A ratio needs a clearly defined numerator and denominator
Incremental revenue ROAS divides estimated incremental revenue by the spend associated with the tested contrast. It differs from attributed ROAS, which assigns credit to observed purchases. Before calculating the ratio, define the treatment and control spending conditions and the outcome window.
- Define incremental outcome
- Match incremental spend
- Carry uncertainty through
- Compare the commercial threshold
The incrementality guide explains the counterfactual. This article focuses on communicating the uncertainty that remains after the effect is estimated.
The effect you can estimate depends on the intervention, comparison and outcome you define. The incrementality and experiments learning path connects this idea to the assumptions, practical checks and next steps.
To work through the connected decisions, read MMM for profit: choose contribution over revenue alone and Meridian ROI priors: translate experiments carefully.
Carry an illustrative interval through the calculation
Suppose a test estimates £30,000 of incremental revenue with an interval from -£5,000 to £65,000. If incremental spend is a fixed £20,000, the point incremental ROAS is 1.5 and the corresponding interval is -0.25 to 3.25. Report the full range rather than clipping the lower bound to zero.
The interval includes both a negative effect and a positive return. It does not establish that the channel is ineffective, but it also does not establish a profitable effect. State the confidence or credible level and the method used to obtain the interval.
Compare against contribution, not a generic ROAS target
If the relevant contribution margin is 40%, revenue ROAS must exceed 2.5 to cover advertising under that simplified cost definition. In the example, the point estimate is below that threshold while the interval includes values above it. The business decision depends on uncertainty, risk and the cost of further evidence.
Use the break-even ROAS guide to define the margin consistently. Avoid multiplying a revenue estimate by a gross margin that omits important variable costs while describing the result as profit.
Be careful when spend or value is uncertain
Dividing interval endpoints by spend is appropriate in the simple example because the denominator is treated as fixed and positive. If incremental spend is estimated, near zero or correlated with the numerator, ratio uncertainty needs a method that reflects those features.
Likewise, converting incremental orders to revenue adds uncertainty in order value and returns. Use a transparent simulation or suitable statistical method where needed, preserving the dependence structure rather than combining unrelated optimistic assumptions.
When using the result for MMM calibration, retain the tested contrast, outcome, period and uncertainty method. The calibration guide explains why a local experimental ratio should not be pasted into every future channel scenario.
Sources and further practice
This article’s worked example is illustrative, not a measured client result. Use the linked methodology to check the assumptions before applying it.