Blended CAC is the number most teams manage to: total spend divided by total new customers. It's simple, and at small budgets it's good enough. At scale, it starts to mislead.
Why the average hides the problem
Every channel has a curve. The first dollars buy the cheapest customers: people already searching for you, high-intent keywords, warm audiences. As spend rises, each extra dollar reaches colder, more expensive buyers.
Your average rises slowly, because the cheap customers are still in the math. The cost of the last customer rises much faster.
Take a channel spending $200k a month at a $400 blended CAC. Against a $1,600 LTV, that's a healthy 4:1. But if the last $50k of that budget is buying customers at $1,200 each, that slice is paying back at about 1.3:1. The blended number never shows it.
What marginal CAC shows
Marginal CAC is the cost of the next customer, not the average one. You can estimate it by comparing periods: when spend went up, how many extra customers came with it? Divide the extra spend by the extra customers. Do it by channel, and the curve becomes visible.
- Well below target: the channel can take more budget.
- Near target: hold steady.
- Above target: the last dollars in that channel are losing money, even when the blended number looks fine.
How to use it
Set a target from your own numbers: LTV, margin, and how long you can wait for payback. Then move budget between channels until the marginal cost in each is close to the same. That is the point where every dollar is working as hard as it can.
Check it monthly, and more often when spend changes. Curves move with seasons, competition, and creative fatigue. Our approach is built around this weekly.
Where to start
Pull the last six months of spend and new customers by channel. Find the months where spend changed the most, and compare the change in customers. If those extra customers cost far more than your average, you've found your worst dollars.