What does returns analytics show?
For any range, the return rate by product, style, category and channel; the profit lost after the refund and the handling cost; the reasons customers gave; the exchanges that followed; and which sizes are responsible where a size was recorded. Everything is joined to the sale, so a return is always attached to the order it came from.
The list is ranked by money. A style with a modest rate on a high-margin, high-volume line can cost more than a style with a dramatic rate on a cheap one, and ranking by profit lost puts the expensive problem first.
- Return rate by product, style, category and channel, for any range.
- Profit lost after the refund, unrecovered shipping, handling and any value lost on resale.
- The reasons customers gave, and the exchanges that followed them.
- Sizing defects per style, from the reasons and exchange directions the returns platform recorded.
- The page fix each product implies: a measurement table, a fit note, better images or a description change.
Why profit lost is bigger than the refund
A return costs the refund, the outbound shipping that is not recovered, the return shipping where the brand pays it, the handling to inspect and restock, and the value lost where the unit cannot be sold as new. The engine computes the total from the records it holds and the rules the brand sets, rather than counting the refund alone.
That is why a product can be profitable on the sale and unprofitable after returns, and why the product analyst's money-leaks list and this surface agree: both are using the same deductions.
How are sizing defects found from your own evidence?
Where the returns platform records a reason and a size, the engine can see that a style is returned for being too small far more often than the range's norm, and that exchanges for it move up a size. That is a sizing defect, and it is found from the brand's own returns rather than from a general rule about apparel.
Fit intelligence that generalises this across the catalogue, such as a label that runs one size small across many styles, is on the roadmap. What ships today is the per-style evidence and the exchange pattern behind it.
The page fix each product needs
Returns cluster by cause, and each cause has a page-level fix. Too small and exchanged up: a fit note and a measurement table. Not as pictured: better images or a colour-accurate shot. Not as described: a description correction. Quality: a supplier conversation rather than a page change. The engine names the fix implied by the reasons on that product.
The fix is a proposal, not an edit. Where the brand publishes through the platform, the change is drafted and goes through Review & Send; where it does not, the proposal is a task for the person who owns the storefront.
What returns analytics cannot tell you
It cannot tell you why a customer who gave no reason returned an item; it reports unreasoned returns as unreasoned rather than guessing. It cannot see returns that never reached the returns platform. And it does not publish an industry return rate to compare against, because the engine reports the brand's own numbers only.
Where a returns integration is not connected, the surface says so instead of showing a rate of zero, which would read as a brand with no returns rather than a brand with no data.
Which roles work this surface on ftes.ai?
The Returns Officer FTE owns return rate and the cost of returns, and is judged on profit recovered from returns reduction. The Fit Officer FTE owns sizing defects and the page fixes that address them. The Buyer FTE reads both before reordering a style that comes back.
A brand connecting a returns platform sees the first full picture within a sync. The fastest first win is usually the top three styles by profit lost, each with its page fix drafted and waiting for approval.