Skip to content
Unit Economics

What is Marginal CAC?

Marginal Customer Acquisition Cost

Marginal CAC is the cost of acquiring one additional customer at your current spend level — the change in spend divided by the change in customers between two points — as distinct from average (blended) CAC, which is total spend divided by total customers.

Blended CAC answers a backward-looking question: what did all your customers cost, on average, so far. Marginal CAC answers the forward-looking one that actually matters for a scaling decision: what would the next customer cost if you spent more right now. The two numbers can diverge enormously, and the gap between them is exactly what a flat, healthy-looking blended CAC hides.

The reason this matters is that almost every acquisition channel has a rising marginal cost curve — the cheapest, most responsive audience segment gets exhausted first, and each additional dollar of spend increasingly buys colder, less-likely-to-convert prospects. Blended CAC averages this all together, so it can look stable even while the actual cost of the next customer is climbing fast.

The practical use isn't the number in isolation — it's comparing marginal CAC against contribution margin per customer. As long as marginal CAC sits below what a customer contributes, scaling adds profit. The moment marginal CAC crosses that line, every additional dollar of spend is buying a loss, even while the blended average still looks fine.

Formula

Marginal CAC = (Spend at Level 2 − Spend at Level 1) ÷ (Customers at Level 2 − Customers at Level 1)

Calculate it between two real spend tiers you've actually run — this week's spend vs. last week's, or a deliberate budget step-up test — not a theoretical curve. It only means something as a comparison between two observed points.

Why Marginal CAC matters

It's the number that actually tells you whether to keep scaling a channel, which blended CAC structurally cannot do — a flat or even improving blended average can coexist with a marginal cost that's already unprofitable, because the average is diluted by the cheaper customers acquired earlier.

The scale-up that looked fine and wasn't

A team spends $10,000 and acquires 100 customers — $100 blended CAC. They scale to $40,000 and get 200 total customers — blended CAC rises to $200, which still looks tolerable against a $250 contribution margin per customer. But the marginal math tells a different story: the additional $30,000 in spend bought only 100 additional customers, a marginal CAC of $300 — already above the $250 contribution margin. The last tranche of spend was losing $50 per customer, and the blended number never showed it because it was still averaged against the cheap first 100.

Common mistakes

  • Reading a healthy blended CAC as room to scale

    Blended CAC is arithmetically incapable of showing where the damage is concentrated. A flat or slowly-rising blended average can hide a marginal cost that's already well past the point of profitability.

  • Comparing marginal CAC to revenue instead of contribution margin

    The relevant ceiling is what a customer actually contributes after variable costs, not their revenue. Comparing marginal CAC to gross revenue per customer overstates how much room a channel actually has.

  • Computing it from too small a spend change

    A tiny budget nudge produces a noisy, unreliable marginal CAC estimate. A meaningful, deliberate spend step — large enough that the resulting customer-count change isn't mostly noise — gives a marginal CAC worth acting on.

  • Treating it as a fixed number instead of a moving one

    Marginal CAC shifts with seasonality, competitive bidding, and creative fatigue — a marginal CAC calculated last quarter isn't a permanent scaling ceiling. It needs recalculating at each meaningful spend change, not set once and referenced indefinitely.

Applied, not theoretical

We'll run these numbers on your account.

A free 30-minute teardown where we calculate this and the rest of your funnel math live. You keep the model whether or not we work together.