Return on ad spend is the most quoted number in performance marketing and one of the least useful on its own. It divides revenue the platform believes it caused by the money you gave that platform. Nothing in that calculation knows what your product costs to make or what shipping costs. It does not know how many orders come back, or whether the sale would have happened anyway.
We have sat in meetings where a founder celebrated a four times return in a category that cannot survive below six. We have also sat in meetings where a two times return quietly funded the whole company. Both rooms were reading the same metric.
ROAS is not profit: the arithmetic nobody puts on the dashboard
Start with your break even return, which is simply one divided by your gross margin. At a forty percent margin you break even at 2.5 times. At a twenty five percent margin, common in electronics and grocery, you need 4 times before a single pound reaches the business. Any target you set without that number in front of you is decoration.
Then subtract the costs the platform never sees. Delivery, packaging and payment fees. Discount codes applied at checkout. Returns, which in this market mostly means cash on delivery orders refused at the door. That outcome carries the shipping cost both ways and brings no revenue at all. Depending on the category and the courier, that write off runs anywhere from ten to thirty percent of reported orders. We explain why in our cash on delivery measurement fix.
Run those through the example and a headline four times return in a twenty five percent margin business is a loss. It looks like success on the screen because the screen was never measuring profit.
Here is a hypothetical illustration, not a client result. Say 100,000 pounds of spend reports 400,000 pounds of revenue, a return of 4.0. If 15 percent of orders come back refused, 340,000 pounds of revenue stays. At a 25 percent margin that is 85,000 pounds of gross profit, against 100,000 pounds of ad spend. The campaign lost 15,000 pounds before a single delivery cost. The dashboard still shows a four.
Platform return versus blended return
The second problem is credit. Every platform reports the conversions it can claim, so the same order can appear in two dashboards and both will count it. Since the tracking changes of the last eighteen months, that overlap has grown rather than shrunk. More of what you see is modeled rather than observed. Our note on how iOS 14.5 breaks Facebook targeting covers where those changes began.
The honest version is blended: total revenue in your own commerce system divided by total media spend everywhere. It is crude, it flatters nobody, and nobody can game it. When platform reported return climbs while blended return stays flat, you have not improved anything. You have simply bought more of the demand that was already coming.
Optimize on contribution, not revenue
The fix is to feed margin back instead of order value. Send the platform the gross profit of each order as the conversion value rather than the sale price. Meta documents how to pass values from your own systems through its Conversions API. The bidding then learns to chase the products that actually pay, instead of the cheap high volume line that inflates revenue and eats the margin. On accounts where we have made that switch the reported return usually drops and the bank balance improves. It is an uncomfortable but very clarifying conversation to have with a board.
Set the target as a contribution figure per order and a blended acquisition cost the business can afford. Keep return on ad spend as a diagnostic for comparing campaigns against each other, which is the job it is genuinely good at. Stop presenting it as proof that the marketing worked.
Building reporting that leads with contribution rather than a flattering multiple is how our paid media advertising and planning team runs an account, and it is usually the first thing we change.
