Lesson 2 of 3 · Returns & Refunds for Sellers
Choose Between a Return, a Refund, a Partial Refund or a Replacement
Price every resolution you could actually offer on one unhappy-customer order, then decide — on cost, on what the customer needs, and on what your platform and policy allow.
- 18 min
- Intermediate
Checking this browser…
By the end you will be able to
- Price several resolution paths on the same order so the comparison is like for like
- Recognise that refunds and replacements solve different problems, and that the cheapest option is not automatically the right one
- Write a resolution decision that accounts for cost, customer outcome and the rules that actually apply to the order
How this lesson runs
- Describe the issue and the order
- List the options actually available
- Cost each one on the same basis
- Check the constraints
- Decide and write down why
Why this matters
Resolution decisions get made in a hurry, under mild social pressure, on whichever option feels most reasonable at the time. That is how sellers end up paying return shipping on items worth less than the label, or sending replacements that cost more than a refund would have. Costing the options once, on a real order, gives you a way of thinking that takes about two minutes on every subsequent one — and it will sometimes tell you that the generous option was also the cheap one.
These options are not interchangeable
Before any arithmetic: a refund and a replacement do not solve the same problem, and treating them as two prices for the same outcome will produce bad decisions cheaply. A customer who no longer wants the item wants their money back. A customer who wants the item, but not this damaged one, wants a working item. Offering a refund to the second customer resolves the transaction and loses the sale; offering a replacement to the first one prolongs the problem. Cost is the second question. What the customer actually needs is the first.
What each option is for
| Option | Solves | Costs you |
|---|---|---|
| Full refund with return | Customer doesn't want it; you want the item back | The whole margin, plus return shipping and handling, less whatever the item is worth back |
| Full refund without return | Customer doesn't want it; getting it back isn't worth the shipping | The whole margin and the product, but no return label and no handling |
| Partial refund, customer keeps it | Item is usable but not as expected, and the customer will accept a discount | Only the refunded amount and the fees on it — the sale survives |
| Replacement | Customer wants a working item | A second unit and a second outbound shipment; you keep the revenue |
| Return plus replacement | Customer wants a working item and you need the faulty one back | A second unit and shipment, plus return shipping and handling, less recovery |
What actually moves the answer
A handful of inputs do almost all the work, and knowing which ones lets you make the call quickly on most orders.
Costing the options on the same basis
Comparison only works if every path starts from the same order and counts the same things. Use the final-order-result view from Lesson 1 for each option, then read the differences.
Full refund with return
− fees retained − outbound label − packaging − return label − handling − (product cost − recovered value)
Revenue and refund cancel, so neither appears. Everything spent on the way out still counts, the return adds two new costs, and the item's recovered value comes off the product cost rather than being added as income.
Full refund without a return
− fees retained − outbound label − packaging − product cost
Same as above minus the return label and handling, but the product is a total loss because it never comes back. Cheaper than a return whenever the item's recoverable value is less than the return label plus handling.
Partial refund, customer keeps the item
(revenue − partial refund) − fees retained − outbound label − packaging − product cost
The only path where you keep revenue, which is why it can stay positive. It scales directly with the refund amount, so the question is not whether a partial refund is cheap but how large a one the customer would accept.
Replacement, no return
revenue − fees − outbound label − packaging − product cost − replacement product − replacement label − replacement packaging
You keep the full revenue and the fees stay as charged, but you fulfil the order twice. Whether that beats a refund depends almost entirely on your margin: a replacement costs you one more unit of product and shipping regardless of what the item sold for.
Writing the decision
The output of this lesson is a written decision, not a preference. Writing it down does two things: it forces the constraint check that gets skipped when you decide in a message box, and it gives you something to compare against next time the same issue appears.
What the written decision contains
- The issue, in one line, and what the customer has actually asked for
- The options genuinely available on this order — not every option that exists
- The total business cost of each, worked on the same basis
- Any platform, policy or legal condition that rules an option in or out, noted as a condition you checked rather than assumed
- The customer outcome each option produces
- The option chosen, and the reason — including if the reason was not cost
- One line on what would change the answer if this happens again
Start: One customer issue Paths: Return?; Partial refund?; Replacement?; Full refund?; Return plus replacement? Converge: Cost; Policy and platform conditions; Customer outcome End: Decision, written down with its reason
Paths
- Return?
- Partial refund?
- Replacement?
- Full refund?
- Return plus replacement?
