Writing a Refund Policy That Is Fair And Defensible

Refund policy gets written once, by whoever is free, usually the week before launch. It then sits untouched while the brand scales, and it becomes the document that decides margin leakage, support load, repeat rate and the outcome of every payment dispute you will ever face. It deserves more thought than it gets.

The mistake is treating it as legal boilerplate. It is a commercial instrument. It sets the price of trust.

The tradeoff nobody prices properly

Refund friction is a lever with two ends, and pulling either end too hard costs money.

Tighten it and your immediate refund outflow drops. What you do not see on the same dashboard is the second order cost: lower first order conversion because the guarantee is weak, lower repeat rate because a bad first experience ended badly, worse reviews, more support escalations, and more customers who skip your support desk and go straight to their bank. That last group is expensive twice over, because a chargeback costs you the order plus a fee and a mark against a ratio your acquirer watches.

Loosen it too far and you invite the opposite problem: refund abuse, wardrobing, and a return rate that quietly eats your contribution margin.

The right setting is category specific. A supplement brand with a consumable product and high repeat frequency should be materially more generous than a furniture brand with a fifteen thousand rupee ticket and a reverse logistics cost that can approach a fifth of order value. Do not copy your policy from a brand in a different category.

Measure it properly. Track refund rate, return rate, support tickets per hundred orders, and ninety day repeat rate together in one view. When you change the policy, watch all four, because a policy tightening that improves refund rate while damaging repeat rate is a loss disguised as a win.

Refund, replacement or store credit

These are not interchangeable, and offering only one is a design failure.

Replacement is right when the product was damaged, defective, or the wrong item was shipped, and the customer still wants what they ordered. It preserves revenue, preserves the relationship, and it is usually cheaper than a refund because you keep the sale. Make replacement the default first offer for fault based cases, and make it fast. Speed is what converts an angry customer into a repeat one.

Store credit is right when the customer changed their mind, the product was fine, and you want to retain the revenue. Credit is genuinely useful when it is generous, easy to spend, and clearly explained. It becomes a trust problem when it is forced on a customer who is entitled to money back, or when it carries a short expiry. Offer it, sweeten it slightly, and never disguise it as the only option.

A cash refund is right when the product was defective and the customer has lost confidence, when the wait for a replacement is unreasonable, or simply when the customer asks for one and your policy entitles them to it. Fighting a customer who is entitled to a refund is the most expensive form of saving money available to a D2C brand.

Write the decision tree down. Support executives should not be improvising this at nine at night during a festive sale.

Prepaid and COD refunds are different machines

Teams assume both flows behave the same. They do not.

For prepaid orders, a refund is a reversal through the original payment method, initiated by you at the gateway. Once you initiate, the money travels through the gateway, the card network or UPI rails, and the customer’s bank before it lands. Each hop adds time and none of it is under your control after initiation. The total time varies by payment method, by gateway and by the customer’s bank, and it is routinely longer than customers expect. So set the expectation explicitly, and give the customer the refund reference number the moment you initiate. Most refund complaints are not about money, they are about silence.

For COD orders, there is no original instrument to reverse. You need bank details, which means a data collection step, an identity risk, and a verification burden. Build a secure form for it rather than collecting account numbers over WhatsApp. Expect a higher failure rate from wrong IFSC codes and mismatched names, and build a retry loop with the customer rather than letting failed transfers sit in a spreadsheet. Many brands offer store credit as the default for COD refunds precisely because it removes this friction, which is fine as an offer and unacceptable as the only option.

One more COD note: refused deliveries and RTO are not refunds at all, because no money was collected. Keep them out of your refund reporting or your metrics will lie to you.

Partial refunds and restocking

Partial refunds are the underused middle path. A customer received four items, one was damaged, and refunding a quarter of the order plus shipping resolves it without a reverse pickup, without a restocking decision, and without losing the rest of the sale. For low value items, a partial refund with no return requested is frequently cheaper than the reverse logistics leg.

Restocking deductions are legitimate in some categories, particularly for large items and made to order goods, but they must be disclosed clearly before checkout, applied consistently, and stated as a specific amount or percentage. A restocking fee that appears for the first time in a refund email reads as a penalty and generates disputes.

Decide your disposition rule at the same time. If a returned unit cannot be resold at full price, the policy should route it to a defined outcome, whether that is a discounted channel, a repair loop or a write off. An undocumented disposition rule is how returned stock disappears.

Your policy is your strongest dispute asset

Here is the operational point that most brands miss. When a payment dispute reaches your bank, the single most useful document you can produce is your own refund policy as it appeared on the date of purchase, alongside the checkout screen that showed it.

That means three habits. Publish the policy on a public, indexable page, not buried in a PDF or a chat widget. Make the customer see it at checkout, not only in the footer. And archive dated versions every time you change it, so that eighteen months later you can prove what the customer agreed to rather than what you have written since.

A written, timestamped, publicly visible policy converts a he said she said argument into a documentary one, and documentary arguments are the ones merchants win.

On statutory obligations, be careful

India has consumer protection law covering ecommerce, and it imposes real obligations on sellers and marketplaces around disclosure, returns and refunds. Those rules have been amended more than once, guidance evolves, and marketplace contract terms often sit on top of them with their own stricter requirements.

So do not lift a fixed number of days from a blog post, including this one, and encode it into your policy. Confirm the current statutory position with a qualified advisor, confirm the contractual position with each marketplace you sell on, and then set your published policy at or better than the stricter of the two. Review it annually and after any regulatory change. A policy that quotes a superseded rule is worse than one that quotes none.