The Only Two Kinds of Fashion Brands: One Optimizes, One Innovates

There's a question I ask early in almost every audit, and nobody has an answer ready for it.

What are you allowed to waste?

Founders find the question strange. Waste is bad — that's the whole answer, isn't it? You cut what you don't need. You negotiate harder. You tighten the supply chain. Every consultant who ever walked through that door said some version of the same thing, and they weren't wrong.

Except the most profitable house in fashion takes fifteen hours to make a handbag, has more demand than it can serve, and publicly refuses to make it in thirteen.

That's revenue it declines to collect. On purpose. Every year. And it sits on a 41% operating margin while doing it.

I've spent the last seven, eight years inside more than 100 brands, the majority of them in fashion. When I line them all up, the ones that actually make money fall into two groups. Not three. Two.

One group optimizes. The other innovates. They are opposite businesses — different cost structures, different margins, different definitions of a good month — and the thing that separates them most sharply is what they do with waste.

And most of the brands that come to me are in neither group. That's usually the whole problem.

The two models

Strip away the storytelling and every fashion business is one of two machines.

The optimizer sells a lot of units at a thin margin. The money is made in the gap between a low cost and a modest price, multiplied by enormous volume. Because the margin per unit is thin, everything depends on the machine running clean: low costs, tight logistics, fast lead times, disciplined inventory, precise finance. There is no room for slack, because slack eats the entire margin. In this model, waste is a defect. Every unit thrown away, every unit marked down, every returned parcel is money that came directly out of the profit, and there wasn't much of it to begin with.

The innovator sells few units at a fat margin. The money is made because the product is worth dramatically more than it costs to make — and it's worth more precisely because it's hard to make, hard to get, and not attempted by everyone else. Getting there means trying things that don't work, rejecting materials that aren't good enough, and spending hours that a spreadsheet would call unproductive. In this model, waste isn't a defect. It's the cost of the product.

Both work. Both produce excellent businesses. What doesn't work is running one model's cost structure on the other model's price list — which is what most of the market is quietly doing.

The optimizer: waste is an error

In this model waste has a name, and the name is markdown.

A markdown is not a pricing decision. It's a receipt for a mistake made months earlier — the wrong product, the wrong quantity, the wrong timing. The discount is just where the error finally shows up on the P&L. Same with returns: every returned parcel is a unit that cost you shipping twice, handling twice, and often can't go back on the shelf at full price.

So the entire machine is built to not make the mistake in the first place. Inditex says half of its end-product manufacturers sit close to headquarters — Spain, Portugal, Turkey, Morocco — and that stock takes an average of 36 hours to travel from the distribution centre to a European store, up to 48 for the Americas or Asia. That proximity isn't romance about local production. It's the ability to decide late, when you already know what's selling, instead of guessing nine months out.

The numbers say how well it works. In its 2025 financial year Inditex did €39.9bn in sales at a 58.3% gross margin and a 15.6% net margin, carrying inventory equal to just 8.2% of sales — it turns its stock roughly every 72 days. H&M, running what looks from the outside like the same playbook, turns inventory every 121 days, carries 15.5% of annual sales as stock, and takes 5.3% to the bottom line.

Read those two lines again, because they're the argument. Same industry, same category, same suppliers available to both — and a three-fold difference in profit, driven mostly by how much product is sitting still. In this model, inventory that isn't moving is the waste.

Shein pushes the logic to its limit. Its listing filing with the Hong Kong exchange describes discovering "approximately 4,700 new apparel styles each day" in the first quarter of 2026, each made in an initial batch of "approximately 100 to 200 items," with reorders placed only where customers actually respond. Inventory turnover in 2025: 36 days.

That is about as close to zero inventory waste as a clothing company has ever run. And the operating margin it produces is 4.1%.

Which is exactly why what happened next matters. When the United States removed its de minimis exemption, Shein's US revenue fell from $10.5bn to $10.1bn over the year, and was down 14% year on year in the first quarter of 2026. To hold the top line up, marketing spending went from 10.7% of revenue to 14.8% — $6.2bn — and to 15.8% in the first quarter. Operating margin slipped from 3.9% to 2.9%.

