Every ad platform reports return on ad spend, and every platform calculates it the same convenient way: revenue attributed to the campaign, divided by what you spent on it. Nothing in that calculation knows what your product costs to make, ship, or take back when a customer returns it.
The number the platform can't see
Consider a product selling at ₹2,000. The campaign reports a 4× ROAS, which sounds comfortable. Then the actual costs arrive:
- Cost of goods: ₹700
- Shipping and packaging: ₹180
- Payment processing: ₹50
- Returns, at a 12% rate: ₹264 averaged across all orders
- Discount code usage, averaged: ₹150
That leaves ₹656 of contribution before any ad spend. At a 4× ROAS you spent ₹500 acquiring that order. You made ₹156 — about 8% of revenue, before a single fixed cost. Scale that campaign and you scale a business that cannot cover its overheads.
The platform is not lying about the arithmetic. It's answering a question that isn't the one you needed answered.
Working out break-even ROAS
Break-even ROAS is simply your selling price divided by your contribution per unit before ad spend. Using the numbers above:
₹2,000 ÷ ₹656 = 3.05
Anything under a 3.05× ROAS loses money on that product. A 4× return is real but thin, and it means the "scale it aggressively" instinct is wrong. You'd want to know that before adding budget.
Why this changes campaign structure
Once you have break-even ROAS per product, campaign structure follows from it rather than from your catalogue. Products fall into margin bands, each band gets its own campaign and its own target, and the algorithm optimises toward something that reflects reality.
The uncomfortable part is that this usually reveals one or two apparently strong products that are losing money on every order — typically high-return-rate items where the return cost was never modelled.
Where to start
- Build a per-SKU contribution model. Landed cost, shipping, processing, average discount and — critically — actual return rate by product.
- Calculate break-even ROAS for each product.
- Group products into margin bands and restructure campaigns around them.
- Set targets per band rather than one account-wide number.
- Re-run the model quarterly. Costs drift, and so do return rates.
None of this is complicated. It's just work that sits outside the ad platform, which is precisely why it so rarely gets done.