Why we think this
is a category.

There is a listing history on Grailed for Apple t-shirts from the nineties that reads like a sneaker market. The first-edition Apple T-Shirts book — a book, of company merch — goes for around $850 on AbeBooks. Figma's Config store had a two-hour line, and the jacket it listed on the day resold for $550 on eBay, which Figma wrote about themselves.

None of that is a merch story. It is a fashion story that happens to be about software. People rep the tools they use the way they rep a band or a team, and they have done for a while now.

The demand moved. The supply didn't.

Here is the gap. A company with genuine fans — a dev tool with a cult following, an AI product people are evangelical about, a platform whose users have opinions about its logo — has a Supreme-level audience and meets it with a print-shop tee. Not because they don't care. Because the alternative is an eight-month engagement with a brand studio in New York, and merch is nobody's priority on a Tuesday.

So the top twenty tech brands do it properly and everybody else ships a conference tote that ends up as a bin liner. There are, roughly, five hundred software companies with real fans and no brand studio. Nobody serves them.

What the economics look like on the other side

The reason the default is bad is that the default is built for volume, not for fans. Amazon's merch programme pays a creator $2.44 on a $19.99 t-shirt. That number is not a scandal, it is just what happens when the product is a blank commodity and the decoration is a print.

Meanwhile Fanatics did $8.1 billion in 2024 on licensed sports merch. There is no equivalent in software, and it is not because the affection isn't there. It is because there is no supply chain pointed at it.

Three things that make a piece good

Nothing here is clever. It is just the difference between a garment someone keeps and one they don't.

1. Embroidered, not printed. A plastisol chest print is the single strongest signal that a garment was given away rather than chosen. It also cracks after ten washes. Embroidery survives the garment.
2. A blank worth putting a mark on. Name the style number. If a vendor won't tell you what blank they're using, it's because you'd rather they didn't.
3. Scarcity that is real. A numbered run of 100 that actually stops at 100. If the number on the tag is decorative, everyone can tell.

Why we built it as software

The constraint that beat everyone else is not manufacturing. It is that merch is never a priority, so any process with a six-week lead time in front of it dies before it starts — even with a warm introduction, even when everyone agrees it is a good idea.

The only way through is to make the first step free and instant, and then make the thing at the end of it good enough that it becomes a priority on its own. Upload a mark, get a real store on your own address in about eight minutes, hold no stock, spend nothing. If it turns out your users want it, you will know from the orders rather than from a survey.

Ask a merch supplier which blank they are using and watch what happens. You will usually get a weight, a fibre content, and a warm adjective. What you will rarely get is a brand and a style number, and the reason is straightforward: the blank is where the margin is, and a style number is searchable.

The four questions

1. What is the blank, by brand and style number? "Heavyweight cotton, 280gsm" is a description. "premium blanks Crew" is a fact you can look up, price, and hold to.
2. Print or embroidery, and where exactly? Ask for the placement in millimetres from a named seam. A vendor who works to millimetres is a vendor who has done this before.
3. What is the minimum, and what happens to the leftovers? The real cost of a 250-piece minimum is not the unit price. It is the eighty extra-smalls in a cupboard eighteen months later, and the fact that nobody will run the second drop because of them.
4. What is the one-time cost, and who pays it? Digitising a mark for embroidery is a real job and costs around $100. It is fine for that to exist. It is not fine for it to be discovered at invoice.

Why the answers matter more than the quote

Two quotes at the same unit price can be completely different products. A printed tee on an unnamed blank at $24 and an embroidered piece on a named blank at $24 are not competing — one of them will be worn in two years and one will not survive ten washes.

The tell is not price. It is specificity. A supplier who names the blank, the thread and the millimetre is a supplier who expects you to check.

What we do about it

We name the blank on the product card, so the garment your buyer receives is the garment the page described. There is no minimum, because we make to order rather than to a warehouse. And we absorb the digitising cost rather than discovering it at invoice, because charging a setup fee to find out whether your users want merch is the fastest way to never find out.

Here is a pattern that took us a while to admit. You get a warm introduction to exactly the right person at exactly the right company. They love the idea. They say so, in writing, with an exclamation mark. And then nothing happens for four months.

It is not a rejection and it is not a lie. It is that merch is a genuinely good idea that is never this week's problem. It loses to the release, the hire, the outage and the board deck, every week, forever, and it will keep losing however good the deck is.

The mistake is treating it as a sales problem

The instinct is to sell harder: a better deck, a stronger case study, a tighter follow-up cadence. That is the wrong instrument. You are not fighting a competitor, you are fighting a queue, and nothing you say moves you up a queue that is sorted by urgency.

What moves you is removing the reason to queue at all. If the first step costs nothing, requires no approval and takes four minutes, it does not need to beat the release for priority. It only needs to beat a coffee.

What that changes about the product

• The first step cannot be a call. A call is a calendar negotiation, and a calendar negotiation is a queue.
• The first step cannot cost money. A card is an approval, and an approval is a queue.
• The first step cannot produce a mockup. A mockup needs a decision to become anything. A real store on a real address does not — it is already the thing.
• The output has to be good enough to forward. The person who uploads the logo is rarely the person who decides. What you are really building is something they can paste into a channel without embarrassment.

Why this is possible now

A store costs almost nothing to build and takes minutes, which is the fact that makes the whole approach possible. If the first step had been expensive, we would have had to qualify before building, and qualifying is a call, and a call is a queue.

That is the entire strategy: make the cheap thing good, put it in front of people with no gate on it, and let the ones who care come back on their own schedule rather than ours.