CAC is not a marketing number

In most plans, customer acquisition cost lives in the marketing section. It sits near the channel strategy, gets a number like "$40," and is never spoken of again.

Then the financial model gets built. The model assumes a growth rate. The growth rate assumes a spend. The spend assumes a CAC — usually a different, more optimistic one, because by then the author is trying to make the model work.

That's the failure. Not the number being wrong; the number being used in two places with two values.

CAC is a cost-of-goods argument

Acquisition cost behaves like a cost of goods, not like an expense line. It's incurred per unit sold, it scales with volume, and it determines gross margin over a customer's life. Call it marketing and you'll budget it annually. Call it what it is and you'll price against it.

Once you see it that way, three things follow that most plans get wrong.

It sets your price floor, not your marketing budget. If a customer costs $200 to acquire and you want them profitable inside a year, you need roughly $600 of contribution in twelve months — because about a third of lifetime value is the most you can spend on acquisition and still fund growth out of margin. At 70% gross margin that's about $71 a month. The price wasn't a positioning decision. It was arithmetic, and positioning had to fit inside it.

It rises with scale, and models almost always assume it falls. Your first hundred customers come from your network, a launch, and people who were already looking — nearly free, and deeply unrepresentative. Customer 5,000 comes from paid acquisition against competitors who have been optimising longer than you have. A model showing CAC declining as volume grows is describing a business with a real network effect. If you don't have one, say so, and draw the curve the other way up.

It's channel-specific, so a blended average hides the only useful information in it. A $60 blended CAC made of one channel at $20 and another at $180 is not a $60 business. It's a $20 business with a leak, and the strategy is obvious the moment you stop averaging. Blending is the most common way founders destroy the signal they most need.

Do it in the right order

This is why order matters more than effort. Work out how you acquire customers and what it actually costs — through the specific channel, with every conversion step multiplied out — then build the model on top of that number.

Do it the other way round and the model tells you what CAC it needs. That number always exists. It's rarely achievable, and it's never tested, because nothing downstream ever checks it against reality.

The one question

If acquisition cost double what I've assumed, does this still work?

Because it might. Channels saturate, auctions get competitive, and the cheap early cohort was never the real number. A plan that survives a 2× CAC has room for the thing that reliably happens. A plan that doesn't isn't a plan — it's a bet on one number staying put, written by the person with the most incentive for it to.