The return you could have prevented

A customer orders a jacket, waits four days for it, and sends it back the following week. The store paid to win that sale, paid to ship it, and now pays to bring it home and put it back on the shelf.

Some of those orders were never going to stay sold. A gift, a change of mind, two sizes ordered on purpose. Others went back because the customer could not tell, from the page alone, whether the product was right for them.

The National Retail Federation, with Happy Returns, put returns at 15.8% of retail sales for 2025, and at 19.3% of online sales. Those are industry averages across every category and every reason. They do not tell you which of your own returns could have been avoided.

Separate the returns you caused from the ones you did not

Sort a month of returns into two piles. One holds the orders that came back for reasons outside your control. The other holds the orders where something the customer needed to know was not on the page.

The second pile is the one you can act on. A dozen returns in a single product line, all citing fit, is a product page problem rather than a customer problem.

Work from your own orders rather than the industry rate. A store selling made-to-order furniture and a store selling t-shirts have little in common here.

The reason code is not the reason

Return forms offer a short list: wrong size, not as described, changed my mind. Customers pick the closest option and move on. The code tells you where to look, not what happened.

"Not as described" can mean the photograph was flattering, the measurements were missing, or the material behaved differently than the page suggested. Each of those needs a different fix.

Read the notes customers leave, and ask your support team what they hear on the phone. The detail that explains a return usually arrives in a sentence, not a dropdown.

Answer the question before the order

Much of what prevents a return is information that arrived too late: the measurement, the compatibility check, the honest comparison between two models.

A customer who is unsure often buys anyway, planning to decide at home. That order counts as a sale in Monday's report and as a return three weeks later. The behaviour behind it is covered in the cost of a question nobody answers.

Live shopping moves that conversation to before the checkout. An advisor who can hold the product up, measure it and say plainly that the smaller model suits this customer better has kept both the sale and the customer.

Count the return against the sale

A report that counts orders without subtracting returns will flatter a product line that costs you money. Look at revenue net of returns by product, not gross.

Include the handling in that cost: the return shipping, the inspection, the restocking, and the item you cannot sell at full price again.

Then check whether the products with the highest return rates are the ones that need the most explaining. Where they are, the answer is usually a better page or a conversation rather than a lower price.