Merch Return Rates: 60 Days of Real Data (2026)

Almost nobody publishes real print-on-demand numbers. So here are ours.

This is 60 days of Merch on Demand sales from a single account, US marketplace, pulled straight from the royalties report. One account is not an industry, and your numbers will differ — but it's real data rather than a guess, and the shape of it turned out to be more useful than the headline figure.

The headline

474 designs with at least one sale
1,645 units sold
  149 units returned
  9.1% return rate

Nine percent is roughly what you'd expect for apparel. If you stopped there, you'd conclude returns are a modest cost of doing business and move on.

That conclusion is wrong, and the reason is in the distribution.

The average hides everything

Only 87 of 474 designs had a single return. The other 387 had none at all. So the 9.1% isn't spread thinly across the catalogue — it's concentrated in about one design in five.

And within those 87, the spread is enormous. Five designs, anonymised, from the same 60 days:

Design	Units sold	Returned	Return rate
A	57	13	23%
B	189	13	7%
C	8	5	63%
D	50	5	10%
E	20	5	25%

Design B is the account's best seller and behaves normally. Design A sells well and comes back at nearly a quarter. Design C is small-volume but more than half of what sold came back.

An account-level return rate would have told us to shave about 9% off every royalty. What the per-design data says is different: most designs need no adjustment at all, and a handful are quietly losing money on every sale.

What that does to break-even

The arithmetic is short. On a $16.99 shirt paying a $2.02 royalty:

Break-even ACoS = royalty ÷ price = 11.9%

Apply each design's own return rate to its royalty:

Design	Effective royalty	True break-even ACoS
B (7%)	$1.88	11.1%
D (10%)	$1.82	10.7%
A (23%)	$1.56	9.2%
E (25%)	$1.52	8.9%
C (63%)	$0.75	4.4%

Design C needs to run under 4.4% ACoS to make money. In practice that means it should not be advertised at all — at a 10% conversion rate its maximum affordable bid is about $0.075, which is below Amazon's minimum bid of $0.02… technically biddable, but nowhere near competitive. The ads cannot work; the design can only earn organically.

We were bidding on it at the account default. That is the entire finding.

The window trap

One more thing worth knowing, because it will mislead you in the opposite direction.

Same account, three windows:

Window	Units	Returns	Rate
14 days	371	42	11.3%
30 days	944	87	9.2%
60 days	1,645	149	9.1%

The shorter the window, the worse the return rate looks — and it's an artefact, not a trend. Returns arrive weeks after the sale that produced them. A 14-day window counts returns generated by sales from before the window against only 14 days of sales, so the denominator is too small.

If you judge returns on a two-week report, you will consistently overestimate them by two points or so. Use 60 days for return rates. Use the shorter windows for what they're good at — recent bid movement.

What we changed

Three things, in order of how much they mattered.

Return rate became a per-design number, not an account number. Every design's royalty is now discounted by its own return rate before any profit calculation. This alone reclassified several campaigns from "fine" to "losing". Designs above roughly 20% returns stopped being advertised. Not paused as products — they still sell organically, and organic sales cost nothing when one comes back. They just cannot carry paid traffic. Returns went into the review table. Seeing 13 next to a campaign explains a low royalty figure instantly, where before it looked like an attribution problem.

What we'd tell you to do with your own data

Pull 60 days of your royalties report. Not 14.

Compute return rate per ASIN, not per account.

Discount each royalty by its own rate, then recompute break-even ACoS per design.

Anything above ~20% returns: stop advertising it. Check the listing — sizing charts and colour accuracy are the usual culprits — but don't fund paid traffic to it in the meantime.

Re-check quarterly. Return rates move when Amazon changes suppliers or you change a mockup.

None of that needs software. It's a pivot table and twenty minutes.

What it does need is your sales report and your ad data in the same place, joined by ASIN — which Amazon does not do for you. That join is what PPC Optimizer Pro exists to make automatic: royalty attributed per design, returns subtracted per design, and profit per campaign, keyword and targeting type calculated from what you actually keep.

Design identifiers are anonymised. Figures are from one US Merch on Demand account over 60 days and are shared as a real sample, not an industry benchmark.

60 Days of Real Merch Sales Data: What Returns Actually Cost