For a growing part of the industry, fashion design now starts with a search bar. Software scours query logs, viral TikTok feeds and celebrity street style for the next thing shoppers want.
James Townsend is the co-founder and CEO of Pattern, a retail technology company. Pattern’s inventory optimisation solutions empower brands and retailers to buy smarter, allocate & replenish faster, and meet demand with precision. Image supplied
Orders go to networked factories overseas, which produce test batches of a few hundred units. If the clicks come, production scales. If they do not, the style disappears.
The fashion journalist Amy Odell has a name for what this process produces. She calls it ‘fashion slop’. This echoes ‘AI slop’, the low-grade, machine-made content clogging social feeds that is hard to miss.
Odell argues that algorithms can process data on what has already sold. But they cannot generate taste.
Quince makes the opposite case for the same technology. The direct-to-consumer brand also relies on machine learning and weekly forecasting. But it points them at durable essentials, such as a $50 cashmere crewneck, rather than at fleeting micro-trends.
In March, investors valued the high-quality fashion brand at $10.1bn.
Both approaches run on data and speed. Trend-chasing uses them to copy what shoppers already like. Quince uses them to strip waste out of making things people keep.
For retail executives and investors, the question is stark: is hyper-reactive algorithmic fashion the future of retail, or a capital-intensive dead end?
Algorithms promise to fix fashion’s 40% overstock problem
The case for the algorithmic model starts with money. For nearly a century, fashion retail ran on speculative bravado. Creative directors bet on a look six to nine months before it reached the shop floor. They committed large sums of working capital to seasonal collections, guided mainly by intuition, historical data and cultural precedent.
Those bets are expensive when they miss. In traditional fashion retail, roughly 40% of manufactured inventory ends up heavily discounted or incinerated as deadstock, because seasonal forecasting is imprecise.
For boards and private equity owners, removing that margin drag is the main prize in the algorithmic gold rush.
The algorithmic model promises to remove it by inverting the supply chain. Traditional retail relies on speculative push, building large inventories months in advance.
Algorithmic retail works on real-time pull. Platforms replace quarterly buying cycles with weekly demand sensing. They test designs in micro-batches of 200-800 units and scale production only once real clicks validate an item.
Speed pays, but an algorithm can only find the median
This cuts inventory lead times from three to six months down to two to four weeks. Capital turns over far faster. Reactive production also promises to meet smartphone-obsessed shoppers where their attention lives, while leaving far less stock unsold.
Edikted, a clothing brand popular with teenagers, shows what the model can deliver. It tracks search behaviour and celebrity outfits to produce about 300 styles a month. Its revenue passed $460m in 2025.
One of its founders has said the company works more like a data business than a traditional apparel maker.
Yet as the practice spreads across the sector, its structural weaknesses are becoming hard to ignore.
Odell’s critique goes to the heart of brand equity. Good taste is human precisely because it cannot be defined mathematically. It is a judgement made against the market. An algorithm can only calculate the median of what has already sold.
When every brand reads the same feeds, pricing power diminishes
Fashion’s most enduring commercial breakthroughs were acts of creative defiance that led the market rather than summarised it. Coco Chanel put women in jersey, a fabric then associated with men’s underwear.
Yves Saint Laurent gave them Le Smoking, his 1966 dinner suit for women. Alexander McQueen staged runway spectacles that unsettled the industry.
When design briefs come entirely from search logs and trend feeds, uniformity follows. Shoppers struggle to tell one brand’s output from another’s.
The commercial consequences are severe. When design costs next to nothing and garments are made to last weeks, pricing power vanishes. Brands lose the ability to defend their margins.
The result is a race to the bottom that pressures wages across the supply chain and sends low-quality garments to landfill.
Trend-chasers cannot beat Shein and Temu on cost or speed
Brands built on pure trend-chasing also face a daunting competitive problem. They are taking on Shein and Temu, multibillion-dollar platforms investing heavily in supply chain planning, automated manufacturing and proprietary demand sensing. Competing with them on unit cost and trend speed alone is a losing proposition for specialist brands.
That raises a hard question for retail leaders: what is the half-life of an algorithmic fashion brand?
Consider a company that relies only on scraping public trend feeds. If it owns no brand equity, material advantage or structural cost edge, its moat evaporates almost at once. As customer acquisition costs rise on digital platforms, that half-life grows shorter.
If pure trend-chasing leads to margin collapse and creative dead ends, where does a durable retail business come from? The answer lies in separating algorithmic supply chain efficiency from hyper-trend capture.
Quince points the same machinery at products people keep
Quince is one example of how. The company has crossed $1bn in annual revenue, and its $10.1bn valuation came with a $500mn round led by Iconiq Capital.
Quince runs what it calls a manufacturer-to-consumer (M2C) model. It strips out sourcing agents, distributors, and shops, so factories effectively act as the retailer. Quince provides the curation, quality, technology, logistics and brand.
Operationally, it looks much like its trend-chasing rivals. Machine learning and weekly product-level forecasting run inventory cycles of two to four weeks. Quince says this keeps overproduction below 5%, against an industry norm of about 40%.
The difference lies in what it makes: timeless, high-quality essentials. Its flagship $50 Mongolian cashmere crewneck shipped 400,000 units in 2021. Quince shows customers where the money goes: $25 on raw materials, $5 on factory labour, $10 on shipping and $10 to Quince as margin.
Let algorithms run logistics and leave taste to people
Quince backs these products with a 365-day return window. It competes on quality and trust rather than speed alone. In 2026, Time magazine named it one of the 10 most influential retail companies in the world. Its rise suggests that while shoppers live on their phones, many still pay for longevity, material value and discernment.
The lesson for fashion retailers is clear. Technology is essential for shrinking the 40% of stock that traditional forecasting leaves to be marked down or destroyed. But algorithms should optimise logistics and working capital. They should not replace human curation and design authority.
The businesses that survive the next decade will pair digital agility behind the scenes with products that have genuine longevity and a distinct identity. Shoppers’ attention will stay on their phones. The winners will give them a reason to keep what they buy.
