The problem with reaching for the rulebook first
Refunds climb. Most founders grab the rulebook. Cut the return window from 30 days to 14. Add a restocking fee. The logic feels right. Make returning harder. Fewer people return.
Here is what that logic misses. In supplements, skincare, and fitness, most refunds are not buyer’s remorse. They are broken expectations. The customer used the product wrong. Or gave up too early. Or never knew what week one should feel like. A shorter window cannot teach anyone to take a vitamin daily. And the friction hits everyone. Good buyers walk away at checkout. Fees come back as one-star reviews.
Every refund you fail to stop costs more than it looks. A $100 order comes back. You do not just hand back $100. You pay about $5 to ship it back. And $2 to shelve it again. And the $20 of ads that won that customer? Gone for good. One refund drains about $127. The Returns line on your shows only the first $100.
This article is about the second one: refunds. A rule change fights a refund after the customer gave up. Teaching stops the refund before it starts, without scaring buyers away at checkout.
1. An example showing you the numbers
You sell supplements on Shopify. Your best seller is a 90-day multivitamin bundle at $100. You do 4,000 orders a month. Your refund rate is 7%. The norm for supplements is 4%. The tell the story. “Did not see results.” “Did not work for me.” “Forgot to take it daily.” Not one is a policy problem.
So you run a fair test. One month of orders. Two of 2,000. Group one gets the tough fix. The return window drops from 30 days to 14. Plus a $5 restocking fee. Group two gets the friendly fix. Five short messages over the first 21 days. How to take the product. What week one feels like. When results usually show. No rule change at all.
At this store’s rates, each return takes about $107 in cash. Each lost order takes $8 of profit. Here is how the two groups ended the month.
Two fixes, one month, 2,000 orders each, only one made money
Cash saved counts the $107 a return takes in cash; the $20 of ads is spent either way.
| What happened | Tighter policy | Education series |
|---|---|---|
| Refund rate before | 7% | 7% |
| Refund rate after | 6.5% | 3.5% |
| Refunds avoided (per 2,000 orders) | 10 | 70 |
| Cash saved on refunds (at $107 each) | +$1,070 | +$7,490 |
| Orders lost at checkout | 160 | 0 |
| Profit lost on those orders (at $8 each) | -$1,280 | $0 |
| Cost to run the fix | $0 | -$800 |
| Net change for the month | -$210 | +$6,690 |
Read the bottom row. The tough fix shaved half a point off the refund rate. It still lost money. The new friction scared off 160 good orders. The friendly fix cut the refund rate in half. Sales held. Review scores rose. The series cost about $800 to send. It returned $7,490.
Per order, it is just as clear. At 7%, refunds cost about $7 of every $100 order. The store kept $8 per order after all six costs. At 3.5%, the refund cost drops to about $4. The store keeps $11. Rolled out to all 4,000 orders a month, that is about $13,400 a month. Roughly $160,000 a year.
One honest note. The drop from 7% to 3.5% is what a well-built series did here. In that category, teaching truly changes the result. Your drop may be smaller. But even a one-point drop saves 20 returns per 2,000 orders. That is $2,140 against $800 of cost.
The sentence that changes how you think about refund policy
Most refunds are not customers gaming your policy. They are customers who never learned how to succeed with your product.
No return window can teach someone to take a vitamin every day for three weeks. Here, the tough fix lost $210. The friendly fix made $6,690. Same 2,000 orders.
2. How to build a 21-day education series
This is writing work, not software work. Plan an afternoon to read your refund data. Then a week to write and load five short messages.
- Read 90 days of refund reason codes first. Every platform collects them. Look for “did not work,” “no results,” or “used it wrong.” More than 30% of refunds? You have an education problem. If “damaged” or “wrong size” lead, this fix is not for you.
- Map your customer’s first 21 days. What does week one feel like? When do results usually show? What mistake do most new customers make early? Write the answers down first. The series is these answers, in order.
- Write three to five short messages. Day 0: a plain how-to-use guide. Day 3: what week one feels like. Day 7: early signs it is working. Day 14: a routine check-in. Day 21: a usage reset and a gentle reorder note.
- Put the same answers on the product page. Buyers who read them sort themselves. A few will not buy. It was not right for them. Good. Those were tomorrow’s refunds.
- Measure three numbers together: refunds, conversion, reviews. A good series moves all three the right way. Conversion dips? Reviews go flat? The messages are too sales-y. Rewrite them from the customer’s side.
3. One warning before you act
Teaching only fixes broken-expectation refunds. Do your reason codes say “broken,” “leaked,” or “wrong item”? No email will help. Take that up with your maker or your warehouse’s inspection step. And never dress a sales pitch up as teaching. Customers can tell. A pushy series kills the trust this needs.
