Workflow · ecommerce · monthly

Automate the monthly returns analysis

A returns analysis tells you how much of what you sell comes back, which styles and sizes drive it, and why. For apparel and other fit-driven brands it is one of the biggest margin levers there is. Most teams pull refunds from the store platform once a month and get the wrong number, because refunds miss exchanges and store credit. A model trained on your returns system and your definitions gets it right every month and says what to do about it.

By Jim Wilson · Updated October 14, 2026

Book a 30-minute audit call

Where the five hours go

StepBeforeAfterBack per year
Export returns and ordersReturns system and store platform1 h012 h
Match returns to the original saleStyle, size, channel, month sold2 h024 h
Group the reasonsFit, quality, changed mind1 h012 h
Write the readWhat changed and what to do1 h20 min8 h

Sample figures to show the format. Five hours down to twenty minutes of review is 56 hours a year.

Same question, two answers

"How did returns do last month?" Illustrative example
Out-of-the-box AI

Refunds were $48,210 in September, 6.2% of gross sales, up from 5.8% in August. Consider reviewing your return policy.

Pulled refunds from the store platform, which misses exchanges and store credit. Month over month on a seasonal brand. Advice nobody asked for.

Trained on your business

Return rate was 14.1% of units on a trailing twelve-month basis, from the returns system, not store refunds. That's flat to last year. September's spike is size exchanges on two new fits, not quality. Worth a look at the size chart before the spring buy.

Your source of truth, your definition, your seasonality, and the judgment call your best analyst would make.

What the model has to learn

  • Your definition of a return. Units or dollars, whether exchanges count, and whether the rate is measured against the month sold or the month returned.
  • Your source of truth. The returns system, not the store's refund field.
  • Your normal. The seasonal pattern, the styles that always run high, and the threshold that makes a change worth a sentence.
  • Who acts on it. Fit issues go to product, quality to sourcing, policy abuse to customer service, each in words that team uses.

The returns read often comes alongside the month-end close pack, since returns are one of the biggest lines between gross and net revenue. It is a common build during an operations audit.

Automate the weekly business recapHow consumer brands turn a four-hour weekly KPI recap into ten minutes of review, with a model that knows your definitions, sources and seasonality.Automate the month-end close packHow consumer brands turn a six-hour monthly close pack into thirty minutes of review, with a model that knows your definitions and sources of truth.The slow-moving inventory report, by SKUBuild a slow-moving inventory report by SKU from receipt dates and trailing sales once, then refresh it monthly for your team, your buyers and your lender.

Questions

Why not use the refunds number from Shopify?

Refunds only count money sent back. Exchanges and store credit, which are often most of an apparel brand's returns, do not show up there. The returns system, such as Loop, Happy Returns or Narvar, has the full picture.

Should return rate be measured in units or dollars?

Both answer different questions. Units show fit and quality problems; dollars show the margin hit. The model reports the one your team uses, and the other when it tells a different story.

Why trailing twelve months?

Returns lag sales by weeks and swing with the season. Month over month on a seasonal brand mostly measures the calendar. Trailing twelve months shows the real trend, with the month called out when something actually changed.

What does it do with return reasons?

It groups the free-text and coded reasons into a few causes your team cares about, like fit, quality and changed mind, and ties them to the style and size so the fix is obvious.

Thirty minutes. Bring one question your team answers every week.

I'll show you how generic AI answers it, how your best analyst would, and what it takes to close the gap.

jim@trimtabops.com