Converge
- Cost
- Policy and platform conditions
- Customer outcome
Order: Item $48.00 + shipping $6.00, product cost $19.00, expected completed-sale profit $22.18 Paths: Label, Result. B — partial refund $16.00, customer keeps item, +$7.25; C — replacement, no return, −$5.32; D — return plus replacement, −$6.52; A — full refund with return, −$29.42 Reversal: At a $40.00 partial refund, path B falls to −$15.15 and becomes worse than both C and D. The crossover against C is $29.46 on these inputs.
Paths
| Label | Result |
|---|---|
| B — partial refund $16.00, customer keeps item | +$7.25 |
| C — replacement, no return | −$5.32 |
| D — return plus replacement | −$6.52 |
| A — full refund with return | −$29.42 |
Worked example
One unhappy customer, four priced options
A customer reports that the item arrived with a cosmetic flaw. It works and is usable, but it is not what they expected. The same product economics as Lesson 1. ILLUSTRATIVE SELLER INPUTS, none of them benchmarks: $48.00 item price, $6.00 shipping charged, $54.00 order revenue, $4.32 of fees on the order made up of a $3.60 variable component and a $0.72 fixed component, $19.00 product cost, $7.40 outbound label, $1.10 packaging, $8.20 return label, $2.50 handling, and a recovered value of $9.50 for the flawed unit if it comes back and is sold as seconds. ILLUSTRATIVE FEE ASSUMPTION: the variable component is credited in proportion to the amount refunded and the fixed component is retained. Check your own order rather than adopting this. POLICY POSITION: all four paths are treated as available here only for the purpose of comparison. Whether they are available to you depends on the platform, your policy and the applicable requirements for that order.
| Line | Value | Note |
|---|---|---|
| Expected completed-sale profit, if there had been no issue | $22.18 | |
| — A. Full refund, item returned — | ||
| Revenue less refund | $0.00 | $54.00 − $54.00 |
| − fees retained | −$0.72 | Variable component credited in full |
| − outbound label and packaging | −$8.50 | |
| − return label and handling | −$10.70 | |
| − product cost net of recovery | −$9.50 | $19.00 cost less $9.50 recovered as seconds |
| Final result | −$29.42 | $51.60 worse than the completed sale |
| — B. Partial refund of $16.00, customer keeps the item — | ||
| Revenue retained | $38.00 | $54.00 − $16.00 |
| − fees retained | −$3.25 | $1.07 of the variable component credited |
| − outbound label and packaging | −$8.50 | |
| − product cost | −$19.00 | |
| Final result | +$7.25 | $14.93 worse than the completed sale |
| — C. Replacement sent, no return — | ||
| Revenue retained | $54.00 | No refund, so no fee credit either |
| − fees | −$4.32 | |
| − original label, packaging and product | −$27.50 | |
| − replacement product, label and packaging | −$27.50 | |
| Final result | −$5.32 | $27.50 worse than the completed sale |
| — D. Return plus replacement — | ||
| As C, plus return label and handling | −$10.70 | |
| Plus recovered value of the returned unit | +$9.50 | |
| Final result | −$6.52 | $28.70 worse than the completed sale |
| — The reversal — | ||
| B at a $40.00 partial refund instead of $16.00 | −$15.15 | Now worse than both C and D |
| Partial refund at which B and C are equal | $29.46 | On these inputs only |
On these numbers the ranking is B (+$7.25), then C (−$5.32), then D (−$6.52), then A (−$29.42) — but B only holds its position while the refund that satisfies the customer stays below $29.46.
Three findings, and the third is the important one. The gap between best and worst is $36.67 on a single order that sold for $54.00. Resolution choice is not a rounding error; on a low-margin item it can be worth more than the sale. The full refund with return is worst here for a specific reason: the item comes back worth $9.50 while recovering it costs $10.70 in return shipping and handling-time value. The economic recovery cost exceeds the recovered value. Change the recovery to $17.00 and that reasoning inverts — which is why the return-or-not question is really a question about the item, not about policy. And the reversal is the reason this lesson refuses to recommend anything. A partial refund is the best option here at $16.00 and the worst of the four at $40.00, with the crossover against the replacement path at $29.46. The number that decides it is what the customer would actually accept, which is not something arithmetic can tell you. What arithmetic can tell you is when to stop negotiating upward — and on this order that point is $29.46, at which sending a replacement costs the same.
