# Judge ads on sales after returns

> A category with strong return on ad spend can lose money once refunds are counted. Connect Shopify, Meta Ads and Google Ads, ask Claude for revenue, refund rate and ad spend by category over six months, and compare ROAS before and after refunds. Lower spend on high-return categories and fix the reason for returns, often sizing.

Type: Play (a chance to grow)
Run it: Quarterly
Source: https://gooseworks.ai/playbooks/measure-roas-after-returns-and-refunds

## 1. Connect

Give your AI assistant read-only access to Shopify, Meta Ads and Google Ads.

## 2. Ask

```text
For each product category, show revenue, refund rate and ad spend for the last 6 months. Work out return on ad spend before and after refunds. Which categories look good before returns and bad after?
```

## 3. Act

Lower bids or spend on high-return categories. Fix the reason for returns (often sizing).

## Steps

1. Connect Shopify, Meta Ads and Google Ads read-only through a Claude connector or MCP server, or export orders, refunds and ad spend and upload them.
2. Ask the playbook prompt to compare ROAS before and after refunds for each product category.
3. Use orders at least 45 days old so refunds have time to come in.
4. Lower bids or spend on high-return categories in steps, and set a review date. Look into the main return reason.
5. Compare net ROAS after the change with the prior period, and re-run each quarter.

## What it tends to find

One category showed 3.1x ROAS, but 34% came back. Net of returns, a 2.4x category won. (Illustrative example, not a result from a named store.)

## Watch out

Give refunds time to come in. Use orders at least 45 days old.

## FAQ

### What is return-adjusted ROAS?

It is return on ad spend after you subtract refunded revenue. Ad platforms usually count the sale when it happens and never see the return, so their ROAS can look better than what you keep.

### Why use orders at least 45 days old?

Returns arrive weeks after the order. If you include recent orders, many refunds have not happened yet, and every category looks better than it really is.

### Should I stop advertising high-return categories?

Not straight away. First lower spend in steps and fix the cause of returns, such as unclear sizing. If net ROAS stays weak after that, cut further.

### Do I need Claude for this, or will any assistant work?

Any assistant that can read your order, refund and ad spend data will work, including ChatGPT, Grok or Muse, from a connection or exported reports.