What's a good magic number for a public SaaS company?

Direct Answer: Magic Number = (Net New ARR added in current quarter × 4) ÷ Prior Quarter S&M Spend. Original definition by Lars Leckie at Scale Venture Partners (Oct 2008, https://blog.scalevp.com/2008/10/the-saas-magic-number/). Public-SaaS healthy band: ≥0.75 (every $1 spent drives ≥$0.75 of annualized new ARR). Below 0.5 you are overspending; sustained >1.0 often signals under-investment in growth. Trend matters more than the absolute value: a falling magic number flags efficiency problems 2–3 quarters before CAC payback gets ugly. As of Q4 2025, the Bessemer Cloud Index median public-SaaS magic number sits at 0.43, down from 0.71 in Q4 2021 (https://www.bvp.com/cloud-index) — the post-ZIRP efficiency reset is visible directly in the metric.

The Detail

Magic Number is the most predictive metric of SaaS unit economics in public markets. Bessemer's State of the Cloud (https://www.bvp.com/atlas/state-of-the-cloud-2026), Meritech's Public SaaS dashboard (https://www.meritechcapital.com/benchmarking/comparables-tables), and ICONIQ's Growth and Efficiency report (https://www.iconiqcapital.com/insights/state-of-saas) all treat it as the primary efficiency lens because it captures the lagged relationship between sales-and-marketing spend (a quarter ago) and net new ARR (this quarter).

Magic Number formula (Scale Venture Partners original, exact):

``` Magic Number = ((ARR_current_qtr − ARR_prior_qtr) × 4) ÷ S&M_spend_prior_qtr

Worked example (Q2 2026):

• Q2 ARR: $50M; Q1 ARR: $47M; Q1 S&M: $8M
• Net new ARR (Q2): $3M → annualized = $12M
• Magic Number: $12M ÷ $8M = 1.50

```

Q4 2025 verified numbers — public-SaaS efficiency reset:

Real public company magic numbers at IPO (filings-sourced):

Bear Case — why magic number can mislead you (adversarial section):

The metric has real limits. Treat it as a coarse signal, not gospel.

1. Numerator gaming via revenue recognition. A company that frontloads multi-year prepaid contracts can show an ARR jump that has nothing to do with this-quarter sales productivity. Salesforce's pre-2019 cRPO disclosure changes are the canonical example — bookings ≠ ARR ≠ revenue, and only ARR (when honestly disclosed) maps to magic number. If "ARR" includes services, professional fees, or one-time setup, the numerator is inflated 5–15%.

1. Denominator distortion via stock-based comp. S&M on the income statement includes SBC for sales leadership and AE equity refreshes. In 2021, Bill.com's S&M-SBC was ~18% of total S&M; magic number computed cash-only would have been ~25% higher (https://www.sec.gov/cgi-bin/browse-edgar?action=getcompany&CIK=0001786352). Compare GAAP-magic vs cash-magic explicitly when SBC > 10% of S&M.

1. Lag mismatch in long sales cycles. The Q-1 → Q lag is right for SMB/mid-market (90–120 day cycle). For enterprise (180–270 days, e.g., Palo Alto Networks, Workday), the correct lag is Q-3 or Q-4. Using Q-1 makes enterprise magic look terrible during ramp and artificially great during contraction.

1. Churn masking. Magic number uses *net* new ARR. A company with 20% gross adds and 18% gross churn looks identical to one with 4% gross adds and 2% gross churn — both yield ~2% net. Always pair magic with gross retention (GRR) and a "gross-magic-number" (gross adds ÷ S&M) to triangulate.

1. PLG distortion. PLG companies (Slack, Zoom, Atlassian, Figma) push acquisition spend into R&D (the product is the funnel). Their reported S&M is artificially low, so magic number is artificially high. Bessemer flagged this in 2024 (https://www.bvp.com/atlas/the-rule-of-40-revisited); cross-check with Rule of 40.

1. The Snowflake counter-example. Snowflake's magic was 1.2 at IPO but cratered to 0.51 by Q4 2024 (https://investors.snowflake.com/financials) despite being a widely-viewed generational compounder. The metric is mean-reverting at scale; using IPO magic to project 5-years-out efficiency overshoots.

1. Roll-up companies. Constellation-Software-style acquirers add inorganic ARR with near-zero marginal S&M, making magic arithmetically infinite for the deal-quarter and grossly understating the M&A spend that did happen. Strip M&A-driven ARR before computing.

Why Magic Number still beats CAC payback for tracking momentum:

Diagnosing a declining magic number:

Step 1 — split marketing magic vs sales magic. Step 2 — AE productivity, sales cycle, deal-size mix. Step 3 — confirm with cohort net retention; if GRR cratered, the magic decline is churn, not acquisition. Step 4 — separate new-logo magic from expansion magic (NRR and gross-new are independent levers; an expansion-driven magic decline is a different problem from a top-of-funnel decline).

Cross-references in this library:

• /knowledge/q73 — CAC payback period and why it lags magic number
• /knowledge/q41 — Net revenue retention (NRR) targets by stage
• /knowledge/q88 — Rule of 40 and how magic number feeds it
• /knowledge/q56 — S&M efficiency benchmarks for PLG vs sales-led GTM
• /knowledge/q22 — Quota attainment and AE productivity bands
• /knowledge/q15 — ARR vs MRR vs cRPO — what to actually report

How to operationalize magic number weekly (not just quarterly):

Public companies report quarterly, but operators should track magic monthly using booked-ARR (not recognized) and trailing-3-month S&M (smoothing seasonality). Build a weekly forward-magic forecast that takes current pipeline × historical close-rate ÷ S&M-spent-90-days-ago. When forward-magic and trailing-magic diverge >15%, the funnel is breaking before the GAAP report shows it.

TAGS: magic-number,saas-metrics,efficiency,go-to-market,unit-economics,scale-venture-partners,bessemer-cloud-index,meritech,iconiq,bear-case,sbc-adjustment,public-saas-benchmarks