Glossary - Acquisition and attribution

Marginal CAC

Marginal CAC is the cost of acquiring the next customer, calculated as the change in acquisition spend divided by the change in new customers between two spend levels. Ecommerce operators use it to decide whether the next dollar of ad spend is still buying customers profitably, which average CAC cannot answer.

Formula

Marginal CAC = Δ acquisition spend ÷ Δ new customers

Δ acquisition spend	The extra spend between two levels: this tranche versus the last, or this month versus last month at a comparable baseline.
Δ new customers	The extra first-time customers that additional spend produced over the same period.

Worked example

Baseline month: $100,000 spend, 800 new customersaverage CAC $125

Scaled month: $120,000 spend, 860 new customersaverage CAC $139.53

Extra spend and extra customers$20,000 → 60 customers

Marginal CAC of the last tranche$20,000 ÷ 60 = $333.33

Contribution margin an incremental customer returns inside the payback window$250

Example numbers. The average moved $15 and still looks healthy; the last $20,000 bought customers at $333 each against $250 of margin, so the scaling tranche runs at a loss the average completely hides.

What is a good marginal CAC?

There is no industry number. The ceiling is what an incremental customer is worth to you: the contribution margin they return inside a payback window your cash can fund, which depends on your margin structure, AOV and repeat behaviour. Keep scaling while marginal CAC sits below that ceiling; hold or pull back once it crosses.

Because auctions reach your best prospects first, returns diminish and marginal CAC almost always sits above average CAC, climbing as spend scales. The question is never whether the curve rises, only where it crosses your ceiling - the cost of the next cohort, not the average of the last, is what governs the scale decision.

Marginal CAC vs related metrics

Metric	What it divides	How it differs
Blended CAC	Total sales and marketing spend ÷ all new customers	The average across everything, organic included; context, never a scaling signal.
New-customer CAC	Paid spend ÷ paid-acquired new customers	The paid engine's average; still backward-looking, still flatter than the margin.
CAC payback	CAC ÷ monthly contribution margin per customer	Prices time instead of cost; pairs with marginal CAC to set the spend ceiling.
MER	Total revenue ÷ total marketing spend	An account-wide average ratio; it dilutes the marginal signal the same way average CAC does.

Common mistakes

Making scale decisions on average CAC. On a rising cost curve the average always lags the margin, so the account looks fine right up until the last tranche is deeply underwater.
Counting Δ customers from platform-attributed conversions. Over-claiming tends to grow with retargeting share, inflating the customer delta exactly when spend scales.
Comparing tranches across unlike periods. A promotion or seasonal week inside one tranche corrupts the delta; compare like periods or run the change as a test.
Treating one delta as a stable curve. Marginal CAC is a point on a moving curve; re-measure at each new spend level rather than assuming last quarter's slope.
Judging the marginal customer on revenue. The comparison is against contribution margin inside your payback window, not against AOV or LTV measured in revenue.

Frequently asked questions

Average CAC divides everything spent by everyone acquired; marginal CAC divides only the extra spend by the extra customers it produced. Averages report money already spent, the margin prices the next decision.

Diminishing returns: ad auctions serve your cheapest, most likely buyers first, so each additional dollar buys attention from colder and more expensive audiences. The next customer nearly always costs more than the average one did.

When marginal CAC rises above what an incremental customer returns in contribution margin inside a payback window you can fund. Until then the rising curve is the acceptable price of growth, not a reason to stop.

Inside Blufire, S2 Unit Economics carries Marginal CAC & Saturation alongside CAC, NCAC, CM-payback and LTV:CAC, so the next tranche is judged on the margin, not the average.

Updated July 2026