06-reference

mostly metrics complete guide to arr

2026-06-11·reference·source: Mostly Metrics·by CJ Gustafson

"Your Complete Guide to ARR" — @CJ Gustafson

Email subject: "A taste of the wild side" (2026-06-11). This issue is a free unlock of one of CJ's most-read paid posts, sent to heavy readers of the free list as a paid-tier funnel. The post content is real and complete, not a teaser. No canonical URL in the email — the post lives behind the mostlymetrics.com paywall, so source_url points at the publication root.

Why this is in the vault

This is the cleanest single reference we have on the six ways SaaS companies describe topline — and, more usefully, on how the dishonest version (CARR-quoted-as-ARR) works mechanically. It doubles as a paid-content quality sample for the on-hold Mostly Metrics paid-tier decision, and as a voice specimen for the Sanity Check study of CJ's register.

The core argument

Not all topline is equal, and the labels are being abused (marketplaces calling GMV "revenue," services firms calling one-time work "ARR"). CJ's taxonomy, paraphrased:

  1. Revenue (GAAP) — accounting view; accrued to match delivery, so it trails ARR and billings in SaaS.
  2. Deferred Revenue — prepaid-but-undelivered; a balance-sheet liability (month 4 of a 12-month contract = 8 months deferred).
  3. RPO — all unrecognized contracted revenue, including the unbilled tail of multi-year deals. Actually IS GAAP, but lives in the financial-statement notes. The "backlog" metric for consumption businesses — illustrated with a Snowflake investor-day slide.
  4. GMV — marketplace/payments throughput, not revenue you keep.
  5. ARR — annualized run-rate of committed subscription contracts as of the measurement date, assuming renewals at existing terms. His PSA: one-time purchases, consulting, and most ecommerce don't qualify.
  6. CARR — contracted ARR using the exit year of a multi-year deal. The trap metric.

The CARR section is the sharpest part. First-year discounting, license ramps, and embedded price escalators all make a deal's final year larger than its current year, which creates "a perverse incentive" to quote the biggest annual figure. Doing so inflates both the ARR level and the YoY growth rate — flattering during a fundraise — but ARR should track GAAP revenue closely, so the gap widens as multi-year deals pile up. Investors eventually spot the divergence, their cash forecasts break, and the company faces a restatement and a trust breach. CJ frames this from experience, not hypothetically. His fix: if you want a de-risked-future number, RPO already does that job honestly.

Closes with the ARR waterfall build: Starting ARR (prior period's Ending, corkscrewed in the model) + New Business + Expansion (upgrades, usage, new products, price increases) + Reactivation, minus Contraction and Churn. The New-vs-Expansion mix is a sales-motion signature — land-and-expand and multi-product companies skew toward expansion; deal-is-the-deal companies don't.

⚠️ Sponsorship

No third-party sponsor in this issue (scanned against the known sponsor set: Brex, Intuit, Samsara, Rivian, MLB, Abacum, Mostly Talent — none present; one "MLB" string match was a tracking-token false positive). The issue IS self-promotional end to end: "giving away our best paid post" is a conversion funnel, bracketed by a 7-day free-trial CTA up top and an upgrade CTA at the bottom. sponsor_entity: self. The content itself is straight definitional material with no product placement, so the bias risk is framing (paid tier looks generous), not substance.

Relevant to the open decision: founder put the Mostly Metrics paid-tier call ON HOLD on 2026-05-11, re-evaluate after 4-6 weeks of free issues. That window is now open, and this unlock is a direct sample of paid-tier quality: high-utility reference material, well written, but definitional rather than novel — the free issues have been delivering comparable analytical value.

Mapping against Ray Data Co

phData consulting (CFO/finance-metrics literacy) — the strongest lens. The RPO-as-backlog framing is the metric of record for consumption businesses, and the post literally illustrates deferred revenue with a Snowflake investor-day slide — Snowflake being the ecosystem the founder consults in and is certifying on. Knowing how RPO, deferred revenue, and ARR diverge on multi-year consumption deals is directly usable vocabulary in client-facing finance conversations. Pairs with [[2026-06-04-mostly-metrics-consumption-based-arr]].

Sanity Check voice study — good specimen. This is CJ's register at full strength: regional self-deprecation ("the Boston in me"), PSA-style bullet rants, joke-deflated jargon ("the super official handbook for bean counters"), and stakes-driven structure — definitional content made readable by attaching it to an investor-trust horror story. The transferable craft lesson: a glossary becomes an essay when each definition is positioned as a way someone lies to you.

RDCO's own products — narrow but pointed. Squarely's one-time app purchases are precisely what CJ's PSA says not to call ARR. If RDCO ever quotes recurring metrics (newsletter subs, app subscriptions), use current-period figures, never exit-year contract values.

Honest verdict: medium. High lookup utility and a strong voice specimen, but it's reference material — no strategic re-frame for RDCO, and no Sanity Check piece should be derived from it (it would restate the source).

Related