The problem. The dashboard reports a return on ad spend of four. By the standard of most e-commerce conversations that is a good number. The bank balance disagrees, and has been disagreeing for two quarters.
Both are accurate. ROAS measures revenue against advertising cost and stops there, and for a physical-product business in India the costs it ignores are frequently larger than the one it counts.
What it costs the business
Optimising toward a metric that excludes most of your costs means scaling whatever is worst for you. The products with the highest ROAS are often the cheap ones with thin margins and high return rates. Budget flows toward them because the dashboard rewards it, and every additional rupee of spend loses slightly more money than the last.
The business grows in revenue while shrinking in profit, which is the most demoralising way to run out of money.
What ROAS leaves out
Cost of goods. The largest omission. A ROAS of four on a product with a thirty per cent gross margin is losing money before anything else is counted.
Shipping, and specifically the free shipping you advertise. A real per-order cost that varies by pin code and weight, and is invisible in ad reporting.
Returns and cancellations. Reported revenue is booked at purchase. The refund arrives later and never reaches the campaign report that claimed the sale.
Cash on delivery refusals. Specific to the Indian market and consistently underestimated. An order that is shipped, refused at the door and returned costs you two-way logistics and produces nothing. In some categories this quietly removes a substantial share of reported revenue.
Payment gateway fees, packaging, and the platform commission if you sell on a marketplace. Small individually, meaningful together.
The number to use instead
Contribution margin after advertising. In plain terms: what is left from an order after everything that varies with the order, including the advertising that produced it.
Take the delivered revenue, subtract cost of goods, shipping, gateway fees, packaging and the expected cost of returns and refusals, then subtract the ad spend. If the result is positive, the sale funded itself. If it is negative, more volume makes things worse.
The important word is delivered. For a business with meaningful cash on delivery volume, revenue at checkout is a forecast, not a fact.
Calculate it per product, not per account
Account-level averages hide the thing you need to see. Almost every store has products that fund the business and products that are quietly subsidised by them, and the blended number makes both invisible.
Build the calculation for your ten highest-spend products. It is a spreadsheet exercise, not a software purchase, and most businesses doing it for the first time find at least one product that should not be advertised at all.
What to do with the answer
Stop advertising anything with a persistently negative contribution. Sell it organically or fix the economics, but stop paying to lose money on it.
Set target ROAS per product, derived from margin. A product with a seventy per cent margin can afford a much lower ROAS than one with twenty five. A single account-wide target is arithmetic applied to the wrong unit.
Attack returns and refusals directly. Better product photography and honest sizing information reduce returns. Prepaid incentives, order confirmation messages and address verification reduce refusals. Both improve profit without touching the ad account.
Then look at repeat purchase. If a customer reliably buys again, you can afford to acquire them at a loss. But that has to be a measured rate, not a hopeful assumption, and it is the calculation most likely to be used to justify losses that never reverse.
The short version
ROAS is a media efficiency metric being used as a profitability one. Work out contribution margin after advertising, per product, using delivered rather than booked revenue. The decisions usually become obvious immediately, and several of them will be uncomfortable.
Our e-commerce growth solutions page covers how we structure this, or send us your margins and we will tell you what your real number is.
