No Room for Virtual Fitting Rooms in Ecommerce
Don’t Expect Cyber Mannequins from Online Apparel Retailers Anytime Soon
Although I regularly skim through news and updates about startup companies, a recent story grabbed my attention because it involved ecommerce, robots, and the quest to get the perfect fitting outfit to online shoppers.
While online retail and the technology that powers this field have matured considerably, one complication that apparel retailers still tussle with is fit. It’s not easy to buy clothes that fit right if you can’t try them on first.
The fit conundrum isn’t new to online retailers (and it’s something that catalogers face as well), but – unlike other aspects of shopping online – it isn’t a problem that’s been widely fixed with more advanced technology.
Cyber Mannequins and My Virtual Model
Some of the My Virtual Mannequins from 2001
The virtual Erin sports a BCBG blouse and Mavi jeans.
A lot of companies have tried, and failed, to use technology to help shoppers determine if that pair of jeans or sexy top they’re contemplating will indeed fit. Land’s End was one of the first companies to dabble with virtual model technology on their website. As early as 2000, Land’s End had added a feature using My Virtual Model technology that created cyber mannequins to help customers truly size up clothing. The virtual modeling technology advanced, and Sears, Lane Bryant, and others followed suit.
Ten years later, My Virtual Model still exists and variations of its offerings have been put to use on many sites. However, in a basic search I wasn’t able to track down any sites that are currently using the customizable virtual models to let shoppers demo different apparel items. The MVM site itself does let you customize a model of yourself and play around a bit with some clothing items in their community site. Even so, I didn’t find the experience very compelling and the technology felt a little clunky.
Few clothing choices exist in the MVM community site for trying on, which I think is indicative of the expensive nature of the program – for clothing manufacturers and retailers to take advantage of the virtual models and cyber fits, every single garment has to be scanned by MVM technology. In a world where new styles come seasonly and designers often do limited cuts of products, the cost and time involved is incredibly prohibitive.
Augmented Reality and Fashionista
Tobi's Fashionista augmented reality.
About a year ago, ecommerce apparel companies started to look at another emerging technology: augmented reality. In the fall of 2009, clothing etailer Tobi launched Fashionista, a program that capitalizes on shoppers with webcams to allow them to virtually try out clothes. Fashionista required customers have a web camera, Flash, and the desire to move around in their living room until they hit the perfect spot where their full body image could be overlayed with outfits.
While perhaps amusing (I never tried it), Fashionista did little to help shoppers determine fit. Like MVM, Fashionista (a product of interactive marketing company Zugara) is an expensive feature to add. In addition to the cost of licensing an augmented reality technology, all product imagery has to be shot or appropriately edited for layering on top of a webcam image. The technology itself is also resource intensive. Interestingly enough, as I browsed through Tobi’s site, I no longer see reference to the Fashionista feature anywhere within their product pages.
Robots and Fits.Me
The Fits.Me Robotic Mannequin
The news that caught my attention last week was about Fits.me. The young company, currently in beta and based in Estonia, has developed robotic mannequins that can be customized by the shopper to match certain body measurements. Shoppers can then use the mannequin to “try on” various sizes of products, looking for the one that fits best. Fits.me is capitalizing on the fact that online apparel sales represent something like a $26billion market in the United States alone this year.
While it sounds interesting, I fail to see the point of the roboticized mannequins. Can’t the same thing be accomplished with an adjustable 3-D rendering? In order to use the technology, Fits.me requires samples of each clothing item (in each size). They then take up to 2,000 photos of each product. They digitalize the whole thing, set up a virtual fitting room, and charge retailers on a CPC basis. The whole thing takes 2-3 weeks of work.
The High Cost of Virtual Fitting Rooms
My take: The product won’t make it big. In order to justify the cost of virtualizing physical garments, you pretty much have to be a big company and be the manufacturer as well as the retailer. You have to have complete control of the timing and production process. Land’s End fits the bill here.
The thing is, many companies like this have already adopted fairly consistent, standardized sizing. And a lot of these companies have retail stores where you can just go and try the clothes on in person.
