The cost of acquiring the next or last customer, measured as the change in spend divided by the change in customers; the decision metric for whether to scale.
Marginal CAC measures how effectively the last dollars of the marketing budget are working: the cost of the next incremental customer, not the average customer. Rutherford describes it as a measure of how the least effective, last dollars of the budget are operating, so operators can decide when and where to modify spend (linkedin.com).
Average CAC (blended or new-customer) tells you about money already spent; marginal CAC tells you about the next dollar. Because of diminishing returns, the marginal customer almost always costs more than the average one. Scaling decisions should be made on the margin: keep spending while the next customer's marginal CAC stays below what that customer is worth.
Worked example (Demonstrative): lifting monthly spend from A$100,000 to A$120,000 grows new customers from 400 to 460. The marginal CAC is A$20,000 / 60 = A$333, well above the A$300 average. If contribution-margin LTV is A$900, the marginal customer is still at 2.7:1 and worth buying; if the next A$20,000 only added 40 customers (A$500 marginal CAC), you are approaching the ceiling.
Blufire surfaces marginal CAC against break-even ROAS and MER precisely so operators judge the next incremental dollar rather than a single vanity blended figure. This is where the dollar-impact recommendation lands: keep, hold, or pull the marginal spend.