Start-ups are developing technology to create made-to-order products that reduce waste and increase personalisation. But, mainstream adoption is still some years away.
By guest author Maghan McDowell from Vogue Business
- On-demand manufacturing at scale can reduce overproduction while providing affordable personalisation to customers.
- Customisation is driven by customer-facing software that works with existing machines or supply chains, which potentially lowers barriers to adoption.
- Knitwear is a key area in which one-of-a-kind production is possible, but scaling is limited by the speed of the machines.
After figuring out how to make clothing faster and at lower prices, fashion is shifting its focus to creating more exclusive products at scale.
On-demand manufacturing, which allows brands to order as little as one piece at a time, aims to produce only what is likely to be sold. Since it can allow for an element of personalisation, items are a better fit for the customer, reducing overproduction and making them less likely to be discarded. “Made-for-me” production is trending, says NPD chief retail analyst Marshal Cohen. “Fashion is no longer about looking like one of a million but one in a million. Brands and retailers that are driving the market today are about innovation and personalisation.”
McKinsey has predicted that automation is key to powering made-to-order production cycles. Among the players are the spate of start-ups testing market appetite for items like USD 200 made-to-order sweaters and glasses moulded perfectly to the face for less than USD 600. And in 2017, Amazon received a patent for a system that produces clothing only after an order is placed.
But personalisation and automation at scale are notoriously hard to pull off. Zozosuit’s body scans failed to deliver flattering custom jeans, while Shoes of Prey found that customers didn’t necessarily aspire to design their own shoes.
The industry is still at least three years from on-demand manufacturing becoming mainstream, says Pano Anthos, founder and managing director of XRC Labs, which has invested in start-ups like on-demand knitwear company Nimbly and custom activewear platform Zielwear. For now, he says, startups, rather than large companies, will still lead innovation in on-demand manufacturing.
Focus on software
Topology offers custom eyewear made from a customer’s unique measurements starting at USD 550. Its augmented reality-enabled app measures and builds a 3D model of a customer’s face, previews how a pair of glasses will look and then converts the design into code for a computer numerical controlled (CNC) machine. Glasses are sculpted one at a time and delivered, with prescription lenses, in two weeks.
The key innovation is the software, not the machinery. Because Topology uses “commodity equipment” that makes everything from medical devices to aeroplane parts, the only limitation on scale is the number of machines the San Francisco-based company owns, says founder and CEO Eric Varady, a trained mechanical engineer. “That’s an easy problem to solve. People assume that we 3D-print our glasses, but we don’t.”
The focus on software lowers the barrier for companies who might assume that on-demand manufacturing requires overhauling the supply chain. Topology has increased its reach by adding retail partnerships with optical boutiques. “The first breakthrough success of applying digitalisation to fashion was to make the product cheaper and in record time,” says Varady. “The next stage of innovation will be using digital to make products better.”
Anomalie sells made-to-order, custom wedding dresses that usually cost less than USD 2000. Customers begin with an online dress builder, which has codified thousands of modularised variables. They then finalise their design with a stylist and send it to one of three factories in China.
Because the core of the dresses — the fabric, bodice construction and skirt lining — do not vastly differ, the customised elements are primarily added by hand at the end. Here, the innovation is in the design process, rather than automating construction. “The way we have thought about it is, how can tech help the process that is already there?” says Leslie Voorhees Means, co-founder and CEO of the San Francisco company.
Production and delivery take about four months. Wedding dresses are well-suited for this model, as customers are willing to wait, says Means. But, a higher order volume would shorten customer wait time because Anomalie could group similar styles and wouldn’t have to wait to run the factories until enough orders are placed.
“In a convenience and efficiency-driven industry, it may take a shift in consumer mindset to adjust to longer delivery times,” says The Future Laboratory senior creative researcher Rachael Stott, who has identified “demand-led design” as a microtrend. “If we educate the consumer on the ecological and environmental benefits of this approach, then it has the potential to be adopted.”
On-demand fast fashion
Social media-scanning artificial intelligence and decades-long relationships with 300 factories in China enable Choosy to produce small-batch fast fashion in as little as three weeks. The brand holds minimal inventory and uses only readily available repurposed fabric, which is eco-friendly and can shave over a month off of the production time.
After identifying trending styles on social media, Choosy’s New York office and Chinese mini-factory design and sample a new item. It tests the item’s popularity with an 80-piece run, and as soon as it starts moving — popular styles sell out on the same day — Choosy orders more units from whichever factory has capacity. “The ones that aren’t best-sellers end up selling out anyway because [80 units] is such a small run,” says co-founder and CEO Jessie Zeng. Choosy has received more than USD 10 million in venture funding.
A demand-led model has the benefit of opening up a dialogue between a company and its consumers, Stott says, allowing it to determine which styles are most commercially successful. But, while waste from overproduction is reduced, this model still creates clothing that is more likely to be discarded after a few wears, XRC’s Anthos says.
Knitwear on demand
Knitwear is particularly well-positioned to be made in small batches, Anthos says. 3D knitting machines can now make one-off items with minimal waste, whereas woven garments generally benefit from being created at scale.
Boston-based science-driven clothing company Ministry of Supply has tested made-to-order 3D-printed knitted blazers and dresses. Customers choose from multiple colours and style permutations. Orders are then routed to its production management system, and within two days, the garment is “printed” and sent to customers. Many do not realise the product is produced on-demand, says co-founder and President Gihan Amarasiriwardena.
Similarly, Variant Malibu has created a 3D customiser tool that allows knitwear designs to be created by the customer online. The tool converts the photorealistic renders into design files that go directly to its factory partner. After testing custom sweaters for Calamigos, the apparel line made for Calamigos Ranch, CEO Garrett Gerson white-labeled the technology so that fashion brands can offer knitwear customisation to customers. Variant also teaches factory partners how to produce on demand.
“In fashion, it has become scary to design something that is audacious because of the minimums. How nice would it be to order just enough and supplement inventory in real time?” Gerson says.
The manufacturers that Ministry of Supply works with can make up to 18 garments per machine per day, with both economic and physical limitations to scaling. “Low resolution” items like sweaters, which have fewer stitches per inch, are profitable. However, the time it takes to make a dense knit T-shirt wouldn’t pay off. And faster-moving needles would just overheat — the same challenges facing early microprocessors, Amarasiriwardena points out.
Anthos says that on-demand knitting could scale the same way servers did. Early efforts were focused on getting the machines to work faster; now, the emphasis is on getting more machines to work together simultaneously. “That is the future — you cannot use one machine and expect it to move 10 times faster.”