Run the fair test before any store-wide rollout. On a small store the numbers will wiggle. Give it a full month. Or ask your accountant to help read the result.
4. Frequently asked questions
My refund rate is already at my category’s norm. Is this still worth doing?
Usually, yes. The norm is a floor, not a finish line. The series costs cents per customer. It lifts reviews and repeat buys even when the drop is small. If the test shows nothing, your pages already teach.
Does this work on Amazon?
Yes, with different tools. A+ Content on your listing teaches before the purchase. Brand Tailored Promotions lets Brand Registry sellers message past buyers. But Amazon owns the customer relationship. This fix pays most on your own store, where you own the email list.
Can I still tighten my policy later?
Yes, in that order. Build the series first. Let the refund rate come down. Most stores then find they never need stricter rules. A generous policy plus a good series usually beats a tight policy alone.
Should the messages go by email or text?
Email does the heavy lifting. It is cheap and carries a real guide. Short Message Service (SMS) texts are the nudge layer: the day-0 “your guide is in your inbox” and the day-21 reorder reminder.
5. Quick reference: what to avoid and apply
What to avoid
- Grabbing the policy lever first, it trades a small refund saving for a bigger sales loss.
- Adding restocking fees to scare people off, they come back as one-star reviews.
- Writing “education” that is really a sales pitch, customers can tell, and it changes nothing.
- Testing on one group with no control, you will never know what caused the change.
- Judging the test on refund rate alone, conversion and reviews are half the story.
What you should do
- Pull 90 days of refund reason codes before you change anything.
- Sort your refunds: broken expectations are teachable; defects and sizing are not.
- Write three to five short messages across the customer’s first 21 days.
- Put the same answers on the product page so buyers sort themselves.
- Track refunds, conversion, and review scores together on matched groups.
Want your six costs mapped for you?
Our free 31-Point Profit Diagnostic puts a dollar figure on where your profit is leaking. Or get a quote for done-for-you eCommerce accounting.
Definitions, modeling notes & rate-basis disclosures
Definitions
- The six profit levers
- (1) Discounts, (2) Refunds, (3) Product cost (landed COGS), (4) Sales channel and payment fees, (5) Shipping and 3PL fees, (6) Advertising spend.
- Refund event cost
- Everything one return costs: the $100 given back, about $5 return shipping, about $2 restocking, plus the $20 of ads already spent on the order. About $127 on a $100 order.
- Refund reason codes
- The reason a customer picks when asking for a refund. Your platform collects them. Most founders never read them.
- Post-purchase education series
- Three to five messages over the first 21 days. They teach correct use and set honest expectations.
- Matched groups
- Two similar sets of customers that differ in one thing only: the fix being tested. The only fair way to know what caused a change.
- Contribution per order
- Selling price minus all six costs. The money one sale really leaves behind.
- Profit and Loss (P&L) statement
- The report of your revenue and costs. Its Returns line shows refunded revenue only. Not the return shipping, restocking, or lost ads.
Modeling notes
- This article uses the series’ standard teaching store: a $100 average order, product cost $40, standing discount 10%, advertising $20 per order, shipping and 3PL $12, payment and channel fees 3%. It changes only the refund lever: this store starts at a 7% refund rate instead of the standard 5%.
- Refund cost per order: each return costs about $107 in cash ($100 back + $5 return shipping + $2 restock). At 7 returns per 100 orders, that spreads to about $7 per order. At 3.5 per 100, about $4. The $20 of lost ads is already counted in the advertising line.
- Contribution per order: $100 − $40 − $10 − $20 − $12 − $3 − $7 = $8 before the fix. The refund line falls to about $4 after, leaving $11.
- Group math: the policy group saves 10 × $107 = $1,070 but loses 160 × $8 = $1,280. Net −$210. The education group saves 70 × $107 = $7,490 and costs $800. Net +$6,690. Rolled out to 4,000 orders a month: 140 × $107 − $1,600 = $13,380 a month, about $160,000 a year.
Rate-basis disclosures
- Category baseline: about 4% refund rate for supplements. This store starts at 7%.
- Return shipping: $5 per returned order. Restocking: $2 per returned unit.
- Education delivery: about 40 cents per customer over 21 days ($800 per 2,000 orders), including email platform fees and a small text budget.
- Policy-test effects: a 30-to-14-day window plus a restocking fee cut refunds 0.5 points and cut conversion 8%, typical of observed tests.
- Store volume: 4,000 orders per month at $100 average order value. All figures rounded to whole dollars.