The three-phase model

Each phase has an entry cost, an exit gate, and a specific question it exists to answer. Do not skip a phase because the next one looks more exciting — the gates protect your cash.

Niche & brand

The niche decision constrains everything downstream — pricing, print method, ad targeting, even which blank you buy. Make it before you draw anything.

What makes a niche work

A viable t-shirt niche satisfies four tests at once. Drop any one and the economics degrade fast.

Niches that structurally work

Not recommendations — patterns. Each has identity, a gathering place, and disposable income:

• Occupational in-groups: nurses, welders, sysadmins, chefs, pilots. Strong identity, dense online communities, and gift demand from family.
• Hobbies with equipment budgets: climbing, sim racing, fly fishing, mechanical keyboards, homebrewing. If they already spend on gear, apparel is a small add.
• Local and regional pride: a specific town, trail, lake, or dialect. Small audiences, but almost no competition and very high conversion.
• Subculture aesthetics: a music scene, a design movement, a game community. Highest passion, but also the highest IP risk — see §11.

Brand minimums

You do not need a brand book. You need enough consistency that a repeat customer recognises you:

• Name — short, pronounceable, .com available, and not a registered trademark in class 25 (apparel). Check your national trademark register and the EUIPO/USPTO databases before you print anything.
• One typeface pairing and a fixed palette. Applied to the shop, the packaging, and the ads alike.
• A neck label and a poly mailer insert. Cheap, and the single biggest driver of "this is a real brand" perception.
• A stated position on quality. If durability is your differentiator (it should be — see §04), say so on the product page and back it with the wash guarantee.

Print technology

Six methods can put an image on a shirt. They differ less in maximum quality than in what they cost per unit, what they demand of you, and which fabrics they tolerate.

The recommendation

Why not the others

• DTG — a good machine is $15k+, needs daily maintenance to stop the heads clogging, and produces the weakest durability of the serious options. Fine as your POD provider's problem in Phase 0; a bad thing to own.
• Sublimation — chemically permanent and completely weightless to the touch, which sounds ideal. But the dye only bonds to polyester, and only shows on light fabrics. A white or heather-grey poly shirt is a narrow catalogue. Worth adding later if your niche is athletic.
• HTV — every piece is hand-weeded, which doesn't scale past a few dozen, and it can't do photographic art. Keep a cutter around for one-off personalisation.
• Embroidery — a genuinely great premium add-on (caps, left-chest logos) but it isn't a print, and a machine is a separate business. Outsource if you want it.

Durability engineering

This is the section that matters most. A lasting print is not something you buy — it is something you verify. What follows is the physics, the process parameters, and the test protocol that turns "lasting" into a number.

How each ink actually bonds

Understanding the mechanism tells you what to control:

• Plastisol (screen) — PVC particles suspended in plasticiser. It does not dry; it fuses. Heat the whole film to ~160 °C / 320 °F and the particles swell and lock into a continuous flexible layer. Below that, you get a film that's solid on top and powder underneath. It will crack.
• Water-based & discharge (screen) — pigment carried in water, which must first evaporate, after which a binder cures and bonds pigment into the fibres. Softest hand feel of anything; excellent durability; harder to run because it dries in the screen while you work.
• DTF — pigment printed on PET film, then dusted with a hot-melt adhesive powder and melted. The heat press pushes that adhesive into the fabric where it solidifies as a mechanical grip. Durability lives and dies on adhesive melt: too cool and it never wets the fibres, too hot and it embrittles.
• Sublimation — at ~200 °C the solid dye becomes a gas, polyester fibres open, gas enters, fibres close. The dye is now inside the fibre. There is no coating to crack, ever. Chemically permanent — but only on polyester.
• DTG — inkjet pigment onto a pretreated surface. On dark shirts a pretreatment layer must be sprayed and pressed first so the white underbase has something to sit on. Skipped or uneven pretreatment is the usual DTG failure.

Process parameters

Start here, then dial in against your own equipment. Every press lies about its temperature — see verification below.

The three tests that define "lasting"

Run all three on every new design/blank/ink combination before it goes on sale. Not once a year — every new combination, because durability is a property of the combination, not of the method.

Fabric choices that make prints last

• 100% combed ring-spun cotton is the durability sweet spot. Tighter, smoother surface than open-end cotton means the ink film sits on a flat plane instead of bridging fibre peaks — fewer stress points, less cracking. It's also what your customer expects at $30.
• Avoid heavy poly-blends for dark garments until you have the dye-migration process nailed. Tri-blends look and feel wonderful and are the hardest thing on this list to print reliably.
• Wash new blanks before printing? No — but know that blanks carry sizing/softener residues from the mill that can inhibit adhesion. If a specific blank fails the crock test for no obvious reason, residue is the suspect.
• Watch fabric weight. Lightweight fashion blanks (110–130 gsm) flex more, which stresses the print more. Mid-weight (150–185 gsm) is more forgiving and reads as better quality anyway.

