In ecommerce, a sale at checkout is not always net profit. Returns, refunds, cancellations, shipping, payment fees, discounts and other costs can erode the margin. Yet advertising platforms may record these sales at full value based on gross revenue, without reflecting what the business actually retains.
The problem with gross ROAS
Using gross ROAS as the main signal for Smart Bidding can cause problems. The algorithm optimizes for the value first reported, even if an order later loses value. If Google Ads receives only the gross order value, it may favor orders and customers that seem profitable at first but ultimately reduce margin.
If the advertising system receives only the gross value of the order, it will tend to optimize towards orders that look good at the moment when conversion takes place. But if a part of those orders degrades later, or arises from low-quality traffic, the budget can be moved to patterns that the business should not reward.
Checkout is only the beginning
Checkout is often considered the goal of advertising conversion. For the business, however, it is only the starting point of a long verification process. Between evasion, delivery, returns, disputes and reconciliation, only in the end can you understand how much value really remained. And this process is far from marginal: according to the National Retail Federation, in 2025 19.3% of online sales will be returned, and of these 9% will be rendered fraudulent.
In Europe, then, the right of withdrawal within 14 days without justification makes the economic value of initial conversions even more unstable.
The right approach: connect Ads, traffic quality and real economic value
The solution is not to abandon ROAS, nor replace it with a more complex metric just to make reporting more sophisticated. The point is to make ROAS less naive, connecting advertising not only to the final economic value, but also to the quality of traffic and orders that generate it.
Google and the main advertising platforms are already recognizing the importance of richer signals. Google Ads allows you to combine conversion data with shopping cart information and sales cost to read better margins and gross profit. Google Analytics 4 supports refund events, useful to measure refunds even at the product level.
Why gross ROAS is only part of the story
A high ROAS can hide repaid orders, low margin products, unsustainable customers and traffic that should not have affected the bidding. If these signals do not fall into advertising platforms, the algorithm continues to learn from an incomplete version of reality.
Adefence helps ecommerce close this gap: it analyzes the quality of traffic, follows the evolution of orders and connects real business events to campaigns. So the budget is not only driven by recorded sales, but by cleaner signals, more reliable and closer to the value that really remains.
Questions and answers
Why does a high ROAS not guarantee a good margin?
The gross ROAS only measures the relationship between advertising spending and generated turnover, without looking at what happens after checkout: returns, chargeback, management costs and fraudulent orders erode the real margin even when the ROAS looks great. A campaign can appear winning on the card and still generate losses, because the data does not distinguish between a healthy order and one that will be refunded or contested.
What happens after the checkout that ROAS does not see?
After purchase, returns, chargebacks, frauds with stolen cards or coupon abuses and loyalty programs can occur: events that reset or reduce the value of the original conversion. The gross ROAS remains unchanged, because it is calculated at the time of purchase and not updated according to what happens next.
How do you connect traffic quality and real economic value?
It takes a flow that brings back to advertising platforms not only the fact that there was a sale, but also its real outcome: if it was confirmed, surrendered or contested. Connecting order management tools to campaigns, the offer algorithm learns to recognize which traffic sources really generate net value and what only apparent turnover.
What is NetROAS and what differs from gross ROAS?
NetROAS is an indicator that removes gross turnover generated by the campaigns the costs of returns, fraud and chargeback, returning the economic value that really remains in the company. Unlike gross ROAS, it allows to optimize campaigns on actual profit and not only on recorded sales volume.
How does Adefence help switch from gross ROAS to NetROAS?
Adefence analyzes traffic after clicking and integrates data on the actual outcome of orders, so advertising platforms can optimize bidding on net value rather than gross turnover. In this way the budget focuses on traffic sources that generate real margin, excluding those that only inflate numbers.
How long does it spend on average before a return is recorded and correct in data?
Usually the return takes a few days to be processed, so the delay between the recorded conversion and the correct data can go from a few hours up to two or three weeks, depending on the return policy of your ecommerce. This delay is one of the reasons why the gross ROAS, calculated immediately at the time of purchase, never reflects the complete reality. Updating data with a controlled delay, but still regular, allows the algorithm to progressively correct its decisions without losing reactivity in the short period. Expecting too long before correcting the signal, instead, is likely to make the update not useful for optimization.
Does the gross ROAS problem concern only those who sell physical products with possibility of return?
Yes, though differently. A digital service has not made physical, but can still have refunds, cancellations after the trial period or chargeback on subscriptions, all events that reduce the real value of a registered conversion as sale. The principle at the base of NetROAS — look at what really remains after conversion, not only what is recorded at the time of purchase — is therefore also applied to business models without physical shipment. The events to be monitored change, but the underlying logic remains identical: the initial gross turnover often does not coincide with the value that the company really holds.
All returns should be considered as a traffic quality problem?
No, and it's a common mistake to think it. Not all returns are a negative signal to be eliminated: a physiological return rate exists in practically every ecommerce sector, linked to wrong sizes, honest rethinks or products that do not match expectations. The problem arises when returns become systematic on specific sources of traffic, orders or patterns of behavior, a sign that conversion was probably never really genuine. The aim of NetROAS is not to zero returns, but to distinguish physiological ones from those that indicate a quality problem of upstream traffic.
What do you really need to start working with NetROAS?
Passing to NetROAS requires first of all to have access to the data on the actual results of orders: returns, chargeback, cancellations and confirmed fraud, usually already present in your ecommerce or management system. The second element is a way to reconnect these results to the original advertising campaign that generated that conversion, which requires a consistent tracking throughout the funnel. Finally, it serves a channel to send this correct data back to advertising platforms, so that the bidding algorithm can actually use it for its future decisions.