We use Google Analytics 4 to see which pages are useful. It loads only if you accept it, and then sets two cookies, _ga and _ga_C372PEK7KT. Nothing is stored until you choose, and nothing it records is used for advertising or sold.
Fashion destroys 30 billion unsold garments yearly — missing body data at the point of production.
TL;DR: Fashion overproduction is not an ethics failure — it is an information failure. The industry produces roughly 30% more than it sells because brands cannot reliably predict demand by size, geography, or season. According to UNEP (2018), the apparel sector generates around 92 million tonnes of textile waste annually. Sizing fragmentation is the single largest correctable source of that uncertainty. Measurement portability is the structural fix.
The United Nations Environment Programme estimates the apparel sector produces around 100 billion garments per year, and a substantial share of that output goes unsold. Many of those unsold garments are destroyed — incinerated or landfilled — to protect brand value or avoid logistics costs. The narrative is typically framed as an ethics problem: brands producing irresponsibly, prioritising volume over sustainability. That framing misses the more important causal story.
Overproduction in fashion is substantially an information problem. Brands produce more than they need because they cannot reliably predict how much they need — and sizing fragmentation is a primary source of that unpredictability. Fix the information layer, and the waste problem becomes structurally solvable.
The fashion industry overproduces because production commitments must be made 6–18 months before garments reach consumers, using demand signals that are inherently imprecise. Inventory accuracy has sat near the top of every operational agenda in the sector for years without moving, which is itself the evidence: the problem is not attention, it is information. The 30% figure is not carelessness; it is a rational response to structural uncertainty.
Fashion brands plan production seasons before actual consumer demand materialises. Lead times are long — 90 to 180 days from order to delivery is standard for Italian production, longer for offshore. Demand signals are imperfect at every stage: wholesale forecasts, retail buyer estimates, and consumer sell-through rarely align with one another. The result is a compounding forecast error that grows at each link in the supply chain.
The standard response to demand uncertainty is a buffer. Produce more than expected demand so that stockouts — the most immediately painful failure mode — are avoided. As the Ellen MacArthur Foundation documented in its landmark 2017 New Textiles Economy report, less than 1% of clothing is recycled into new garments, meaning overproduction waste is almost entirely permanent value destruction. The Foundation put the annual value destroyed through clothing underutilisation and disposal at over $500 billion globally.
On UNEP's accounting, the fashion industry is among the largest industrial consumers of water and is responsible for 8 to 10% of global carbon emissions: more than all international flights and maritime shipping combined.
Sizing adds a specific and largely invisible layer of noise to this problem. It is the one driver of overproduction most directly addressable through digital infrastructure.
Sizing fragmentation forces brands to hedge across a wide size distribution for every new collection, market, and fit block — because historical sell-through data by size does not transfer reliably when any of those variables change. A material share of per-SKU overproduction traces directly to inaccurate size-distribution forecasting, which produces waste and lost revenue at the same time: the units cut at the tails do not sell, and the sizes that would have sold ran out.
Because size labels are not standardised across brands or markets, historical sales data by size does not cleanly translate to production planning for a new collection, a new fit block, or a new geography. A brand expanding from Italy into Germany or the UK finds that the size distribution of its new market differs from its home market — but it cannot predict precisely how it differs until it has two or three seasons of sell-through data.
Returns compound overproduction waste. A garment bought and sent back because it did not fit represents a second round of logistics cost with no value delivered, and fit drives return rates into the tens of percent in some online categories.
Two distinct waste mechanisms operate simultaneously when sizing fragmentation is unresolved:
Made-to-order is the production model that structurally eliminates overproduction: no garment is manufactured without a confirmed, paid order. There is no speculative inventory and no size-distribution bet. A well-run made-to-order operation produces no unsold inventory by design. Its unit cost is higher than batch production, and the question is whether the premium is smaller than the end-of-season markdowns and destruction costs it removes. Above a certain price point it plainly is.
This is the model that Italian sartorial tradition has operated for centuries. A bespoke tailor does not maintain a warehouse of jackets waiting for buyers. They hold a relationship with a client — measurements known, preferences documented — and produce to that relationship. The Neapolitan tailoring houses of Kiton and Cesare Attolini built global reputations on exactly this model: no speculative cutting, no unsold inventory, no size-range hedging required.
The challenge historically has been scale. The bespoke model, in its traditional form, is inherently one-to-one and cannot operate at the volume even a premium ready-to-wear brand requires. A tailor can serve perhaps 200 clients per year. A brand serving 200,000 customers cannot replicate that manually.
The innovation that measurement portability makes possible is applying made-to-order precision at a scale that was previously unachievable. When body measurements are digital, storable, and algorithmically matchable against production specifications, the personalisation step can be executed at volume. The made-to-order model is no longer limited to bespoke workshops — it can operate as an infrastructure layer for any brand that adopts it.
Three proven levers reduce the gap between production commitment and actual demand. Each lever attacks the same variable: the gap between what was cut and what a body could actually wear. None of them requires a brand to abandon its fit block, which is why they are adoptable without a strategy change. None requires a full business model transformation to begin — each can be introduced incrementally at the SKU or category level.
