Return on ad spend (ROAS)
ROAS is a number measuring how many dinars of revenue each dinar spent on ads brought back. Calculate it by dividing revenue by ad spend in the same period, giving a number like 2.5 meaning each ad dinar brought back two and a half dinars of revenue. In COD, revenue counts delivered and paid orders, not confirmed orders.
Also calledROASReturn on ad spendAd spend returnReturn on advertisingAd efficiency
The formula
ROAS = Revenue from delivered orders ÷ Ad spend
- Revenue from delivered orders: total value of orders that reached the customer and were paid in cash, excluding refused or returned orders.
- Ad spend: what you paid ad platforms in the same period, including all campaigns that fed these orders.
- The result is a number, not a percentage: 2 means two dinars of revenue per dinar spent on ads.
Worked example
A COD campaign in Algeria
- 1Ad spend for one week: 40,000 DZD.
- 2Confirmed orders: 25, product price 4,500 DZD.
- 3Revenue from confirmed orders: 25 × 4,500 = 112,500 DZD, apparent return 112,500 ÷ 40,000 = 2.81.
- 4Actual delivery rate 60%: 15 delivered orders × 4,500 = 67,500 DZD.
- 5True ROAS: 67,500 ÷ 40,000 = 1.69.
Takeaway: The number on your ad dashboard is built on confirmed orders, and the number that enters your account is built on delivered orders: 2.81 versus 1.69 for the same campaign.
What it means in practice
ROAS measures revenue, not profit. A campaign with ROAS of 2 can lose money if product cost, shipping, and packaging eat more than half the price. The right call is comparing the number to your own break-even threshold, not a rule like "3 is good."
Ad platforms in Algeria, Morocco, and Egypt log a "purchase" the moment the form is submitted—before confirmation call and before delivery. So the platform's number is always optimistic. Multiply the platform's number by (confirmation rate × delivery rate) to get closer to reality before any budget call.
Calculate ROAS per product and per ad angle separately. An average across your whole account hides a product eating budget at 0.8 while another funds it at 3.2, unnoticed.
ROAS moves over time within the same campaign: a product taking three days to confirm and deliver gives a false number if you read it on day one. Lock in a measurement window equal to your delivery cycle.
Common mistakes
- Calculating ROAS from confirmed order revenue in a market where a meaningful share of packages refuse at the door: the number looks profitable but the bank says otherwise.
- Comparing your ROAS to a benchmark you heard in a Facebook group instead of your own break-even, which shifts with product price and cost.
- Counting collection fees the carrier deducts as part of revenue, so you count money that never reaches you.
- Judging a campaign on a small order count: a sample of five orders gives a wobbly ROAS that's no basis for a decision.
Questions and answers
What's a good ROAS for COD?
There's no single number. Good is any number above your break-even threshold, calculated from your margin after product, shipping, and return costs. A seller with 30% margin needs much higher ROAS than one at 60%. Calculate your threshold first, then judge.
Why does my ad dashboard number differ from my bank account?
Because the platform counts the order when the form is submitted, and you count it when cash arrives. Between them is a confirmation call some buyers reject, delivery that fails for some, and packages that return. Same campaign, two numbers—the truer one starts from delivered orders.
Is this number alone enough to judge a campaign?
No. It measures revenue against ad spend only, and ignores product cost, shipping, packaging, and confirmation calls. A campaign with ROAS 3 and weak margin may lose, and one at 1.8 with high margin may win. Read it alongside net margin and order cost.
How do I calculate it for one product inside an account with several?
Put each product in its own campaign or ad set, then tie its spend to its delivered orders from the same period. If two products share one campaign, split the spend by each product's order share; the split is approximate but far better than averaging the whole account.