When we look to the future of fashion, the need for better fit is a foregone conclusion. From the brand’s perspective, fit is one of the primary culprits in industry-wide return rates, which makes providing customers better-fitting clothes a bottom-line imperative. From the shopper’s point of view, fit is a key criterion in brand loyalty: we buy more from a particular marque when we’re confident that what we order will flatter us, make us feel comfortable, or, for sports and athletic gear, perform to our expectations.
And with that level of pressure from both ends of the market, change is inevitable.
At an industry-wide level, the forces that have begun steering fashion towards a revolution in fit are just as stark. Pre-pandemic, eCommerce had already reached the majority tipping point in at least one major market, creating a clear demand for better matching shoppers with the right size – remotely. During the pandemic, and likely post-COVID, socially distanced shopping and contactless try-ons – if try-ons were offered at all – became the watchwords for physical retail.
The heightened level of importance placed on fit has seen an entirely new industry spring up around new methods of assessing, analysing, and improving it through technology. And as more brands take an inclusive approach to their sizing strategies, limiting the usefulness of historical data and population-level sizing surveys, a lot of technology investment has been concentrated on new ways of capturing body data at the individual customer level. The objective then is to make use of that more granular, demographic-specific body data in both shopper-facing experiences like smart size recommendations, and in internal tools that can incorporate insights gleaned from aggregate body data into future cycles of product design and development.
This is, of course, a positive thing. Over the last few years – and especially during COVID – fashion has been taking steps to address longstanding problems with fit by making wide use of accessible, turnkey consumer body measurement solutions.
But as the industry begins to emerge from the pandemic, the demands placed on those body data capture methods are about to increase beyond the levels of scalability and precision that brands’ current systems are capable of. And since so much of the future of fit relies on the output of those capture tools, we could soon see a disconnect between what the future of fit requires, and what current fit technology can provide.
How circumstances are stress-testing fit technology.
How do we see that demand evolving to where it starts to test, and even surpass, the capabilities of existing capture methods?
As the pandemic wanes, the expectation is that the appetite loungewear and casual wear that have dominated so many assortments will be replaced by consumers seeking more fitted garments. And while it is unlikely the sector will rebound to precisely the shape it was pre-COVID, where formalwear constituted the majority of fashion sales, lower volume will not ease the expectation for evening wear, office wear, and other fitted garments to fit right. In fact it’s likely to place even greater weight on the specificity of size ranges, the accuracy of size recommendations, and the consistency and precision of fit.
And if any product category – besides masks – can be said to have been a pandemic success story, it would be athletic apparel and footwear. COVID has made several generations more aware of their health than ever, in addition to emphasizing solo and team sports that can be played outdoors. Gym membership numbers may be rebounding, but from street hoops to trail running, shoppers are investing in shoes and apparel that perform elsewhere. And while not everyone who took up a new sport during the pandemic will stick with it, those who do will ascend the ladder – from casual to committed amateur, and potentially to professional – of what they expect from their sportswear both in terms of fit and technical performance, which are in many cases intertwined.
While these are currently behavioural trends, it’s easy to see how quickly they could translate into data problems. A new cohort of consumers will be seeking recommendations from activewear brands that transcend the usual sizing brackets, and that address their specific, individual needs. Soon, a runner who once would have been happy buying sneakers based on prior measurements ported from one brand to another will be seeking more granular insights into how a specific shoe is going to fit and how it’s going to perform. And at the same time, the brand selling that shoe could quickly discover that the broad-brush, historical body data it holds is not calibrated for the changing landscape of athletes.
As it stands, fit is set to become a more significant factor in purchasing decisions than perhaps ever before. And in a market where consumers are buying fewer products that they intend to keep for longer, and for which their expectations are at an all-time high, the already-high bar for fit technology is about to be raised to a level that one new entrant, NetVirta, is prepared to meet.
The precision problem.
When that bar is raised, where does it leave existing methods of capturing consumer body data? And what does it mean for the long-term viability of the data they produce?
The most obvious limitation for any size capture system is finding the right balance of precision, accessibility, and user experience. It’s no secret that extremely precise body scans can be obtained from dedicated capture hardware – this was, after all, how modern population-wide sizing surveys were conducted. But precision is only part of the future-facing picture; standalone body scanners can be costly and impractical to deploy at the store level, and while specialist retailers such as sportswear stores may wish to make that investment anyway, the return on that spend could be compromised by post-pandemic reductions in foot traffic and by the public, offline nature of the engagement. While foot scanning is only a partially intrusive process, full-body scanning requires greater intimacy, and is not something that shoppers will necessarily want to participate in even if they have taken the decision to return to in-store shopping, rather than buying online.
Crucially, the future of fit will require body data that has a lifespan beyond the point of capture – however conveniently it’s captured. After being anonymised, aggregated, and with regular updates ensured, body data should then be used to shape design and development in a way that improves fit for the target consumer with each iteration. This is the difference between static body data and an iterative loop of fit improvement, supported by not just an intuitive method of initial scanning, but by enterprise-wide data structures, archival-quality standards, and innovative solutions. By contrast, the pathway from dedicated scanning hardware to a universal view of body data is not guaranteed – in the same way that integrations to bridge that initial scan with the ongoing customer journey are likely to be bespoke rather than standardised.