What you tell the customer

Durability is only a differentiator if you make a claim and honour it. Put the care instructions on the product page, in the confirmation email, and on a card in the box:

Blanks & sourcing

The blank is 40% of your cost and 90% of what the customer physically feels. Nobody returns a shirt because the design was 2mm off-centre; they return it because it fit badly or felt cheap.

Size distribution

Ordering equal quantities across sizes is the classic first mistake: you sell out of M and L and sit on XS and 2XL forever. A typical unisex adult distribution:

Skew it toward your actual niche — a climbing audience runs smaller, a trucking audience runs larger. After Phase 0 you will have your real distribution from POD order data. Use it. That data is one of the main reasons Phase 0 exists.

Sourcing rules

• Buy from a distributor (S&S Activewear, SanMar, AlphaBroder in the US; Stanley/Stella stockists or Ralawise in the EU/UK) rather than retail. You need a business account and usually a tax ID.
• Always order a sample first and wash it five times before committing. Blanks change spec between production runs.
• Buy the same dye lot for a design where you can. Black blanks in particular vary noticeably between lots, and customers who buy two notice.
• 2XL and 3XL cost more — $1.50–3 extra. Price them higher; nearly every brand does and nobody objects.

Artwork specs

Files that meet these specs press correctly the first time. Files that don't will waste blanks — and a wasted blank is pure loss.

Design rules that survive contact with fabric

• Fabric is not paper. It's a textured, flexing surface. Fine detail, hairlines, and subtle gradients that look crisp on screen fragment on cotton. Design bolder than feels necessary.
• Less ink lasts longer and feels better. A big solid block is heavier, hotter to wear, and has more area to crack. Distressed textures and halftones reduce coverage while looking intentional — this is why so much good apparel art is distressed.
• Check contrast against every blank colour you sell. A design that sings on black can vanish on navy. Mock every colourway before listing.
• Keep a master file at 2× print size in vector where possible. You will want it for hoodies, caps, and stickers later, and re-drawing is expensive.
• Screen printing charges per colour. A 3-colour design and a 7-colour design have very different Phase 3 economics — design for 1–3 colours if you intend to scale that design.

Unit economics

Move the sliders with your own real numbers. The point of this section is to show you exactly why Phase 2 exists — and why POD cannot carry paid advertising.

The two scenarios that matter

Levers, in order of impact

1. Raise the price. $32 → $36 adds $4 to the bottom line with no cost increase. In a well-chosen niche the volume loss is smaller than you fear — you're not selling a commodity. Test it before assuming otherwise.
2. Raise AOV. Two shirts in one parcel share one shipping cost and one CAC. A "2 for $58" bundle can nearly double net profit per order. This is usually the single biggest available win.
3. Free shipping over $50. Pushes AOV up and reframes shipping as a customer choice.
4. Sell to the same people again. A repeat order has zero CAC — a $17.46 gross profit becomes $17.46 net. Email flows (§09) are the cheapest revenue in the business.
5. Cut product cost. Listed last deliberately: squeezing $0.50 from a blank is the smallest lever and the one most likely to cost you a return. Do not solve a margin problem by buying a worse shirt.

Startup capital

The web shop

A companion container in this project holds a working storefront. Here's the reasoning behind it and what has to be true before you take real money.

Platform choice

What the product page must do

• Photos on a real body, not a flat lay. The single largest conversion factor in apparel. Flat lays don't show fit, and fit is what the buyer is anxious about.
• A size chart in actual measurements — pit-to-pit and length in cm. "Model is 183 cm, wearing L" beats any chart.
• A close-up of the print. If durability is your claim, show the texture where the ink meets the fabric.
• The durability guarantee, stated plainly, near the buy button.
• Care instructions on the page. Also sets expectations that reduce claims.
• Every colourway mocked accurately. Contrast changes; see §06.

Before you take real money

• Payments tested end-to-end with a real card, then refunded
• Tax/VAT configured for every region you'll ship to
• Shipping rates that reflect what carriers actually charge you
• Order confirmation and shipping-notification emails firing
• Returns policy published and legally compliant (§11)
• Privacy policy and cookie consent live if you have EU visitors
• Abandoned-cart flow armed
• Mobile checkout tested on a real phone — most of your traffic

Marketing

You have one job before scaling: find a single channel that produces orders below your gross profit. One channel, not five.

Organic — start here

• Show the process, not the product. Heat press reveals, peel shots, and packing videos consistently outperform product photos on TikTok and Reels. The moment of pulling the film off a fresh print is genuinely satisfying to watch and costs nothing to film.
• Post where your niche already gathers. The subreddit, the Discord, the forum, the local event. Participate as a member first — a store link dropped by a stranger gets removed and resented.
• Seed micro-influencers. 5–20k followers inside your niche, gifted a free shirt. Vastly better ROI than large generalist accounts, because their audience is your niche.
• Photograph real customers. Ask, offer a discount, use it everywhere. Nothing you shoot yourself converts like it.

