Understand why ROAS misses inventory, margin, returns, customer mix, and attribution—and how to make safer ecommerce budget decisions.
Short answer: ROAS divides the revenue attributed to advertising by spend; it does not alone show profitability, stock risk, return rate, new customer quality and attribution difference between platforms. For the right budget decision, advertising data should be evaluated together with store, margin, stock and customer data.
What you need to know before acting on this guide.
ROAS formula:
“Revenue attributable to advertising ÷ ad spend
Example:
This campaign appears to generate 4 TL of attributed revenue for every 1 TL of advertising spend.
But the following information is still missing to decide whether 4 ROAS is good or bad:
A campaign's ROAS might be 5. However, if there is only 8 days of stock left of the product the ad is selling, the budget increase is:
may cause.
Two products can have the same ROAS:
| Product | ROAS | Gross margin | Business result |
|---|---|---|---|
| Product A | 4 | 70% | Healthy |
| Product B | 4 | 25% | Risky |
The same ROAS does not mean the same profitability.
While a retargeting campaign shows high ROAS, most of the sales may come from previous customers who are already close to purchasing. Expanding this campaign without control will not improve new customer acquisition.
Ad revenue is visible instantly; Refunds may occur weeks later. Particularly in fashion and high return rate categories, raw ROAS may overstate true contribution.
Meta, Google Ads, GA4 and the store can write the same order to their own channel with different rules. Therefore, calculating “total advertising revenue” by directly adding platform revenues may create double counting.
MER:
“Total store revenue ÷ total ad spend
Blended shows performance. However, the channel alone does not make the decision.
“New customer acquisition expenditure ÷ number of new customers
If possible, it should be calculated on a new customer basis.
It is evaluated by subtracting the product cost, discount, payment cost, shipping and variable expenses from the revenue.
Is the campaign capturing growth or existing demand?
A customer who contributes low on the initial order can be valuable in the long run. However, this assumption must be validated with segment and cohort data.
“Current stock ÷ average daily sales
It shows the compatibility of the scaling decision with the supply capacity.
Signals that can be observed together:
Large budget decisions should not be made with a small number of conversions.
Policy example:
These values are not the same for every brand; it should be adjusted according to product speed and budget.
“Is this campaign good?” instead of:
“Can the budget for this campaign be safely increased in the next 7 days?
Compared data:
Make sure it is.
-CAC
-CPM
Suggestion example:
“Increase the campaign's budget by 10%. Trust: Medium. There are 54 conversions in the last 7 days, CAC is 12% below target, stock coverage time is 41 days. But creative frequency is rising; New creative must be deployed within 72 hours.
This suggestion is much more usable than the sentence “increase the budget.”
A profitable campaign should not be automatically closed just because the auxiliary KPI is below the target.
For example, CTR may be lower than expected; but:
The commercial outcome of the campaign is a priority.
KPIs are diagnostic signals, not rules.
Optimisium Decide:
“Instead of looking at the ROAS screen, let's create the real business decision together.
There is no universal number. The required ROAS should be calculated based on margin, returns, product cost, customer value and growth targets.
ROAS compares specific ad spend to attributable revenue, while MER compares total ad spend to total store revenue.
There is no single ratio. Incremental changes limited to data volume, campaign stability, inventory and risk appetite are preferred.
Platforms use different attribution windows, models and signals. The same order can be attributed on more than one platform.
Varies depending on scope. It is recommended to use human approval and policy limit for critical actions.