A fashion week produces an enormous quantity of clothing and a very small quantity of information. Thirty-one shows and roughly 1,900 exits later, the number of looks that will genuinely alter what gets made, bought and copied is somewhere around eleven. Identifying which eleven is the entire job, and getting it wrong is how trend forecasting acquired its reputation.
How to tell a signal from a stunt
The test is not whether a look is striking. It is whether it is reproducible. A garment that requires a specific body, a specific fabric that only one mill makes, or forty minutes of dressing is a photograph, not a signal. The eleven exits we counted all shared one quality: a normally proportioned person could get into them unassisted and sit down.
Three of the eleven were coats and all three did the same thing — moved the shoulder seam outward while keeping the armhole high. That combination is difficult and expensive, and the fact that three unconnected houses arrived at it in the same season is the clearest signal of the week. Expect a compromised version of it in every mid-market coat by next autumn.
“Half the room photographs the loudest look. The buyers are all photographing the third one.”
A buyer for a European department store
What the room got wrong
The most photographed exit of the week — a spectacular, essentially unwearable construction on day four — will not be produced in any meaningful quantity, and everybody in the room knew it while applauding. Meanwhile a genuinely new trouser cut on day two, shown quietly in the middle of a long collection, went almost entirely unremarked and has already been ordered by three retailers.
Colour was the one place the week spoke with a single voice. Across thirty-one shows the palette narrowed sharply towards warm neutrals, and where colour appeared it was almost always a single saturated accent against them. That is a mill decision as much as a design one, and it means next autumn will look considerably more coherent than usual.
What we got wrong last time
It is worth checking this exercise against itself. We ran the same count two years ago and named nine exits; on review, five of them turned out to matter, three were commercially irrelevant, and one has become a genuine staple that we ranked near the bottom. The failure mode was consistent: we over-weighted novelty and under-weighted anything that looked like something already selling, which is precisely the bias the format encourages.
The correction we have applied since is to ask a duller question about every candidate exit — not whether it is new, but whether a version of it already exists in somebody's wardrobe that this improves on. The looks that survive tend to be small, specific upgrades to garments people already own. That is an unexciting thing to write about and it has been considerably more accurate.
We will publish the eleven in full at the end of the season, with the reasoning attached, so it can be checked against what actually reaches shops next year. That is a slightly uncomfortable commitment to make in print and it is the only way this kind of piece is worth anything. A prediction that is never scored is entertainment, and there is already a great deal of that in this part of the calendar.
None of this is a complaint about spectacle, which is what gets people to look. It is an argument for reading the middle of a collection, where the commercial decisions actually live, rather than the finale, where the photographs do.