Definition
Measurement-based pre-orders
Taking confirmed orders with body measurement specifications before committing to production quantities allows a brand to observe the actual demand distribution — by size, fit preference, and geography — before cutting fabric. A pre-ordered line arrives at production with the demand distribution already observed rather than assumed, which is the whole of the advantage.
Definition
Made-to-order for high-ASP categories
The economics of made-to-order are most favourable at the top of the product range, where the cost of overproduction per unit is highest and where buyers most readily accept a 3–6 week production lead time in exchange for precision fit. A jacket retailing at €1,800 that goes unsold represents a capital write-off an order of magnitude larger than a €180 T-shirt — the buffer economics are completely different.
Definition
Measurement data as a size-distribution signal
A brand with access to body measurement profiles across its customer base can use that aggregate data to plan size distributions more accurately for standard-production lines — reducing the buffer required to cover demand uncertainty without moving to full made-to-order. This is the minimum viable application of measurement infrastructure: better planning, not structural model change.
Each of these levers reduces the gap between production commitment and actual need. The EU Ecodesign for Sustainable Products Regulation (2024/1781) — which entered into force in July 2024 — creates mandatory digital product passports for textiles that will make overproduction patterns traceable and regulatorily relevant. Brands that reduce speculative inventory before that data becomes auditable are building a structural advantage, not merely a sustainability credential.
Measurement portability — the ability for a consumer's body measurement profile to travel with them across purchase contexts — is the infrastructure that makes sizing-accurate production planning possible at scale. When a brand accesses the actual body measurement distribution of its buyer base, rather than inferring it from size-label purchase history, it gains a fundamentally better demand signal, and the buffer it needs to carry against uncertainty shrinks from the first full season of data.
The GS1 Digital Link standard provides the interoperability layer that makes measurement data portable across systems: garment identifiers linked to product specifications, and buyer profiles linked to measurement records, can be connected algorithmically at the point of production planning. This transforms measurement data from a CRM feature into production infrastructure.
In made-to-measure operations, the pattern is consistent and measurable: when a buyer's measurement profile is available at the point of order, production specification error drops to near zero, and per-order contribution margin improves because alteration and remake costs are eliminated. When a client's measurements travel with them from their first fitting to every subsequent order — regardless of which brand or atelier they are ordering from — the production planning step becomes a lookup, not a guess.
The aggregate effect across a customer base is where it compounds. Knowing what share of your European buyers actually falls in a given chest range — rather than inferring it from size-M sell-through — turns the size distribution from an estimate into a measurement, and a buffer sized against a measurement is a fraction of one sized against an estimate. That is not a marginal improvement. It is structural waste elimination.
The United Nations Environment Programme estimates that the global apparel sector produces approximately 100 billion garments per year, of which a substantial share goes unsold. A significant share of those unsold garments is destroyed rather than redistributed, primarily to protect brand positioning and avoid secondary-market price dilution. The Ellen MacArthur Foundation's 2017 New Textiles Economy report put the annual value destroyed through clothing underutilisation and disposal at more than $500 billion globally.
Producing less requires more accurate demand prediction. Demand prediction in fashion is constrained by three structural factors: long lead times that require production commitments before demand is observed; sizing fragmentation that makes historical sell-through data unreliable when fit blocks, markets, or collections change; and distribution chains that add multiple layers of forecast uncertainty between brand and end consumer. Without better information — particularly sizing data — reducing production quantities risks stockouts, which damage revenue and brand equity more immediately than overproduction does.
Poor sizing generates waste through two mechanisms. First, brands must produce wider size distributions than their actual customer base requires, because they cannot predict how demand will distribute across sizes for a new collection or market — resulting in unsold inventory at the size tails. Second, poor fit drives returns: fit is the primary driver of online fashion returns in EU markets, and return rates reach the tens of percent in some online categories. Returned garments that cannot be resold at full price add to the overproduction waste stream.
Made-to-order structurally eliminates speculative inventory overproduction — no garment is cut without a confirmed order. However, it introduces a different operational challenge: customer willingness to accept a production lead time. In premium and luxury segments, 3–6 weeks is broadly accepted, particularly when the trade-off is communicated as personalisation and fit accuracy. In mass-market segments, the model requires a consumer education investment that most brands have not yet made. The economics are most favourable at high average selling prices, where per-unit overproduction cost is highest.
The EU Ecodesign for Sustainable Products Regulation (ESPR, 2024/1781) requires digital product passports for textiles, which will eventually make production volumes, material composition, and end-of-life data traceable at the garment level. While the regulation does not directly cap production quantities, it creates an audit trail that will make large-scale overproduction — and the destruction of unsold stock — regulatorily visible and potentially subject to future restriction. Brands with measurement-driven production planning already in place are building a compliance advantage ahead of that framework's full implementation.
Standard size data tells a brand that a customer bought a size M — but it cannot reveal whether that customer is 175 cm with a 98 cm chest, or 182 cm with a 92 cm chest. Both buyers might purchase an M, but they need different garments. Measurement portability replaces the size label with actual body measurements — 30 to 50 data points per buyer — that travel with the buyer across every purchase context. The production planning signal becomes a body-measurement distribution, not a size-label distribution. That is why it reduces overproduction structurally: the uncertainty it eliminates was never about quantity alone, but about which bodies the brand was actually serving.
Sources
Related concepts