Assumptions: Every figure is an illustrative seller input, not a benchmark, an average or a recommendation. The proportional credit of a variable fee component and the retention of a fixed component is a modelling assumption made to show the mechanics, not a statement about any platform's rules. Recovered values are estimates the seller makes, not realised sales. No return window, refund deadline, restocking-fee rule, shipping obligation or consumer-law requirement is stated or implied; whether any of these four paths is available on a given order depends on the platform, the seller's own policy, the jurisdiction and the circumstances, all of which must be checked separately.
Illustrative figures for teaching only — not a benchmark, average or guarantee.
Do it with your own numbers
Price two resolutions for one real order
Before you start, have ready
- One real customer issue, with the original order figures in front of you
- Your product cost and what a replacement unit would cost you today
- Outbound label and packaging cost, and a real quote for the return label if you would pay it
- An honest estimate of what the item would be worth back, given the actual issue
- What the platform credited or would credit on this order, and what it would retain
Open the Refund vs Replacement Calculator
You will produce: A cost comparison between refunding and replacing on your own order, which you can extend by hand into the partial-refund and no-return paths.
Then ask yourself: Check what the tool does with the original outbound shipping. If the replacement side counts a second label but the refund side ignores the first one, the comparison is tilted and refunds will look better than they are. Then check whether it lets you enter a recovered value for a returned item, and whether it treats that value as reducing product cost rather than as income. If it can only model a total loss, it will make asking for the item back look pointless even when it isn't. Finally, extend it yourself. Work out the partial refund at which the two options cost the same. That figure is the most useful thing this lesson produces, because it tells you the point at which negotiating a larger discount stops being the cheaper path.
Common mistakes
Choosing the resolution before pricing it
The options differ by more than most people expect — $36.67 on the worked example's $54.00 order — and the intuitive choice is often not the cheap one.
Instead: Cost the paths that are actually available, on the same basis, before replying.
Assuming a partial refund is the cheapest option
It scales directly with the amount refunded. It is only cheap while the amount that satisfies the customer stays small, and it becomes the most expensive option above a crossover you can calculate.
Instead: Work out the refund amount at which the partial refund matches your next-best option, and treat that as the ceiling.
Assuming a replacement is cheaper than a refund
A replacement is a second full fulfilment — another unit, another label, another box. On a thin-margin item it can cost more than refunding the whole order.
Instead: Price the replacement as product plus shipping plus packaging, and compare it against the refund path on the same order.
Asking for the item back out of principle
If the return label plus your handling-time value comes to more than the item is worth to you when it arrives, the economic recovery cost exceeds the recovered value — and part of that cost is your time rather than cash, which makes it easy to miss.
Instead: Compare return shipping plus your handling-time value against your honest recovered value, and let that decide — subject to whatever the platform and your policy require.
Deciding on cost alone
A cheaper resolution that leaves the customer with an unusable item hasn't resolved anything, and it may not be permitted by the platform, your published policy or the applicable requirements.
Instead: Treat cost as one of three inputs alongside customer outcome and the conditions that apply to that order.
Counting the lost sale as well as the refund
On a full-refund path the revenue and the refund already cancel. Subtracting the sale again double-counts it and makes every refund path look far worse than it is.
Instead: Let them net, and capture the missing profit once by comparing against the completed-sale figure.
Reusing last time's decision without rechecking the inputs
The ordering depends on recovered value, replacement cost and return shipping, all of which move by product and over time.
Instead: Keep the written decision and the numbers behind it, so next time you are rechecking inputs rather than starting over.
Illustrative Beginner Scenario
Marcus, defaulting to the return
- Situation
- Marcus handled every complaint the same way: ask for the item back, refund on receipt. It felt fair and consistent, and it meant he never had to decide anything.
- What went wrong
- The default was expensive on exactly the orders where it was least justified. On his lower-priced items, the return label plus the time to inspect and repack cost more than the item was worth once it arrived — and much of what came back could only be sold as seconds anyway. He was paying to recover things he could not really recover.
- What changed
- He costed four paths on one recent complaint and found a spread of more than thirty dollars on a fifty-four dollar order. Then he worked out the one number that generalised: the point at which a partial refund stopped being cheaper than sending a replacement. He started checking recovered value against return shipping before deciding, and writing the decision down with its reason.
- Result
- He now asks for the item back when it is worth getting back and resolves it another way when it isn't, within what his platform and his own policy allow. Some customers get a partial refund, some get a replacement, and a few still get a full refund with a return. What changed is that each choice has a reason attached to it, and he can tell you what the alternative would have cost.
