Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Write a clear, fair returns, refunds & exchanges policy for an online store. Use when asked to write a return policy, refund/exchange policy, or store returns page. Produces a customer-friendly policy — window, conditions, process, refund method/timing, exceptions, and shipping — in plain language that reduces support tickets and builds trust. Not legal advice.
.claude/skills/mohitagw15856-return-refund-policy/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 5% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 31% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 54% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 46% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 21% | 0% |
A clear returns policy is a conversion tool, not just fine print — shoppers check it before buying, and a fair, plain-English one removes a purchase objection. This skill writes a policy that's easy to understand and easy to act on, so customers trust it and your support team isn't answering the same questions all day.
> Note: this is a drafting aid, not legal advice. Consumer-protection rules (statutory return rights, > distance-selling/cooling-off, warranty law) vary by country and platform — have it reviewed against your > jurisdiction and marketplace policies before publishing. Flag, don't rule on, legal questions.
Given "write a returns policy for my Shopify store", produce the full policy anyway — infer sensible, customer-friendly defaults (e.g. a 30-day window), and clearly mark each business-specific choice (set your value) so the owner confirms window, who pays return shipping, and exceptions. Never withhold for missing detail.
Ask for these only if they aren't already provided (else use a labelled default):
Mark each business-specific value (set your value) and add a note to confirm jurisdiction-specific rights.
E-commerce trust & CRO practice — transparent, plain-language returns policies that reduce purchase friction and support load.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 21,135 | 16,328 | -23% | 1 | 1 | 0% | 2,835 | 2,974 | +5% | 0 | 0 | — |
case-02 | fail→pass | 16,952 | 16,908 | -0% | 1 | 1 | 0% | 1,882 | 2,461 | +31% | 0 | 0 | — |
case-03 | fail→pass | 11,084 | 15,885 | +43% | 1 | 1 | 0% | 1,762 | 2,714 | +54% | 0 | 0 | — |
case-04 | fail→pass | 19,456 | 18,541 | -5% | 1 | 1 | 0% | 2,086 | 3,036 | +46% | 0 | 0 | — |
case-05 | fail→pass | 18,721 | 19,581 | +5% | 1 | 1 | 0% | 2,318 | 2,815 | +21% | 0 | 0 | — |
case-06 | pass→pass | 21,102 | 15,616 | -26% | 1 | 1 | 0% | 2,166 | 2,670 | +23% | 0 | 0 | — |
case-07 | fail→fail | 22,146 | 17,722 | -20% | 1 | 1 | 0% | 2,282 | 3,000 | +31% | 0 | 0 | — |
case-08 | fail→pass | 17,208 | 18,764 | +9% | 1 | 1 | 0% | 1,940 | 3,213 | +66% | 0 | 0 | — |
case-09 | pass→pass | 20,918 | 16,561 | -21% | 1 | 1 | 0% | 3,030 | 3,165 | +4% | 0 | 0 | — |
case-10 | pass→pass | 15,018 | 12,755 | -15% | 1 | 1 | 0% | 1,670 | 2,718 | +63% | 0 | 0 | — |
case-11 | fail→pass | 12,287 | 15,272 | +24% | 1 | 1 | 0% | 1,894 | 2,962 | +56% | 0 | 0 | — |
case-12 | pass→pass | 18,811 | 16,855 | -10% | 1 | 1 | 0% | 2,103 | 2,759 | +31% | 0 | 0 | — |
case-13 | fail→fail | 19,026 | 17,497 | -8% | 1 | 1 | 0% | 2,037 | 2,793 | +37% | 0 | 0 | — |
case-14 | fail→pass | 20,903 | 19,527 | -7% | 1 | 1 | 0% | 2,254 | 3,056 | +36% | 0 | 0 | — |
case-15 | fail→pass | 9,801 | 17,536 | +79% | 1 | 1 | 0% | 1,694 | 2,857 | +69% | 0 | 0 | — |
case-16 | fail→pass | 21,481 | 18,823 | -12% | 1 | 1 | 0% | 2,285 | 2,787 | +22% | 0 | 0 | — |
case-17 | fail→pass | 16,851 | 12,026 | -29% | 1 | 1 | 0% | 1,805 | 2,770 | +53% | 0 | 0 | — |
case-18 | fail→pass | 17,210 | 20,043 | +16% | 1 | 1 | 0% | 1,982 | 2,857 | +44% | 0 | 0 | — |
case-19 | fail→pass | 33,444 | 15,664 | -53% | 1 | 1 | 0% | 2,308 | 2,635 | +14% | 0 | 0 | — |
case-20 | pass→pass | 28,124 | 26,755 | -5% | 1 | 1 | 0% | 4,209 | 4,640 | +10% | 0 | 0 | — |
case-21 | pass→pass | 18,153 | 23,453 | +29% | 1 | 1 | 0% | 2,406 | 3,241 | +35% | 0 | 0 | — |
case-22 | pass→pass | 6,885 | 12,273 | +78% | 1 | 1 | 0% | 1,069 | 1,816 | +70% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +59 percentage points is the difference between those two pass rates over the 22 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.