Return-adjusted ROAS: your platform counts the sale, not the refund
Return-adjusted ROAS is the sales your ads brought in after refunded orders are taken out, judged against a break-even that also carries the cost of handling returns. Your ad platform reports on the sale and your refund arrives later. Enter your own numbers below. The first figure you give is gross: reported before any return adjustment.
Profit per $1 of sales = (1 − r) × m − r × (h + u × (1 − m)). Gross break-even = 1 ÷ that profit. Net break-even = gross break-even × (1 − r), for reports where every refund is already adjusted out. r is the share of value returned, m your margin on a kept sale, h your cost of handling a return as a share of returned value, u the share of returns you cannot resell. Each return refunds the full item value. Use values on the basis your platform reports (tax-inclusive or not). Defaults are illustrative.
Compare gross 4.00 with the gross break-even 3.09, or net 3.20 with the net break-even 2.47, never across. Each $1 of spend keeps about $0.29 (illustrative inputs).
Pick one basis. If you send no adjustments, your platform shows the gross 4.00 and your target is the gross 3.09. Once you upload every refund, it shows the net 3.20 and your target is the net 2.47. If you mix them, you count returns twice. Break-even climbs faster than your return rate: 0% returned needs 2.22 and 40% needs 5.08 (illustrative). Our marketing ROI calculators cover the wider profit maths, and our performance marketing work starts from this figure.
Returns land after the ad system stops listening
Google gives "up to 7 days after they're first recorded for autobidding readability" and says adjustments can be made "within 54 days". If you need autobidding to read refunds, treat day 7 as the deadline; Google does not state the effect of later adjustments.
ROAS seen = gross ROAS × (1 − return rate × share of returned value adjusted in time). Assumes full refunds, each uploaded as soon as you book it (Google suggests uploading at least 24 hours after the conversion), and reads the 7 days as the cut-off for bidding. Enter the shares from your returns system. Defaults are illustrative.
If bidding reads only adjustments made within 7 days, it may see 3.76 against a real 3.20 (illustrative): 0.56 of ROAS it cannot see.
What Google lets you correct, and the deadlines
Google lets you retract or restate a conversion, with up to 7 days for autobidding readability and 54 days for adjustments overall. The table lists each rule beside what you do with it.
| Rule | What Google says | What you do with it |
|---|---|---|
| Retraction | Removes a conversion; value becomes 0.00 | Use for a fully refunded order |
| Restatement | Changes the value, not the count | Use for a part refund |
| Autobidding readability | Up to 7 days after first recorded | Send refunds inside it |
| Overall limit | Adjustments within 54 days; start not stated | Refunds after it stay in your books |
| Final state | Retracted or restated to 0 cannot be adjusted again | Send part refunds as a restatement |
| Conversion window | 30-day default; 1 to 30, 60 or 90 by source | Set it to your buying cycle |
| Reporting date | Click time; conversion-time columns exist | Compare finance by conversion time |
"you have up to 7 days after they're first recorded for autobidding readability"
Google Ads Help, answer 7686280, read 19 November 2026
Conversion value rules "let you adjust your conversion value reporting and Smart Bidding optimization" by audience, location or device. They act on a segment, not an order, and apply to Search, Shopping, Display, Travel and Performance Max. Google cautions that Smart Bidding "already uses signals like geography, device, and first-party audience lists (when applied)", so a rule may add little. Use one only where you can show a segment returns at a different rate; the order-level correction is still the adjustment.
Meta's windows and long buying cycles
In its Insights API, Meta still returns 1, 7 and 28 day click windows, and no longer returns 7 day or 28 day view. Neither platform credits you with a sale that lands after your window, so you lose those sales from your reports.
| Window | Status from 12 January 2026 |
|---|---|
| 1-day click, 7-day click, 28-day click | Continue to be available |
| 1-day engaged view, 1-day view | Continue to be available |
| 7-day view, 28-day view | No longer returned |
A window is the time after an ad interaction in which a sale is credited to it. That caps long cycles: a buyer who clicks, thinks for six weeks and buys falls outside Meta's longest click window and Google's 30-day default. Widen your Google window to your buying cycle, then test your long-cycle prospecting with a holdout, as our incrementality testing guide sets out.
We could not read Meta's own help page on its default setting, so you should check each of your ad sets in Ads Manager. The Meta pages we read describe windows, not return adjustments: compare Meta's value against your net-of-returns sales.
Read your return rate by order cohort, not by refund date
Your return rate for March orders is only final once the March return window closes. Read at day 7, your same orders can show 30% of the real figure, and you would feed that early rate to your calculators.
| When you read it | Refunded so far | Return rate shown |
|---|---|---|
| Day 7 after first recorded | 6,000 | 6% |
| Day 30 | 15,000 | 15% |
| Day 54 | 18,000 | 18% |
| Window closed | 20,000 | 20% |
This cohort gives the second calculator its shares: 30% of returned value inside 7 days and 60% after, with the last 10% after day 54. You will find published rates thin. ASOS's results of 21 November 2025 report a fall of c.150bps in its underlying returns rate, year to 31 August 2025, which it credits to size guides, reviews and a fair use policy, without stating a level. We found no statistics-agency online returns rate for Singapore, Malaysia, Australia, Canada or the US.
Five changes for Monday
The pages we read do not state market restrictions; confirm availability in your market. Your refund timing under local consumer law is a question for your adviser; we did not research it.
- Pull your own return rate by order cohort and product group, after the window closes.
- Run the first calculator and pick one basis: the gross break-even if you send no adjustments, the net break-even if you send every refund.
- Ask your developer to send each refund to Google as an adjustment with the order ID, inside 7 days where you can. Then set Target ROAS from the net break-even, never the gross one.
- Add a conversion value rule only for a segment whose return rate you can show to your team.
- Put your finance team's net-of-returns ROAS next to the platform figure in your monthly report.
Leapbuzz sets this up for ecommerce and retail advertisers on Google Ads, and our ecommerce marketing partner guide lists what to ask of any partner. You can talk to us.
