06-reference

mostly metrics ndr benchmarks

2026-06-21·reference·source: Mostly Metrics·by CJ Gustafson
saas-metricsndrretentionbenchmarksvaluationfpa

"Net Dollar Retention Benchmarks: Where'd All the 130s Go?" — @cjgustafson222

Why this is in the vault

The promised quantitative follow-up to the NDR disclosure piece: original dataset of 95 public software companies across 1,500+ quarterly disclosures, proving the 130% NDR club effectively collapsed and that the published median has ~6-8 points of survivorship bias baked in.

⚠️ Sponsorship

This issue is sponsored by Abacum (FP&A/AI-native planning platform). The sponsor block appears at the top ("Mostly metrics is proudly powered by Abacum") and promotes a CJ-moderated Abacum webinar on FP&A tooling vs. AI-native platforms. Clearly demarcated; no bleed into the NDR analysis. Abacum is a repeat sponsor on this newsletter.

The core argument

Gustafson built an original dataset: 95 public software companies, 1,500+ quarterly NDR disclosures from early 2020 onward. The headline: the 130% club had 18 members in 2020-2021 and now has exactly two (Palantir and Figma).

Key benchmark data:

The expected narrative was a K-shaped bifurcation — winners escaping while losers got crushed. The data rejected that. All quartiles slid in parallel, roughly "everyone moved down a floor."

Survivor bias correction: Companies that stopped disclosing (Fastly, CrowdStrike, DocuSign) had NDRs near or below median at the time of exit. The disclosed median is a median of the survivors. Gustafson estimates the real aggregate compression is 19-21 points, not the headline 13.

Peak-to-trough standout: Snowflake dropped 53 points (177% → 124%) — described as "the most extreme revenue retention compression in public software history." Still top-quartile by current standards, which puts current benchmarks in perspective.

What drove the bleeders:

Valuation translation: Each 10 pts of NDR ≈ 1x forward revenue multiple (0.07x per point at <20% growth, 0.18x per point at >30% growth). For a $250M NTM revenue business at 8x, a 10-pt NDR improvement adds ~$250M EV immediately; if it also bumps analyst growth assumptions, closer to $500M over 18 months. This is the quantitative reason companies hide declining NDR — it's a direct valuation lever, not a vanity metric.

Mapping against Ray Data Co

Mapping strength: medium, with one sharp consulting-context angle.

Honest read: core mechanics (B2B SaaS NDR) are one step removed from RDCO's own surfaces, but the client-facing utility and the Sanity Check content angle both have real legs. The Snowflake data point alone is load-bearing for any phData client conversation touching Snowflake spend justification.

Related