SaaS Magic Number

How do you calculate it?#

1. Choose and label the version#

Never move between them without saying so. Recognised revenue is not ARR, and reported S&M is not acquisition spend.

2. Test the lag rather than assuming it#

One quarter is the convention because it fits a mid-market sales cycle. Enterprise motions with nine-month cycles may need a two- or three-quarter lag; product-led conversion may need less than one. Run the ratio at several lags across your own history and use the one that actually tracks.

3. Decide what is in the denominator#

For the internal version, split reported S&M into new-logo acquisition, expansion and customer marketing, brand and category creation, partner incentives, and commission amortisation. Only the first is causally connected to new-logo output.

4. Bridge the numerator#

Separate new logos, expansion, price, contraction, churn, currency, acquisitions and services. A numerator driven by installed-base expansion set against a denominator of new-logo spend is not a sales-efficiency measurement — it is two unrelated quantities in a fraction.

5. Use rolling and segmented views#

Track the quarterly figure and a rolling four-quarter version. Segment by SMB, mid-market and enterprise, by product and by channel wherever the data allows. A blended 0.75 can be one channel at 1.4 and another at 0.2.

6. Corroborate before acting#

Pair with CAC payback, gross margin, NRR and GRR, quota attainment, win rate and burn multiple. The Magic Number tells you to look; those tell you what you found.

Worked example#

A SaaS company reports Q1 subscription revenue of $3.00M, Q2 subscription revenue of $3.45M, and Q1 sales-and-marketing expense of $2.40M.

sequential increase = $3.45M − $3.00M = $0.45M
annualised          = $0.45M × 4      = $1.80M
Magic Number        = $1.80M ÷ $2.40M = 0.75

By the 2008 thresholds that sits exactly on the line between "step back" and "pour on the gas" — which is a good illustration of why a single reading should not drive a hiring decision.

Decompose the numerator before believing it. Of the $450,000 sequential increase, $250,000 came from existing-customer expansion and a price increase, $150,000 from new customers, and $50,000 from a recently acquired product. If the question is whether the new-logo engine is efficient:

new-logo-only Magic Number = ($150,000 × 4) ÷ $2.40M = 0.25

The headline is 0.75. The part of it attributable to new customers is 0.25. Those two numbers support opposite hiring decisions, and only the decomposition distinguishes them.

The internal version, on the right denominator. Suppose the ARR bridge shows $900,000 of net new ARR from new logos, against $1.50M of Q1 acquisition-specific S&M:

new-logo sales efficiency = $900,000 ÷ $1.50M = 0.60

This is more actionable than the headline and not comparable to it — different numerator basis, narrower denominator. Label both; never present them as versions of the same number.

Sensitivity to the lag. Suppose Q2 S&M was $2.60M and someone uses the current quarter by mistake:

same-quarter denominator = $1.80M ÷ $2.60M = 0.69

A 0.06 swing from a single definitional slip, in a metric whose published decision threshold is 0.75. That is the whole argument for stating the lag every time.

Checking the arithmetic against the original. Leckie's 2008 illustration — recurring revenue of $1.0M, $1.2M, $1.5M across three quarters, with S&M of $800K and $900K in Q1 and Q2 — gives (1.2 − 1.0) × 4 ÷ 0.8 = 1.00 for Q2 and (1.5 − 1.2) × 4 ÷ 0.9 = 1.33 for Q3. Both reproduce exactly.

The caveat that matters: the × 4 asserts that a single quarter's increment recurs for a year. For a subscription business closing evenly through the quarter that is roughly fair. For a business with a large December or a lumpy enterprise close it is a fiction, and the annualisation multiplies the noise by four along with the signal.

What are the common mistakes?#

• Using the current quarter's S&M. The lag is structural, not stylistic — spend precedes revenue. Worth 0.06 in the example above.
• Treating all revenue growth as new-logo output. Expansion, price, acquisitions, currency and services all lift the numerator without the new-logo team touching them.
• Ignoring commission accounting. With commissions amortised over a five-year period of benefit, reported S&M is a weighted blend of five years of selling, not this quarter's cash cost of sales.
• Mixing the ARR and GAAP-revenue versions. They differ by exactly the things ARR excludes: services, non-subscription revenue and recognition timing.
• Applying 0.75 as a grade. In the source it is a spending instruction for a company with a working motion, calibrated on 2008 public SaaS. Scale's own decade-long private median is 0.7x, so "below 0.75" describes roughly half the market.

When does the Magic Number break?#

With seasonality or lumpy enterprise closes. Sequential quarterly revenue is noisy where implementations are scheduled, consumption is seasonal, or a handful of large contracts dominate. Use matched seasonal comparisons or an ARR bridge instead.

When prior-quarter S&M is unusually small. A hiring freeze, a reorganisation of cost classification, or a quarter of deliberate underspend produces a spuriously high multiple in the following period — and a spuriously low one when spend normalises.

After acquisitions. Acquired revenue enters the numerator with no corresponding organic spend in the denominator. Recast prior periods or disclose the discontinuity.

As a measure of returns. The numerator is revenue, not gross profit or cash. A company with 45% gross margins, heavy onboarding cost or weak retention can post an attractive Magic Number and have poor lifetime economics. Cross-check with gross margin and contribution margin.

When pipeline creation and closing sit in different periods. A one-quarter lag punishes healthy pipeline building and then flatters the quarter in which it converts. Look at several lags and cohort payback before changing headcount.