The case against blended CAC.

A client once fired us for being honest about their CAC. They had a target — let's call it $42 — that the previous agency had been hitting. We told them, after our first month of clean tracking, that the real number was closer to $74. They left. The next agency went back to telling them $42.

This happens more than it should. Blended CAC — total paid spend divided by total new customers, measured monthly — is the most commonly reported marketing metric in DTC. It's also one of the least useful. Most of the founders we work with use it as if it tells them something about whether their paid program is healthy.

It mostly doesn't.

What blended CAC actually measures

It measures the ratio between two numbers, one of which contains a lot of organic, brand, and word-of-mouth traffic the paid program had nothing to do with. The bigger your organic base, the better your blended CAC looks — and the less it tells you about your paid efficiency.

Two brands can have identical blended CACs and wildly different paid programs. The brand with strong organic gravity looks "efficient" no matter how badly its paid is performing. The brand without organic looks "expensive" no matter how well-managed its paid is. Same metric. Different stories. Often the wrong conclusions.

Worse: blended CAC moves in response to things that have nothing to do with marketing. A press hit. A seasonal trend. A new product launch with high organic demand. The number is a financial echo of whatever else is happening in the business.

Blended CAC is a financial echo of whatever else is happening in the business. The paid program is one of those things, but not the only one — and rarely the loudest.

The four CACs we actually report

Here's what we put in our monthly reports for clients instead.

Marginal CAC. The CAC of the last $10K (or $50K, or $100K) of spend, depending on scale. This tells you what one additional dollar of paid spend is currently buying. If marginal CAC is well above your target, you're spending too much. If it's well below, you're underspending. This is the number we optimize against day-to-day.

Channel CAC, properly attributed. Not platform-reported CAC. We use server-side tracking, post-purchase surveys ("how did you hear about us?"), and quarterly incrementality holdouts to triangulate the real CAC of each channel. The numbers usually differ from the platform's by 20-50%.

Paid-only CAC. Paid spend divided by paid-attributed customers only. This is the cleaner version of "blended" CAC: it isolates the program we're actually responsible for. We typically report this alongside organic and direct as separate lines.

Contribution-margin CAC. CAC measured against contribution margin per customer (not against revenue). This is what your CFO actually wants. It accounts for COGS, fulfillment, and variable margin. A "low" CAC against a low-margin product is sometimes worse than a "high" CAC against a high-margin product.

What CFOs should be looking at

If you're a CFO or finance lead reading this, here's the smallest set of metrics you can ask for and still see what's happening:

• Marginal CAC, reported weekly. This is the leading indicator.
• Contribution-margin payback period, reported monthly. How fast does a new customer pay back the cost of acquiring them?
• Incrementality-tested channel CAC, reported quarterly. The platform isn't measuring this for you.
• LTV/CAC by cohort, reported quarterly. The only ratio that matters at the business level.

If your marketing team can't produce these, the issue is the measurement infrastructure, not the marketing. Fix that first — usually a six-to-eight-week project — before optimizing anything else.

What to do tomorrow

If you're running a brand today and you've been managing to blended CAC, the move isn't to throw it out. Keep it as one number among several. But add marginal CAC and a paid-only CAC to your weekly reporting this month. The conversation about what's actually working will change inside a quarter.

That client who fired us went back to optimizing toward a number that wasn't real. We hope they're doing well. We suspect they aren't.