Return on ad spend is the most quoted number in performance marketing and one of the least useful on its own. It tells you what revenue a campaign produced per pound spent. It tells you nothing about whether that revenue was worth producing.
The number that hides the problem
Two campaigns each return 4x. The first sells a product with a 70% gross margin. The second sells one at 25% after shipping and returns. The first is a business. The second is a way to turn marketing budget into warehouse activity. Reported side by side on a dashboard, they look identical, and the optimisation algorithm will happily scale whichever one is easier to buy.
This is not an exotic edge case. It is the default state of most accounts we audit, because platform reporting is organised around revenue and almost never has your cost data in it.
What to do instead
Get contribution margin per product into the same view as spend. That usually means a spreadsheet before it means a tool, and the spreadsheet is fine. Then set targets per product group rather than per account, because a blended target quietly subsidises your worst performers with your best.
Expect the first version to be uncomfortable. Most accounts, seen this way, turn out to have one or two campaigns carrying everything and a long tail that has been losing money in plain sight for months.