Free tool
Marginal ROAS measures the return on the next pound of ad spend - the additional revenue it creates divided by the additional spend. Use it to decide whether scaling a channel still pays, even when its average ROAS looks healthy.
Marginal ROAS
3×
Blended ROAS
3.67×
Current avg ROAS
4×
Efficiency change: -25% vs current average
Diminishing but still profitable - the next pound returns less than your average, so scale deliberately and watch for saturation.
Enter what a channel currently spends and returns, then the additional spend you are considering and the revenue you expect it to generate. The calculator returns the marginal ROAS of that increment, your new blended ROAS, and how the next pound compares with your current average. If the marginal ROAS sits below your average, the channel is beginning to saturate - the dynamic explained in marginal ROAS vs average ROAS.
Marginal ROAS is closely related to iROAS (incremental return on ad spend): both ask what spend genuinely creates rather than what it is credited with. To project these curves across a whole plan rather than a single channel, see ad spend forecasting or balance several channels at once with the budget allocation simulator.
Marginal ROAS is the return on ad spend generated by the next increment of budget - the additional revenue produced divided by the additional spend that produced it. It measures the efficiency of your next pound, not the blended average of all spend to date.
Marginal ROAS = additional revenue ÷ additional spend. If increasing spend by £5,000 is forecast to add £15,000 in revenue, the marginal ROAS is 3.0× - regardless of how efficient the existing budget has been.
Budget is decided at the margin. A channel can show a strong average ROAS while the next pound returns far less as it saturates. Planning on marginal ROAS prevents over-investing in channels that look efficient on average but have stopped scaling.
ElenIQ forecasts the marginal return of each channel across your whole plan, so budget moves to where the next pound works hardest.
Explore ad spend forecasting