Your SaaS metrics are probably right and still misleading

Most SaaS metric arguments are not disagreements about the arithmetic. They are disagreements about which definition was used.

Two people can look at the same spreadsheet, both be correct, and still quote CAC numbers that differ by 40%.

Here are the three definitions that cause the most damage, and how to pin them down before anyone quotes a ratio.

1. CAC: blended, paid, or fully loaded?

Customer acquisition cost is sales and marketing spend divided by new customers. The division is easy. The two halves are where it goes wrong.

Blended CAC uses total S&M spend over all new customers, including the ones who arrived through word of mouth, organic search, or a partner. It answers "what does growth cost us overall" and it is the number a CFO usually wants.

Paid CAC uses only paid acquisition spend over only the customers attributed to paid channels. It answers "is this channel working" and it is the number a growth team usually wants.

Fully loaded CAC adds the costs most teams leave out: sales tooling, the SDR team's salaries, onboarding engineers if they are dedicated to new logos, and any commission that has not been paid yet. It can be 20-50% higher than the blended figure at companies with heavy sales-assisted motion.

None of these is wrong. Quoting one while the other person assumes another is the bug. The fix is boring and effective: write the definition at the top of the model, and never compare your blended CAC to a benchmark that was computed on a paid basis.

If you want the full breakdown, including which time period the spend should cover, there is a longer walkthrough here: how to calculate CAC without three different answers.

2. ARPU: which denominator, and what counts as revenue?

ARPU is revenue divided by users. Both halves are ambiguous.

Take a product with 10,000 monthly active users, of whom 800 pay. MRR is $32,000.

• Revenue per active user: $3.20
• Revenue per paying user (this is ARPPU): $40

Those are not two estimates of the same thing. They answer different questions. ARPU is useful for a freemium funnel and for capacity planning. ARPPU is useful for pricing. Reporting ARPPU and calling it ARPU usually inflates the number by an order of magnitude and makes every benchmark comparison meaningless.

The numerator is just as slippery. Do you include one-time setup fees, overage charges, professional services, or a prepaid annual plan booked in a single month? The usual convention is to normalise recurring revenue to a monthly basis and keep non-recurring revenue out, because including it makes ARPU spike in the month the contract lands and then fall back.

I wrote the per-plan and per-region version of this, including what to do when a customer moves between plans mid-month, in the ARPU formula guide.

3. Gross margin: what is actually inside COGS?

This one decides whether every downstream metric is right. LTV is built on gross margin, CAC payback is built on gross margin, and the Rule of 40 comparison is only fair if margin is defined consistently.

A defensible SaaS COGS usually contains:

• hosting and infrastructure
• third-party APIs and data costs the product cannot run without
• payment processing fees
• support and customer success staff who serve existing customers
• the delivery/implementation cost, if it is recurring rather than a one-off services line

It usually does not contain R&D, sales, or marketing.

The tricky boundary is support: if the same people handle pre-sales questions and post-sale tickets, you need an allocation rule, and you need to keep it constant across periods or your margin trend becomes an artefact of reclassification.

A 75-80% gross margin is a common target for SaaS, but the benchmark is worthless if half the companies in it book customer support under operating expenses. The COGS boundary discussion is in SaaS gross margin, defined.

Why this compounds

Change the definition of CAC and gross margin together and CAC payback can move by a factor of two:

Same company. Same quarter. Three very different conclusions about whether the growth engine is efficient. Only one of them is comparable to a public benchmark, and you cannot tell which one from the slide.

That is why it is worth fixing the definitions before drawing the chart. I keep a free LTV:CAC calculator that shows blended and fully loaded side by side, precisely so the gap is visible instead of hidden. There is also a longer CAC payback period guide if you want to see how the month boundaries are handled.

A short checklist

Before quoting any SaaS ratio, write down four things:

• Which acquisition cost basis (blended / paid / fully loaded)
• Which denominator for ARPU (active users or paying accounts)
• What is inside COGS, especially support
• Which period the spend covers relative to the revenue it produced

If those four lines are in the model, most metric arguments disappear. If they are not, the argument is not about the numbers at all - it is about the definitions, and it will happen again next month.

If runway and burn rate are also on your plate, the same "which definition" problem applies there, and the burn rate formula breakdown covers net vs gross burn and why the denominator changes the answer.