR measures only losses. If a single dollar of expansion touches the calculation, GDR can read above 100%, which is impossible, and you've quietly computed NRR instead, losing the diagnostic power to see how much the base is actually leaking.
• Counting downgrades as if they don't matter. GDR subtracts both churn and downsell — a base that holds every logo but bleeds revenue to downgrades is still eroding. Treating GDR as churn-only overstates retention and hides a pricing or value-delivery problem.
• Running GDR off live ARR instead of the opening cohort. Calculate from the period's Starting MRR (prior-month Closing MRR), not a moving live-revenue figure that already nets in mid-period changes. A shifted base destroys the month-over-month trend and makes comparisons meaningless. Use live, billing contracts only — signed-but-not-live deals inflate the base and mask churn.
• Reading company-wide GDR in isolation. A blended 88% can hide a segment at 65% and another at 98%. Slice by vintage, segment, product line, and deal size to find where churn and downgrades concentrate, and read GDR dollar-weighted — losing one $50K account hurts far more than losing ten $1K accounts.

Worked example