Online fashion sizing fails because measurements are never stored. Fix the infrastructure, fix the returns.
TL;DR — Online fashion sizing causes 30–40% return rates not because buyers choose wrong, but because body measurements are never stored or carried across brands. Size charts, fit quizzes, and recommendation engines all treat the symptom. The only structural fix is a persistent, portable measurement profile that replaces size labels entirely.
Online fashion sizing is broken at a structural level. Return rates of 30–40% industry-wide — with fit cited as the primary driver in over half of cases according to Barclaycard and ASOS annual reporting — are not a consumer behaviour problem. Every brand, from Zara to luxury maisons, faces the same infrastructure failure: the buyer's body measurements are unknown at the moment of purchase and discarded immediately after.
Online fashion sizing fails because it relies on size labels rather than measurements. A size label is a brand-specific shorthand — it encodes one brand's interpretation of a body type into a single character. When a buyer moves to a different brand, that character means something different. The label transfers; the information it represents does not.
Research published in the International Journal of Fashion Design, Technology and Education found that chest measurement ranges for 'size M' across major European retailers varied by as much as 8cm — a full size band — within the same nominal label. A buyer who correctly identifies as a size M in one brand can purchase a garment that is effectively a size S or size L in another.
Definition
Ease allowance
A garment's finished chest measurement is always larger than the body measurement it targets — the extra room is called ease, and it varies by silhouette. A slim-fit size M and a regular-fit size M share the same body measurement range but may differ by 6–10cm in finished chest width.
Definition
Cut variation within a brand
The same brand's size M in a classic oxford shirt is patterned differently from their size M in a sport shirt. A brand-wide size chart presents an average that is correct for no specific product.
Definition
Shrinkage and care
Cotton and wool garments may shrink 3–5% after a single standard wash cycle. Size charts describe garments as manufactured, not as they exist in use.
Definition
Inter-brand non-transferability
A buyer who has verified their size in Brand A has acquired a brand-specific data point, not a portable measurement. McKinsey's State of Fashion research identified this as a primary driver of first-order returns in cross-brand shopping journeys.
The root cause of the online fashion sizing problem is that body measurements are captured by no one, stored by no one, and carried nowhere. Every purchase begins from zero measurement knowledge.
In practice, when measuring a shirt fit — for example, comparing a 42cm shoulder width against a garment's shoulder seam specification — the result is immediately useful and then immediately discarded. The next purchase at a different brand starts the same guessing process from scratch.
Fit recommendation engines — tools like True Fit, Fit Analytics, and brand-native quizzes — reduce the approximation error of size selection. They do not solve the underlying sizing problem because they operate on size labels, not on body-to-garment measurement comparison. ASOS data suggests recommendation adoption reduces return rates by 5–8 percentage points — a meaningful improvement — but the fundamental gap remains: the buyer's actual body measurements are not part of the transaction.
Definition
Measurement portability
The capacity of a buyer's verified body measurements to persist across brands, platforms, and purchases — replacing size-label estimation with direct body-to-garment comparison at the point of purchase. Analogous to how a passport carries verified identity data across jurisdictions, a measurement profile carries verified body data across retailers.
With portable measurements, the transaction changes entirely. Instead of asking 'what size am I in this brand?', the system asks 'does this garment's shoulder measurement match this buyer's shoulder width?' The answer is deterministic rather than probabilistic. The fit is specified before the garment ships.
Size Passport addresses the structural problem directly: it captures body measurements precisely, stores them in a portable profile, and applies them to every subsequent garment specification. The profile is created once — through a structured measurement session or a tailoring appointment — and carries forward across every Bespoke purchase. There is no size label in the transaction at any point.
Because size guidance improvements — better photography, expanded size charts, fit quizzes, recommendation engines — all operate on the same flawed input: the size label. They improve the accuracy of mapping one label to another label, but none of them capture or store the buyer's actual body measurements.
It affects luxury brands more severely in financial terms. A fast-fashion return on a €30 garment may cost €6–9 in processing. A luxury return on a €1,500 jacket costs €300–450 in processing and often cannot be re-sold as new, requiring markdown to outlet.
Virtual try-on tools address the visualisation gap but not the measurement gap. A realistic render of a garment on an avatar does not confirm that the garment's shoulder seam will fall at the correct point on this specific buyer's shoulder. Visualisation and measurement are different problems.
For shirts and tailored jackets: chest circumference, waist circumference, hip circumference, shoulder width (seam to seam), sleeve length, neck circumference, and body length. For trousers: waist, hip, inseam, outseam, and thigh circumference. These 12–15 measurements resolve the majority of fit decisions for structured garments.
Several structural incentives work against a solution. Returns processing is predictable — brands have built logistics infrastructure around it. Sharing measurement data across brands would require interoperability standards that competing retailers have little incentive to establish unilaterally. And the sizing problem's cost is diffuse: distributed across logistics, inventory, and customer lifetime value rather than appearing as a single visible line item.
Sources
Related concepts