← Ad spend saturation optimizer
Average ROAS is revenue divided by spend over everything the channel has done. Marginal ROAS is what the next dollar returns at today's spend. Ad channels saturate, so the first dollars earn the most and the last earn the least. Every dashboard shows the average; budget decisions are made on the margin.
The tool's built-in invented example: four channels, 78 weeks of spend and revenue, a 60% margin, so a dollar of spend needs about $1.67 of revenue to break even. Each channel's curve was fitted to its own weeks; the marginal figure is the curve's slope at the last eight weeks' average spend, with a 10th to 90th percentile range from 100 resamples.
| Channel | Weekly spend | Average ROAS | Marginal ROAS (range) | Suggested weekly spend |
|---|---|---|---|---|
| Google Search | $5,107 | 3.14 | 2.60 (2.57–2.65) | $8,658 |
| Meta Prospecting | $11,429 | 2.03 | 2.03 (1.97–2.09) | $13,653 |
| TikTok | $4,788 | 1.95 | 0.97 (0.93–1.03) | $3,063 |
| Influencer | $5,316 | 1.35 | 1.11 (1.03–1.18) | $1,265 |
TikTok's average ROAS of 1.95 clears the 1.67 break-even line, so a dashboard calls it healthy. Its marginal ROAS of 0.97 is below the line, and below 1.00: the next dollar returns less than a dollar of revenue. Google Search is the opposite, with a marginal return of 2.60 and room to take more. Moving the same total budget along those slopes lifts fitted revenue by about $4,323 a week (about 7%), and the gain is positive in all 100 resamples. Fitted on the earlier weeks, the curve predicted the final 12 weeks to within about 5% a week, against about 13% for a flat-ROAS line.
This is one invented account whose curves were known in advance, so it shows the method works when its assumptions hold, not that your account looks like this. Real data carries seasonality, promotions, carry-over between weeks and overstated platform attribution, none of which the tool models.
Find the margin in your own channels
Marginal return can only be learned from variation. If a channel spent the same amount every week, no method can tell you the slope.
Divide the extra revenue in your heaviest weeks by the extra spend, relative to your lightest. That ratio is a crude marginal ROAS, and it is usually well below the average.
One divided by your margin. Anything under it on the margin is spend that does not pay for itself on attributed revenue.
Shift part of the budget, wait a few weeks, and check that revenue moved the way the curve said.
Marginal ROAS is the extra revenue you get from the next dollar of ad spend in a channel at your current spend level. Average ROAS is total revenue divided by total spend. With diminishing returns the marginal figure is lower than the average.
One divided by your margin. At a 60% margin you need 1.67 revenue per extra dollar; at 40% you need 2.50. Use the margin on the product sold, before ad spend, and treat the figure as approximate because platform-reported revenue is often overstated.
Yes, if spend varied from week to week. Fit a diminishing-returns curve to weekly spend and revenue per channel and read its slope at today's spend. The free tool does this in your browser and shows a range and a back-test so you can judge how far to trust it.