Returns and Size Data Feedback Loops

A size feedback loop is a routine that sends what you learn from returns and size data back into your size chart, patterns and product pages. Each month you collect fit-related returns, find the pattern, make one change, and check the next month whether it worked.

It is the practical side of AI size recommendation for fashion stores: a recommendation is only as good as the chart behind it, and a feedback loop keeps that chart accurate.

What is a size feedback loop?

It is a cycle with four steps:

1. Collect. Record why items come back and which size was ordered.
2. Find the pattern. Group the data by product and size.
3. Change one thing. Fix the chart, the pattern or the page.
4. Check. See whether the next batch of returns improves.

Without the loop, the same fit problem repeats every season. With it, every return teaches you something.

What data should you collect?

You need only a few fields on every return:

Field	Why it matters
Product and variant	Shows which size has the problem
Reason code	Separates fit from quality or taste
Fit direction	Too small, too large, too short or too long
Size recommended, if any	Shows if the suggestion was followed
Date	Lets you compare months

Use fixed choices for reasons, not free text. The guide to return reason codes shows how to set these up, and analyzing return reasons by product shows how to read them.

How do you read the pattern?

Look for repeated signals, not single returns. A few examples:

• One size returns "too small" more than others. That size's range in your chart may be too low.
• All sizes of one product return "too small". The product may run small. Add a fit note, or review the pattern.
• A size returns both "too small" and "too large". The range may be too wide for that size.
• A product returns for "not as pictured". The problem is the page, not the size.

Compare each product with others in the same category in your own store. A single high number means little without that context.

What can you change?

Each finding points to a different fix. Change one thing at a time so you can tell what helped.

Fix the size chart

If a size is the problem, adjust its ranges. The guide on fixing a size chart after a spike in returns walks through the steps.

Fix the product page

Add a plain fit note such as "runs small, size up if between sizes". Show the model's height and size worn.

Fix the pattern

If one product keeps returning for the same fit reason across sizes, pass the finding to your designer or maker. Describe the problem in measurements, such as "too tight across the upper arm", not "bad fit".

How does size recommendation data fit in?

If you use a body scan, you gain a second signal. You can compare what the scan suggested with what the shopper ordered, and with how the order went.

ZeDaq.ai matches a 30-second body scan to your own size chart, with one chart for each garment type. When you correct the chart, later recommendations use the new ranges. If you want to check how well a tool performs on your products, read how to evaluate a size recommendation tool's accuracy.

Set aside time to look at borderline cases too. Shoppers between two sizes are where chart gaps show up first.

How do you set a realistic goal?

A feedback loop improves fit-related returns, but it will not remove returns entirely. People still change their minds and order extras. The guide to setting a realistic returns target helps you choose a goal based on your own history, not on a figure from another store.

Who should see the findings?

Share a one-page summary each month with whoever decides on sizes, patterns and product pages. In a small brand, that may be one person. In a larger one, it may be the designer, the buyer and the customer service lead.

What does a monthly routine look like?

Keep it short enough to repeat:

1. Export last month's returns.
2. Group fit-related returns by product and size.
3. Pick the one biggest pattern.
4. Make one change and write it down.
5. Check the same numbers next month.

Keep a simple log of changes with dates. After a few months, it becomes the most useful document you have about your own sizing.

For the wider benefits of fit tools, the article on how virtual try-on affects clothing returns looks at what is known, with sources.

FAQs

How often should I review size data?

Monthly works for most stores. Do it more often only if you launch many new products.

Do I need special software?

No. A spreadsheet with product, size, reason and date is enough to start.

How long before I see results?

Wait for at least one full return window after a change, since returns arrive after delivery. Compare the same product across the same period.

What if the data is too thin?

Combine similar products in the same category, or wait longer. Do not change a chart on a handful of returns.