Why not every smartphone scan is a solution.
The search for a solution that blends the accuracy of dedicated scanning hardware with universal accessibility, privacy, and usability has driven the explosion of body scanning at home, which many brands and retailers have deployed during COVID (as well as ahead of the pandemic. From the shopper’s end, this is typically accomplished using hardware they already have – a smartphone – to take several photographs, front and profile, which are then synthesised into a 3D model of the customer using machine learning.
That model is then used to match the shopper to their closest size within a brand’s catalogue, or across a multi-brand retailer’s inventory, giving the customer greater confidence in their buying decisions. Once separated from the customer’s identifying features – for privacy reasons – the model becomes part of a pool of captures that are aggregated and analysed to form a reservoir of body data intelligence for the brand.
But as you might imagine, brands have had to make compromises to obtain body scans from $500 smartphones rather than $20,000 standalone scanning hardware. And while high-end smartphones that incorporate LIDAR and depth-sensing camera setups can be considered a middle ground from a pure sensing input point of view, those hardware features remain confined to the most expensive consumer devices, meaning that they do not pass the universal accessibility test.
There are two ways that consumer body scanning has tried to overcome this input discrepancy: through innovations and improvements in computer vision, or by supplementing its scans with additional cross-brand size correlation.
The latter is what might be referred to as a brute force approach, since it relies on comparing sizes between brands the customer has shopped before to determine their fit, rather than adding additional precision to the underlying data. And as the market continues its push towards placing greater emphasis on true, technical fit across a range of applications (and especially as different brands approaches to sizing diverge in line with their inclusivity strategies) the brute force method could start to come apart at the seams.
Indeed, the limitations of this approach are already being demonstrated. While cross-brand correlation has been shown to increase conversions, it has had minimal impact on returns, suggesting that customers are buying with false confidence that the recommendations they are being provided are generated at the product level, rather than being averaged out by inference from one brand to the next.
The alternative, computer vision, is where fit technology stands the best chance of measuring up to the level of precision and usability the market could demand of it in the very near future. Like a lot of applications of computer algorithms, though, the computer vision models used to generate accurate fit data from a simple smartphone scan are both incredibly complex in their construction and equally variable in their results.
So far, that variability in the level of precision delivered by different computer vision algorithms has not been a major issue for fashion – primarily because consumer-directed body scanning is still an emerging market, and one primarily populated by relatively immature companies. But as we’ve established, circumstances are changing, and the future of fit technology is highly to rely on the mapping of customer body measurements to the dimensional data of each individual product with the greatest possible accuracy.
As a consequence, fit technology is about to enter a computer vision battleground, with body data solutions evaluated not just on the ease with which a customer can scan themselves and receive recommendations, but whether the precision of the results will deliver against consumers’ evolving expectations.
And this battle is one that NetVirta has been quietly preparing for more than a decade. Founded in 2013 by two MIT graduates, NetVirta’s innovative computer vision and body scanning technology has been widely-adopted and tested time and time again in the crucibles of the medical 3D scanning industry and professional athletics (including the NFL) where uncompromising precision is the baseline, not the exception.
Based on computer vision algo-fusion technology originally developed at MIT, the company’s scientific approach to fit has seen the precision it’s able to achieve from affordable smartphone scanning reach a tolerance of 0.5mm in its medical-facing app, and across those demanding applications in orthotics, protheses, and other regulated medical applications, NetVirta has racked up more than 375 million data points that have been used behind the scenes over the last eight years to both deliver absolute precision in fit for its customers, and to perfect its algorithm in the most demanding scenarios. This differs from other fit solutions which are trained primarily on narrow datasets before being extended to real-world applications. By contrast, over the last two years, NetVirta has been partnering with several leading apparel and footwear brands to convert its medical-facing app into a consumer-facing experience built for fashion, and built on real-world category-specific data, and this is now ready to roll out to consumers across the full spectrum of body, foot, and head scanning.
On top of the work the company has also put into ensuring that its holistic body data platform meets the usability and accessibility needs of both the near-term and more distant future, this commitment to precision is how NetVirta seeks to distinguish itself from other body scanning technologies that operate within much wider tolerances.
From this point of view, NetVirta is aiming to play ahead of the game. As fashion consumers recalibrate their expectations for fit technology, and come to demand detailed, product-level recommendations, that accuracy could be critical. And conversely, a lack of accuracy body scanning at this stage could serve to potentially undermine brands’ plans for the future – plans that include capitalising on the rapidly-growing market for personalisation and custom fit.
But whether or not a brand is targeting truly custom fit, or simply looking to meet the ever-raising bar of expectations for fit technology post-COVID, it’s becoming increasingly clear that next generation companies will need to offer next generation fit. And to deliver it they will require body data technology that comes without compromises – and scanning technology that’s both universally accessible and uniquely precise.
About our partner: NetVirta is a fast-growing, venture-backed, software start-up in the emerging field of 3D scanning, computer vision & graphics. Founded by two MIT graduates, and located in the heart of Boston, we offer a precision 3D body scanning mobile app and technology platform that enables brands to seamlessly offer their customer’s a personalized shopping experience, either through suggesting the best fitting apparel/footwear, or offering custom-fit products. NetVirta’s mission is to help brands and retailers usher in the next generation of customization and personalization during the buying experience.