That's the part founders miss about this model. A thin margin has no shock absorber. When you've engineered every ounce of slack out of the system, there's nothing left to absorb a bad quarter, a tariff change, a freight spike. The efficiency that makes the model work is the same thing that makes it fragile. Optimization isn't the safe choice. It's a different risk, taken on purpose.

The innovator: waste is the product

Now the other machine.

There's a story everyone tells about luxury leather: houses like Hermès reject enormous quantities of hide to find the few flawless panels, and that mountain of discarded leather is what you're really paying for. It's a great story. It isn't true.

Hermès' own filings say close to the opposite. Cutters are trained from the day they arrive to use leather as sparingly as possible. Digital hide-imaging systems identify defects specifically to maximise usable area. Offcuts are sorted and sold on — the company doesn't even count them as waste in its reporting, and it built an entire métier, petit h, out of consuming what the other métiers leave behind. The line that settles it, from its 2025 universal registration document: "the analysis of hide usage is a management indicator used in the workshops."

Hide yield is a KPI. They are ruthless about material.

So where is the waste?

It's in the hours. And in the growth they refuse to take.

Here is Axel Dumas, the executive chairman, asked about exactly this:

"It takes 15 hours for an Hermès bag. Even if there's a lot of demand, I'm not going to start doing them in 13 hours to raise production."

Sit with that for a second, because every optimizer instinct in your body should be screaming. There is demand you cannot serve. You have found two hours. Take them. Every consultant, every operations review, every board would tell you to find those two hours.

And the answer is no. Permanently.

The constraint isn't rhetorical either — it's physical, and they publish it. Hermès opens on average one production unit per year, each bringing around 300 new hires: the 24th workshop in 2025, the 25th in 2026, then 2027, 2028, 2030, each one named and scheduled years in advance. To staff them it runs its own school, with 12 training centres and 800 candidates enrolled. You cannot buy your way past that. Money doesn't compress it.

What does that "inefficiency" buy? A 71.1% gross margin and a 41.0% recurring operating margin on €16bn of revenue — against 22.0% for LVMH and 11.1% for Kering. About €604,000 of revenue per employee, against €382,000 at LVMH and €336,000 at Kering. And it spends 3.9% of revenue on communication, against 11.4% for LVMH — barely a third — because a product people queue for doesn't need to be advertised as hard.

So the reframe holds, it just sits somewhere different from where the folklore puts it. You're not paying for discarded leather. You're paying for the two hours they refused to save, and for all the bags that were never made because the workshop doesn't exist yet.

That's the most expensive waste on the menu — revenue voluntarily left uncollected — and it's precisely what makes the thing worth queueing for.

Every brand has a waste budget — the question is where you spend it

This is where most people get the idea wrong, so let me be precise.

It isn't that innovators waste and optimizers don't. Everybody wastes. The difference is where they're allowed to do it — and every business that works is ruthless in one place and generous in another.

The fast-fashion machine is generous with bets on product. Look again at what Shein is actually doing: 4,700 new styles a day, 100 to 200 units each. Almost all of them fail. That is the point. The tiny batch is not a production constraint — it's the price of the option, paid thousands of times a day, on the assumption that most of those options expire worthless. Inditex does a slower version of the same thing: more than 700 designers across the group, producing only a few thousand pieces per style until the shop floor says otherwise.

Both companies are spending real money on things that won't work. They just refuse to spend it in units of finished, committed, unsold inventory — because that's the one form of waste this model can't survive. Generous with attempts, ruthless with commitment.

The luxury house is the mirror image, and note carefully that it is not generous with materials — we just saw that hide yield is a managed number. It is generous with time and with forgone volume: the hours it won't compress, the workshops it won't build faster, the orders it won't fill this year. And it is ruthless about the things fast fashion is relaxed about — distribution, price, who gets to sell it.

So the strategic question was never "how do I eliminate waste?" It's:

Which waste am I buying, and can I charge for it?

Get that backwards — generous where you should be ruthless, ruthless where you should be generous — and you get the worst of both. You'll cut the very spending that would have made you different, and tolerate the very spending that's making you slow.

Why you can't be both

This isn't my idea, and it isn't new. It's one of the most cited papers in management, published in 1991.