- Lesson
- A consistent default is not the same as a good decision. The right resolution depends on what the item is worth back, what a replacement costs and what the customer would accept — and those change from order to order.
This is a composite teaching example, not a guaranteed result.
Try it yourself
Write one resolution decision
- Take one real customer issue, open or recent, and write the problem in a single line along with what the customer has actually asked for.
- List the resolutions genuinely available on this order, and note any that are ruled out by the platform, your own policy or applicable requirements — check rather than assume.
- For each available option, work out the final business result using the Lesson 1 method: what you keep, minus every cost, plus any recovered value.
- Cost the full refund with return, including the return label and your handling time, and net the recovered value against product cost rather than adding it as income.
- Cost the same refund without a return, and see whether skipping the return saves more than the item is worth back.
- Cost a partial refund at the amount you think the customer would accept, and then find the amount at which it stops being cheaper than your next-best option.
- Cost a replacement as a second full fulfilment — product, label and packaging — and add a return leg if you would need the original back.
- Check each option against what the customer actually needs, not just its price.
- Write down the option you chose, its cost, the alternatives you rejected with theirs, and one line on what would change the answer next time.
You end up with: One written resolution decision with every available option priced on the same basis, the constraints you checked, and the reason for the choice.
Confidence check
Before you reply to the customer, check you can answer these:
- Can you say what the customer actually needs, as distinct from what would cost you least?
- Have you priced every option that is genuinely available on this order?
- Do you know what the returned item would be worth to you, and how that compares with the return label plus your handling time?
- Do you know the partial refund amount at which your next-best option becomes cheaper?
- Have you checked which options the platform and your own policy actually allow here, rather than assuming?
- Can you state the reason for your choice — including if the reason wasn't cost?
Your next small step · 15 minutes
Take one open or recent customer issue, price every resolution available on it, and write the decision down with its reason and the cost of the option you rejected.
The first one takes fifteen minutes. Every one after that takes two, because you will already know which inputs move the answer on your products.
Related on SimpleLifeCalc
Guides
Calculators
Free resources
Hubs
Questions people ask
Is a partial refund cheaper than a full refund?
Sometimes, and it depends entirely on how large the partial refund is. It is the only resolution where you keep part of the revenue, so a small one can leave the order profitable. But it scales directly with the amount, so a large one can cost more than sending a replacement. On this lesson's worked example a $16.00 partial refund leaves the order at +$7.25 while a $40.00 partial refund leaves it at −$15.15, worse than every other option compared. The useful move is to calculate the amount at which it stops being your cheapest path — $29.46 on that order — and treat it as a ceiling.
Should I ask the customer to send the item back?
Compare what the item would be worth to you when it arrives against the return label plus the value of your inspection and repacking time. If the item comes back sellable at more than those costs, recovering it is worth doing; if it comes back worth less, the economic recovery cost exceeds the recovered value. Note that only part of that cost is cash — the label is money, the handling may just be your time. That is the economic answer. Whether you can decline a return, or must accept one, depends on the platform, your published policy and the applicable requirements for that order, which you should check separately.
Is sending a replacement cheaper than refunding?
Not automatically, because a replacement is a second complete fulfilment: another unit of product, another shipping label and more packaging. You keep the revenue, which helps, but on a thin-margin item the cost of fulfilling twice can exceed what a refund would have cost. On the worked example the replacement path lands at −$5.32 against −$29.42 for a full refund with return, but that ordering comes from those particular numbers and changes with product cost, shipping and what the returned item would be worth.
How do I compare the options fairly?
Start every path from the same order and count the same things: what you keep after any refund, minus the fees actually retained, minus the outbound shipping and packaging you already spent, minus any return shipping and handling, minus product cost, plus any recovered inventory value. Two traps to avoid: don't subtract the lost sale on top of the refund, because the revenue and the refund already cancel; and don't add recovered value as income while also reducing product cost, because that counts it twice.
Should I always pick the cheapest resolution?
No. Cost is one of three inputs. The resolution also has to solve the customer's actual problem — a refund and a replacement answer different needs — and it has to be permitted by the platform, your own policy and any applicable legal requirements, which vary by platform, jurisdiction, product and circumstance. Knowing the cost of each option means you can choose to absorb more than the minimum deliberately, which is a different thing from doing it by accident.