Paid — only after organic proves the design

• Don't run ads for an unvalidated design. Ads amplify; they don't create demand. Amplifying a design nobody wants just buys you a faster no.
• Meta first. Interest targeting maps onto niches well. Advantage+ campaigns need roughly 50 purchases in the pixel before they learn — budget for that as tuition.
• Retargeting has the best ROAS you'll see and the smallest ceiling. It's a closer, not a growth channel.
• Judge on contribution margin, not ROAS. A 2× ROAS at 38% gross margin loses money. The calculator in §07 tells you your real break-even CAC — that's the number to optimise against.

Email — the cheapest revenue you have

• Abandoned cart — three emails at 1h, 24h, 72h. Routinely the highest-ROI automation in ecommerce, by a wide margin.
• Welcome flow — capture email with 10% off, then tell the brand story. This is where your niche credibility does its work.
• Post-purchase — care instructions (which lowers claim rates), then a review request at day 14.
• New design announcements — your list is the reason design #7 doesn't need paid traffic at all.

Operations

Phase 2 makes you a small factory. Boring consistency here is what makes the durability claim in §04 true for every order, not just the sample.

The fulfilment loop

1. Batch daily. Print orders once a day, not as they arrive. Context-switching is what makes small-scale fulfilment miserable.
2. Pre-press the blank for 5 seconds to drive off moisture and flatten wrinkles. Skipping this causes adhesion failures that look like ink problems.
3. Use a placement jig — a cardboard template for collar distance and centring. Eyeballing produces the crooked prints that generate returns.
4. Press to spec (§04), cold peel, post-press under parchment.
5. Stretch-test every single unit. Two seconds. Catches cure drift before the customer does.
6. Fold, bag, label, insert card.
7. Ship within 48 hours. Faster than POD is a real, defensible advantage — use it in your copy.

Inventory discipline

• Reorder at 30% of a size's stock, not at zero. Distributor lead times are 3–7 days and stockouts on M/L are pure lost revenue.
• Track sell-through per design per size. A spreadsheet is fine. What you learn: which sizes to buy next, and which designs to promote to Phase 3.
• Dead stock is a decision, not an accident. If a design hasn't sold in 90 days, bundle it, discount it, or give it away as seeded influencer stock. Cash beats cardboard boxes.
• Never stock more than 90 days of projected sales for any design. Tastes shift; blanks don't spoil but capital does get trapped.

Returns

• Sizing is the reason. Attack it with better measurements (§08), not a stricter policy.
• Replace defects immediately, no return shipping. Getting the shirt back costs more than the shirt. Refund or replace, apologise, move on — and log it, because a defect cluster is a cure problem.
• Expect 3–6%. Above 8%, the problem is your size chart or your print quality, and the data will tell you which.

Legal & tax

Nothing here is exotic, but two items on this page can end the business outright: intellectual property and unregistered tax obligations.

Setup checklist

• Register a business entity (LLC / Ltd / AS / ENK / sole proprietorship — depends on your country)
• Separate business bank account from day one — mixing funds creates an accounting nightmare
• Register for VAT/sales tax where you cross the threshold (EU distance-selling and US economic nexus rules both bite on where the customer is, not where you are)
• Trademark search in class 25 (apparel) before committing to the brand name
• Commercial-use licences for every font and stock asset — desktop licences usually do not cover merchandise
• Written agreement with any freelance designer transferring copyright to you — "I paid for it" does not transfer rights by default in most jurisdictions

Selling rules you must comply with

• EU 14-day right of withdrawal. Customers can return for any reason. Note the exemption: genuinely personalised goods are excluded — but a stock design in a chosen size is not personalised. Don't try to use this exemption for a normal catalogue; it won't hold.
• Fibre content and care labelling. Required in both the EU and US. Your blanks arrive labelled — if you remove the neck label to add your own, you must reproduce the fibre content and care information.
• GDPR if you touch EU customer data: a privacy policy, a lawful basis for marketing email, and a working deletion path.
• Country of origin marking on imports.

12-month roadmap

Targets are deliberately modest. A business doing 150 shirts a month at 55% margin is real, profitable, and — unlike most stores — still alive at month 12.

Risk register

Ranked by what actually ends small apparel brands, not by what feels scariest.

Do this first

This week, in order. Nothing here costs more than a few hours or a few dollars.

• Write your niche down in one sentence and run it against the four tests in §02. If it fails one, keep going — this is the highest-leverage hour in the whole plan.
• Find the gathering place. Name the specific subreddit, Discord, hashtag, or event. If you can't name one, the niche isn't findable and the rest won't work.
• Trademark-search your brand name in class 25 before you fall in love with it.
• Sketch three designs that pass the insider test — outsiders puzzled, insiders delighted.
• Order the blanks you're considering and wash them five times. Wear them. You're about to bet on this fabric.
• Stand up the shop with POD connected. Ugly and live beats beautiful and pending.
• Post once in the gathering place — as a member, not a store. Watch what happens.