James March named the two modes exploitation and exploration. Exploitation is refinement, efficiency, selection, execution. Exploration is search, variation, risk-taking, experimentation. His point wasn't that a company should do both. It's that it can't do both freely, because the two "compete for scarce resources." Every euro, every hour, every hire goes to one or the other.

And the contest is rigged. March's description of why is the most useful sentence in the paper. The returns to exploitation, he wrote, are "positive, proximate, and predictable." The returns to exploration are "uncertain, distant, and often negative."

Put those two side by side in any management meeting and optimization wins. Every time. Not because someone chose it, but because it's the only one of the two that can prove itself this quarter.

That's not a theory about how people behave — it's measured. In a survey of 401 financial executives, 80% said they would cut spending on R&D, advertising and maintenance to hit an earnings target. Not villains. Just people responding rationally to what they're measured on.

In fashion, that same instinct looks like this: source the cheaper fabric, cut the sampling round, reorder last season's bestseller instead of backing the new idea, drop the collaboration. Each decision defensible on its own. Together, over about three years, they turn a brand into a commodity — and nobody can point to the meeting where it happened.

Then there's the second reason, and it's the harsher one: efficiency doesn't protect you. Porter's 1996 argument in What Is Strategy? is that operational effectiveness — doing the same things better than rivals — isn't strategy at all, because everyone converges on the same practices. "The more benchmarking companies do, the more they look alike." The gains get passed to customers as lower prices, and nobody keeps the advantage.

His conclusion is the one that should worry any founder who thinks efficiency is the safe path: competing this way produces "pressures on costs that compromise companies' ability to invest in the business for the long term." Optimize hard enough and you consume the surplus that would have funded being different.

So if you're going to be an optimizer, you have to be one at a level competitors structurally cannot copy — an advantage built into your supply chain, not into your effort. That's what those 72 days versus 121 days actually represent. "Better run than average" is not a business model.

And Porter had already written the verdict on the alternative back in 1980: the firm stuck in the middle "is almost guaranteed low profitability."

"But what about Toyota?"

Fair objection, and worth answering properly rather than dodging.

Toyota built the most waste-hostile production system in industrial history — and still invented the Prius. Researchers who studied its plants found high efficiency and high flexibility at the same time. So the trade-off isn't a law of physics.

But look at how that gets done. The research on companies that manage both is consistent: they don't blend the two. They separate them — exploratory work is structurally walled off from the operating business, connected only at the very top. In one study of 35 breakthrough initiatives, more than 90% of the organizations built that way hit their goals. The ones that asked existing teams to do both hit essentially none.

Which is exactly why the big groups can look like they're doing both. LVMH is a portfolio of separated houses. Inditex runs distinct chains. They bought the right to make the choice twice — by keeping the two halves apart.

You have one P&L, one team, and one set of decisions about where next month's money goes. You don't get to make it twice.

Your waste budget is funded by your gross margin

Here's where this stops being philosophy and starts being arithmetic.

You can only afford as much waste as your margin pays for. That's the entire constraint, and it's a subtraction, not a slogan.

Your gross margin has to cover four things, in this order: your fixed costs, your marketing and distribution, the net profit you actually intend to keep — and then whatever is left. That last number is your waste budget. It's the money available to reject materials, scrap a run because the hand feel is wrong, pay for small batches, and run the extra sampling rounds.

Waste budget = gross margin − fixed costs − marketing and distribution − target net profit

Three of those four I've already put numbers on in previous pieces. For a brand selling mainly through its own e-commerce, fixed costs should sit under 10% of revenue. A realistic net profit target is 10% — the good operators land between 10 and 15%. Marketing and distribution, in the brands I see, runs somewhere between 15% and 25%.

If you trade mainly through your own stores, the split moves. Your fixed costs are considerably higher — rent, staff, fit-out — but you spend far less on marketing, because the store does a large part of the acquiring for you. The two lines trade against each other.

What doesn't move is the subtraction. Whatever the split, the number at the end of it is the only money you have to spend on the product itself.

So here's what's left, at each level of gross margin, for a brand selling mainly online:

Read the right-hand column first, because that's where most brands live. Below about 45% gross margin, there is no waste budget at all — the number is zero or negative. Not "tight." Nonexistent. You are spending money you do not have, and the only question is how many years of losses it takes before someone notices.

