Where a 4x return actually goes
- Ad spend: AED 100,000. Revenue attributed at 4x: AED 400,000.
- Less the 5% VAT inside that gross figure: net revenue AED 380,952.
- Less cost of goods at 60% of net revenue: minus AED 228,571, leaving AED 152,381.
- Less delivery and fulfilment at 10%: minus AED 38,095, leaving AED 114,286.
- Less payment and marketplace fees at 2.5%: minus AED 9,524, leaving AED 104,762.
- Less a 15% return and cancellation rate applied to what is left: minus AED 15,714, leaving AED 89,048.
- Against AED 100,000 of ad spend, that is a loss of AED 10,952 on the month.
Deriving your own break-even ROAS
Take one AED of net revenue and subtract cost of goods, delivery and fees as shares of it. Reduce what remains by your return rate. Divide by 1.05 to move from the gross order value the platform reports back to net revenue. Break-even ROAS is one divided by that figure.
On the inputs above: 1 minus 0.60 minus 0.10 minus 0.025 leaves 0.275; after a 15% return rate, 0.234; divided by 1.05, 0.223. Break-even is 4.5x, which is why 4x loses money.
- Contribution of 30% of gross order value: break-even ROAS 3.5x.
- Contribution of 20%: break-even ROAS 5.3x.
- Contribution of 15%: break-even ROAS 7.0x.
- The same 4x is comfortably profitable in the first business and heavily loss-making in the third.
A target nobody derived is a target nobody can defend
Advertising platforms optimise towards a ROAS target that someone types in. When that target was inherited, copied from a competitor, or set at a round number, the platform will hit it precisely and the business can still lose money at the same time.
The break-even figure above is the floor, not the goal. Fixed costs, the cost of holding stock, and the return the capital could earn elsewhere all sit above it.
Contribution is still not the whole answer
Some of the revenue a channel reports would have happened without it. Existing customers search for a brand they already intended to buy from, and the platform books the sale. A channel reporting 6x can be closer to 2x once that is removed.
No amount of reporting settles this, because every platform is measuring its own contribution with its own rules. A holdout, a geographic split, or a staged reduction measured against a baseline agreed in advance is the only evidence that separates caused revenue from recorded revenue.