The Market Stall Never Went Away

Most companies lose something when they scale

Most companies lose something when they scale. The founder used to stand in the shop and hear what people asked for. Then there were twelve shops, then forty, and the feedback loop got replaced by a quarterly report and a buying team making bets nine months out.

Ghanda didn't lose it. They industrialised it.

The brand started in 2003, with Josh Rudd selling T-shirts out of the back of his car at markets around Torquay. In an interview with Forte Magazine he described the loop he was running: he'd sell at the market, see exactly which designs moved, go back that night, print the same ones again, and sell them the next day.

That's a sell-then-make cycle measured in hours. Nearly every apparel business on earth runs the opposite one — make, then hope.

The interesting part is what happened when they got bigger. Twenty-plus stores across Australia now, and according to their own About page the majority of what's on those racks is still hand-printed by their own team, in-house, printed to order. The sentence they use is quiet and easy to skim past: their range is directed by what customers want, so they don't overproduce styles that won't sell.

That's not a sustainability line. It's an operating model, and it has a name.

The step they chose to own

Vertical integration is usually discussed as though it's one decision — you're integrated or you're not. It isn't. It's a decision about which step, and almost all the value is in picking the right one.

Ghanda doesn't spin yarn. They don't weave fabric or cut and sew garments. Those steps are capital-heavy, globally competitive on cost, and — this is the part that matters — they don't create variety. A blank fleece crew is a blank fleece crew.

Printing is different. Printing is the step where one SKU becomes forty. It's where all the forecasting risk lives, because that's the moment you commit to guessing which of the forty someone will want. Own that step and you can hold inventory in its undifferentiated form, then commit at the last possible moment.

Operations people call this postponement, and the textbook case is Benetton, which switched from knitting garments in coloured yarn to knitting them undyed and dyeing them once orders came in. Same garment, same factory, later commitment point. The forecasting problem doesn't disappear, it just shrinks from “which forty colours in which quantities” to “how many units total” — a far easier question that you get much closer to being right about.

Ghanda's version is blanks and ink. Same principle, different industry.

What the discount rack tells you

Rudd's critique of how everyone else operates, in the same interview, is sharper than most consultants would put it. Other companies guess what customers want, and when the guess is wrong they mark it down until it clears — which means people end up buying things nobody actually wanted, at a price that destroys the margin.

That's worth sitting with, because it reframes what a markdown is.

A markdown is not a pricing decision. It's the settlement of a forecasting error made months earlier by someone who has probably moved on.

By the time it hits the floor it's already too late to fix — the fabric was bought, the garment was made, the container shipped, the shelf space allocated. The discount is just the cost of that error becoming visible.

Which means the sale rack is diagnostic. If you want to know how good a retailer's demand sensing is, don't read their forecast accuracy report. Count the racks at the back of the store.

A print-to-order model doesn't have that rack, because the error never got made. Note: it can still get the design wrong. It just doesn't pay for the mistake in units.

What it actually costs

This is where most write-ups of vertically integrated brands go soft, so let's be honest about the trade.

In-house printing converts a variable cost into a fixed one. When you outsource, you pay per unit and your cost base flexes with demand. When you own the presses and employ the printers — around 25 warehouse jobs as of a 2014 report — you pay whether the machines run or not. A quiet quarter doesn't reduce that bill.

It also caps you. Outsourced production scales by signing another supplier. In-house production scales by buying another press and hiring people who know how to use it, which takes months and capital. If a design goes unexpectedly viral, the constraint isn't demand, it's your own throughput. That's a good problem to have exactly once, and an expensive one after that.

And there's a geographic constraint. Printing in Australia means printing at Australian labour rates, competing against brands printing at a fraction of that. The model only works if the responsiveness is worth more than the unit cost difference — which it is at their scale and positioning, and would not be for everyone.

So this isn't a strategy you adopt because it sounds virtuous. It's a bet that speed and accuracy beat unit cost in your specific market. For a design-led brand whose product is the artwork rather than the garment, that bet looks right.

The capability nobody else kept

One detail from that 2014 report stopped me. At the time, Ghanda was said to be one of only two screen printers in Victoria still actively printing their own product — down from fourteen.

I'd treat the exact figures as a snapshot from a magazine rather than an audit, but the direction is the point, and it's the most strategically interesting thing about the company.

An entire regional capability was offshored over roughly a decade. Every one of those decisions was individually rational: printing is cheaper elsewhere, the quality is fine, why tie up capital in machines. And collectively they produced a market where almost nobody can do the thing anymore.

If you're one of the two who kept it, you now hold something that can't be quickly bought back. Not because the equipment is exotic — presses are purchasable — but because the people who know how to run them, fix them, and hit a colour consistently have moved on to other work. Capability decays faster than equipment does, and it comes back slower.

This is the part I'd want any operator to take from Ghanda, because it generalises well beyond apparel. Every time you outsource a step, you're also selling an option. The option to respond quickly. The option to change your mind late. The option to do something your competitors structurally cannot. Those options don't show up on the business case, because the business case compares unit costs — and the value of optionality never appears in a unit cost comparison.

That doesn't mean don't outsource. Most of the time the option isn't worth what it costs to hold. But it should be a line in the analysis, priced deliberately, rather than a thing you discover you gave away when the market shifts.

The exposure

There's a vulnerability built into this that's worth naming, and it surfaced publicly.

In December 2021 the ABC reported that Ghanda was in a dispute with Cotton On over alleged copies and trademarks. I don't know how it resolved, and the merits aren't the point here.

The point is structural. When your differentiator is speed of production, a larger competitor can't easily copy you — they'd have to rebuild the capability. But when your differentiator is the design, they can copy you in a week and out-distribute you instantly.

So a business built on in-house design and in-house printing has two assets with very different defensive profiles. The printing is a moat. The artwork is not, unless you defend it. Which means design IP stops being a legal housekeeping matter and becomes a core strategic asset — the thing you register, monitor and are prepared to fight over.

Plenty of founder-led businesses figure that out several years after they needed to.

What transfers

Three things, if you're running something that isn't a clothing brand.

Pick your integration step by asking where the forecast error lives, not where the cost is. The cheapest step to own is rarely the most valuable one.

Treat your markdowns, write-offs and idle inventory as a measurement of how badly you're guessing, not as a pricing problem. They're the bill for a decision made months upstream.

And before you outsource a capability, write down what you can do today because you have it in-house that you won't be able to do afterwards. If the answer is nothing, outsource it without a second thought. If the answer is something, you now know the actual price.