Allbirds sat at 41–43%. Five straight years of losses, roughly $470m in total, and a gross margin that fell from 52.9% in 2021 to 41.0% by 2023 after the excess inventory went out at markdown — and never came back, even after revenue was cut nearly in half. Nothing there was a failure of taste or product. It was this table.

It also explains how Hermès does what it does. At 71.1% gross margin, spending 3.9% on communication instead of 25% on marketing and distribution, the residue is enormous. Those fifteen hours a bag are paid for by the subtraction.

One caveat on that middle column. For a house that manufactures its own product, the reported cost of sales already contains the making of it — artisans' wages, workshops, tanneries — so Hermès' equivalent markup reads as a modest 3.4× while the price on the shelf is anything but. The fifteen hours sit inside that cost line, not below it. A brand buying finished goods from a supplier carries only the supplier's invoice there. So line yourself up against the theoretical rows, not against the two companies.

And notice what the two columns do to each other. Every point you take out of marketing and distribution lands, one for one, in the product. That is the most underrated trade in this industry: spend less on finding the customer and you can afford to be worth finding.

Now put Reformation next to Allbirds. Everything about both companies reads as an innovator — proprietary or considered materials, a craft-forward story, a similar customer. But Reformation takes about **80% of its DTC revenue at full price**, year after year. Result: profitable in every year from 2018 to 2025 except 2020, on $507m of revenue.

Same segment. Opposite discipline. The variable isn't the product — it's whether the price was allowed to fund the model.

And that's the trap I see over and over. A brand buys the innovator's inputs — small runs, better fabric, local production, a craft story — then prices like a volume player and discounts like one too. They're paying for waste they never charge for.

Waste you can't charge for isn't a strategy. It's just waste.

The test

Four questions. The answers have to point the same direction — if they don't, you're in the middle.

1. Raise your prices 20% tomorrow. How many customers do you actually lose? If the answer is "not many," you have innovator pricing power and you're not using it.
2. Double your batch size. Does the P&L get better or worse? Better means you're an optimizer, and you should be behaving like one — bigger runs, fewer SKUs, tighter costs.
3. What percentage of your revenue comes in at full price? An optimizer needs high sell-through to survive thin margins. An innovator that discounts is destroying the only thing it sells.
4. Where does your money actually go? Into product, materials and people — or into systems, logistics and media? Both are valid. Being unclear which one you're funding is not.

Mixed answers aren't nuance. They're a diagnosis.

Pick one

Let me be precise about what I'm not saying, because it matters.

I'm not saying the mid-market is dead. That was fashionable advice for years, and the data has turned hard against it. Here is McKinsey and The Business of Fashion, in The State of Fashion 2026:

"While luxury players raised prices without corresponding improvements in product quality or creativity, design-led brands from the value segment up through affordable luxury elevated their products and store experiences. Now, the midmarket is the fastest-growing one, and it's replacing luxury as fashion's main value creator."

Read that first sentence again, because it's this whole article in miniature. One group raised the price without raising the spend. The other raised the spend and earned the price. Meanwhile luxury has been losing aspirational customers for two years running.

So sitting at a mid-market price point is fine.

What kills brands isn't the price segment. It's the incoherence: an innovator's cost base underneath an optimizer's price list. That combination is fatal at any price point, and it's the single most common thing I find in an audit.

The evidence is everywhere once you look for it. Direct-to-consumer gross margins run roughly 23 points higher than wholesale — and yet DTC operating margins come in lower, because the extra margin is entirely consumed by the cost of acquiring the customer. As Simeon Siegel put it: "No one eliminates the middle person, they simply become the middle person." Outdoor Voices burned about $2m a month on $40m of sales, buying volume it could never convert into margin while carrying the fixed costs of a craft brand. And Everlane — a company whose entire identity was built on the transparency of its costs — was sold to Shein in 2026 for $100m, in a deal that cleared roughly $90m of debt.

That's what the middle actually looks like. Not a market position. A cost structure that ran out of time.

So: if you're going to optimize, then optimize. Cut deep, get the fixed costs under control, build the volume, and stop paying for a craft story your price can't support.

If you're going to innovate, then charge for it. Raise prices until they fund the waste that makes you worth buying. Make less. Hold full price.

Both are good businesses. Pick one, and let the waste follow the decision instead of the other way round.