Factoring returns into ad profitability: correcting your ROAS target

Same ROAS, different result

The ROAS in your account says nothing about how much revenue is left once the return period has ended. Google counts the value that was reported at purchase. If the customer sends the goods back, the conversion stays in the account unless you correct it yourself. The ad spend never comes back.

It follows that two products with identical ROAS can be economically opposite. One earns money, the other burns it. What makes the difference is almost always the return rate, and that varies widely between product groups.

The calculation with return rate and cost per return

A return costs you twice. The contribution margin of the order is lost, and costs for return shipping, inspection and restocking are added. For break-even ROAS you therefore calculate the expected contribution margin per reported order:

Break-even ROAS = reported value ÷ (CM × (1 − r) − r × C)

Here CM is the contribution margin of an order that is kept, before advertising, r is the return rate by value and C is the cost per return. The formula assumes that returned goods are resold. That is the normal case: according to a retailer survey by the University of Bamberg, 93.2% of returned items went back on sale as new. For goods you have to write off, the value of the goods is added to C.

Worked example, hypothetical: €100 reported value per order, €40 contribution margin, €10 cost per return.

Break-even does not rise linearly with the return rate but ever faster, because the denominator shrinks. From 0 to 20% it moves by 0.83, from 20 to 40% by 1.67. In this example, a ROAS of 4 stays profitable only up to a 30% return rate.

At very high rates the denominator tends towards zero, and then no realistic ROAS is enough. A rule of thumb from Handelsverband Schweiz shows how quickly this builds up: at a 50% return rate, an item is shipped twice on average before it is sold.

The basic formula without returns, and which other costs belong in the contribution margin, are covered in the article Calculating break-even ROAS.

Which rate belongs in the formula

For advertising, the rate that counts is the one by value: returned revenue divided by reported revenue, not the number of items sent back. A shop that keeps many cheap accessories and gets expensive dresses returned has a low rate by units and a high rate by value. Industry studies mostly report the rate by items, and for your calculation it only serves as a guide.

You measure the rate over orders whose return period has ended. If you include the most recent orders, you underestimate the rate, because some returns are still on their way. I would take at least three completed months per product group and look at seasonal outliers such as Christmas separately.

C is rarely known either. According to the EHI study on returns management 2026, about a third of retailers report costs of up to €5 per returned item, roughly a fifth each fall between €5 and €10 and between €10 and €20, and more than a quarter cannot put an exact figure on their own costs. 60.6% now bear the return shipping costs entirely themselves. If you do not know C, work with a cautious assumption and check whether the decision flips when C is twice as high.

Which products stand out

The rate varies by a multiple between categories. According to EHI 2026, 67.8% of fashion retailers report return rates above 20%, while in consumer electronics 94.5% of retailers are at 10% or below and none is above 20%. Across all retailers, two thirds report a rate by items of 10% or below.

For cost per return the order is different. An older EHI analysis puts home and furnishing at €10 to €20 per return because of size and value of the goods. Furniture and lighting therefore often have no extreme rate but a high C.

In your own range, these patterns usually stand out, as an assessment and not as a rule:

• Products with sizes and fit: shoes, clothing, glasses. High rate, because customers order several sizes.
• Bulky or heavy goods: high C, even at a medium rate.
• Products that look different in photos than in reality, such as colours and fabrics.
• High-priced single items, where one return eats up the contribution margin of several orders.

In Germany, the payment method comes on top. The Bamberg survey cites the high share of purchases on invoice and the mostly free returns as causes of the high return rates. A range that runs largely on invoice payment is therefore better off with its own rate than with an industry value.

Returns by channel: a check, not a benchmark

There are no reliable comparative figures on whether Google Ads customers send back more often than customers from other channels. You can check it in your own back end: link orders with their source (UTM parameters or click ID) and return status, and compare the rate by value per channel and product group. Behind a difference between channels, I would suspect a range effect first, for instance because one channel sells more fashion than the other. So compare by product group first, then by channel.

Why a global target ROAS spreads returns wrongly

Target ROAS steers towards an average. According to Google, the system predicts conversions and their values based on the values you report and tries to reach an overall ROAS equal to the target, while individual conversions may come in above or below it. For the bid, one euro of reported revenue from a dress is worth exactly as much as one euro from a coffee grinder.

If a campaign contains products with 5% and with 40% return rates, the bid pays the same price per euro of revenue for both. A global target derived from the average rate is then too strict for the low-return products and too lax for the high-return ones. My assessment: where customers order several sizes and send them back, reported value is easy to get, and that is exactly where a bid that sees only gross values pulls.

Set the target per return class instead of globally

Form product groups by return class, not by margin alone. For each class, you calculate the return-adjusted break-even with its own rate and its own C and set the target above it. In the example above, that would mean the low-return group can run with a target ROAS of around 3, while the high-return group only reaches break-even at 5.0.

The cleanest way to implement this is separate campaigns, because Google does allow individual targets per ad group but, according to its help page, does not recommend them, and because target ROAS is not available as a portfolio strategy for Performance Max. There the target is set per campaign. You do the grouping in the feed with custom labels, and the logic behind it is the same as in target ROAS by margin group. Margin and return class can be combined in one label, as long as this does not create too many groups.

The other option is to report returns straight into the account. Google allows you to retract conversions after a return or to reduce the value for partial returns. This changes the conversion value and affects the ROAS bid strategy, but it requires a transaction ID to have been passed (Google Ads Help). Smart Bidding then sees net values. Google still does not know the cost per return, so it stays in your target. How the transfer works technically is described in Sending returns to Google Ads.

When splitting is not worth it

Separate targets cost data volume. Each campaign needs enough conversions for Smart Bidding to learn stably, see how many conversions target ROAS needs. If the rates of your groups are only a few percentage points apart, the difference in break-even is smaller than the normal fluctuation of ROAS, and a global correction using the average rate is enough.

My rule of thumb: split when the return-adjusted break-even of two groups differs by more than about a quarter and each group has enough volume on its own. Otherwise, check the individual products with the highest return rates first. Often it is enough to exclude a handful of them or to move them into a separate campaign with a strict target.