Skip to main content
    Back to blog
    2 November 2026 7 min read

    Blended ROAS Is Lying to You: A Marginal CAC Primer

    Blended ROAS tells you what your average dollar did. Marginal CAC tells you what your next dollar will do — and only one of them should decide where budget goes.

    The comfortable number that hides the truth

    Blended ROAS is the most reassuring metric in DTC — and the easiest to misuse. It takes all revenue, divides it by all spend, and hands you a number the whole team can celebrate. The problem is what it averages away: it treats your first dollar and your ten-thousandth dollar as if they work equally hard. They never do.

    Every channel has diminishing returns. The first $10K into Meta hits your warmest audiences and cleanest intent; the next $10K pushes into colder pools at a worse return. Blended ROAS blends those realities into one average, and averages are exactly where budget decisions go to die. A portfolio can sit at a healthy-looking 3.5 blended ROAS while the last 20% of spend returns 1.2 — money that would work twice as hard anywhere else.

    What marginal CAC actually is

    Marginal CAC answers a sharper question: if I add (or remove) the next $1,000 of spend, what happens to my acquisitions?

    It is not channel-reported CPA. Platform CPA is an average within an auction that credits itself generously. Marginal CAC is measured against a counterfactual: what would we have acquired without that incremental spend?

    You do not need a data science team to estimate it. Three practical approaches, in ascending rigour:

    1. The weekly curve. Plot weekly spend against weekly acquisition volume over 12+ weeks. The slope of that curve near your current spend level is a rough marginal estimate. If doubling spend across a month never came close to doubling volume, your marginal CAC is already well above your reported CPA.
    2. Planned spend steps. Deliberately step budget up (or down) on a channel for two full purchase cycles and watch total outcomes — not platform-reported ones. This is a poor person's incrementality test, and it is far better than nothing.
    3. Geo or audience holdouts. The cleanest read. Withhold the channel in comparable regions or hold back an audience slice, and measure the difference in total sales. This is the same incrementality logic I covered in the incrementality testing starting guide, applied at the budget level.

    Using it without drowning in analysis

    The point is not to compute a perfect marginal CAC for every channel every week. It is to change the budget question from "what is our ROAS?" to "where does the next dollar work hardest?" In practice:

    • Rank channels by best estimate of marginal return, not by platform-reported ROAS.
    • Move budget in steps, not leaps — 10–20% shifts, held for a purchase cycle, so the signal stays readable.
    • Accept a lower blended ROAS as you scale. Growth means buying progressively more expensive customers on purpose. The job is to know the price, not to pretend it is not rising.

    Why this matters more as you grow

    At small spend levels, blended numbers roughly track reality — everything is marginal. The gap between average and marginal widens exactly when the stakes rise: new capital, a board asking for growth, a scale-up mandate. That is when teams that have been managing to blended ROAS discover their "scalable" channels have ceilings.

    If you are approaching that stage, it is worth building the measurement discipline before the pressure arrives. I cover the broader approach — MER, contribution margin, incrementality — on the DTC paid media consulting page, and this is core to how I structure full-funnel media plans with budget gates set at the margins, not the average.

    Frequently asked questions

    Is MER the same problem as blended ROAS? MER is blended ROAS with the brand's blessing — useful as a directional health metric, dangerous as a budget allocation tool. Allocating on MER tells you nothing about which channel deserved the next dollar; it only tells you the portfolio average moved.

    How often should I re-estimate marginal returns? Quarterly for a formal read, continuously through planned spend steps. Marginal curves shift with seasonality, creative freshness and auction competition — a number from last Christmas is not a number for this quarter.