How to Test a Return Fee Without Guessing: A Four-Week Experiment Design

Industry roundups now report that a majority of retailers charge for at least some mail-in returns, while surveys find a large share of shoppers say a fee makes them less likely to buy. Both can be true for your store, or neither. The only way to know is to test.

Start with a hypothesis you can fail

"A $7 return fee will recover label cost without reducing conversion by more than a set amount" is a testable statement. "Fees are good" is not. Write the threshold before you start, so the result decides the question and not your mood.

Pick three variants, not two

• A. Current policy. The control.
• B. Flat fee on refund returns. The fee comes out of the refund.
• C. Fee on refunds, waived for exchanges or store credit. The version that steers customers toward keeping the revenue.

Testing C alongside B tells you whether the fee itself or the way you structure it drives the result.

Choose metrics that capture net margin

1. Conversion rate on product and checkout pages.
2. Return rate as a share of delivered orders.
3. Exchange and store credit share of returns.
4. Net margin per order after return costs and fee revenue.
5. Support contacts about the return policy.

A fee that cuts the return rate but also cuts conversion can lose money, and you will only see it if you look at net margin per order.

Plan for the lag

Returns arrive days or weeks after the order. A four-week test window needs a follow-up period long enough for most returns from the final orders to come in, or the variants that return slowly will look better than they are. Match the follow-up to your return window.

Keep the test clean

• Split traffic by a stable rule, so a shopper always sees one policy.
• Do not run a sale, a shipping change or a new theme during the test.
• Avoid peak weeks, which distort both conversion and return behavior.
• Record the policy text each variant displayed.

Shopify's built-in tools do not split a return policy by variant, so many merchants run the test by market, by customer segment or in sequential time blocks. Each approach has trade-offs, and sequential blocks are the weakest, so use them only if you control for seasonality.

Read the result honestly

If your store sees a few hundred orders in the test window, differences in conversion may be noise. Treat small samples as directional. Decide in advance what would convince you, and what you will do if the result is unclear. Often the right answer is to extend the test.

Check fairness before rollout

If the fee wins, look at who pays it. High-value customers and repeat buyers may deserve a waiver. A fee that protects margin by annoying your best customers is a poor trade.

ShipAid Returns & Exchanges gives merchants discounted labels with no monthly software fee and merchant-controlled fees, so you can run this kind of test and keep the margin it creates. Learn more at shipaid.com.