Still the Best Bet for Virtual Fitting
Some companies have tried to combat the problem of unknown fit by offering the stats of the model under the clothes. This is great – if you happen to be 5’10 with a 34B chest, 24 inch waist, and wear a size small and 26 in jeans. In my opinion, clear product imagery, detailed descriptions (that include fit notes), and a lenient return/exchange policy are still the most cost-effective ways for ecommerce apparel companies to help customers find the best fit.
If any non-manufacturing online apparel retailer had room to play with virtual sizing technology like Fits.me, I would point my finger at ShopBop. As a powerhouse of designer fashion and an arm of Amazon, the company might have the leveraging power to play around with expensive virtual fitting. With a free shipping and free returns policy, I’m sure any money saved on returns and exchanges would also be a boon.
Virtualized sizing and fit help is most needed by retailers who sell clothing from a variety of designers, many of whom often take liberties with size and offer limited runs of styles. Unfortunately, the more exclusive the clothing is, the fewer the number of cuts done and the more quickly retailers work to get the products out into the online world. This is why it’s tricky for many apparel etailers to get successful product review programs going – by the time people come to review the item, it’s already sold out.
As much fun as I think it would be to have a robot or virtualized avatar scour the web and return with a plentitude of products that fit me perfectly, I don’t see it happening in the near future. For now, determining online fit will remain a complication that ecommerce clothing companies must face.
I’m from Fits.me and wanted to thank you for this insightful article. You point out the most pressing troubles for any of the virtual fitting room technologies.
Some other interesting technologies in this area are being developed by http://www.mimicme.nl, Japanese Rakuten is using a variation that helps choose the right size, Tukatech’s Styku is nearly coming to the market, and Myshape.com gives great size and fit recommendations.
An international sizing standardization would help apparel e-commerce a lot – however, it should have a new system which, in addition to size, would register the customers body shape.
Currently 25-30% of clothes bought online are being returned, with 60% of the reason of returns the poor fit. 15% of the reason of returns is the “feel” of the fabric, 5% color misrepresentation. Return rates vary by category, with t-shirts lowest at about 8%.
And I really wanted to brag a bit: making 2000 photos of clothes in different size iterations is simply cheaper than 3D modeling. It takes about 2 hours. 2-3 weeks is the whole setup time, which mostly goes under logistics and integration. Per brand Fits.me digitizes only about 4-7 generic fit samples.
I really enjoy this type of articles as your feedback really helps us pinpoint what’s missing.
Great post!
Harovia
Hi Heikki, Thanks for your reply and for sharing the information on other technologies in the area. I agree that returns because of poor fit are still a huge consideration for ecommerce; cutting the number of returns by even a small percentage can represent a great cost savings. Good luck with the development of Fits.me. I look forward to seeing the progress.
I develop software for the fashion industry. Heikki was dead on – without body shape information, clothing size is practically unusable information, especially for women. Think about it: in every size you have the possibility of being large, medium or small chested; large, medium or small waisted and large, medium or small hipped. This results in over 27 different possible combinations per size and we haven’t even begun to look at the various body segment lengths (torso, rise, etc)! Clothing manufacturers create fit profiles and savy consumers have learned, via trial and error, which brands will offer more hip room, better cleavage coverage, etc….that is until the brand changes their fit profile which happens all too often. Now add to this complexity the fact that there are no algorithms that properly predict all types of fabric draping and stretch on a mannequin(This is also true for skin draping which is why it is easier for 3D annimation studios to create realistic looking annimated toys but not realistic humans). And, unlike a mannequin, customers have various conditions of muscle tone, which also affect fit, especially in swimwear and intimates.
It is for all these reasons that 3D modelling is not yet ready for prime time in fashion.
I think it is important to keep the fast improving capabilities of technology in mind. I am from MimicMe and we are able to digitise all garments based on existing product pictures, so instead of hours the production of 3D garments takes minutes. And actually I expect different companies will be able to automate this process including simulating different sizes within a couple of years.
In my opinion virtual fitting rooms should not focus on reducing reverse logistics only. In fact upselling items by visualising how to complete a look or providing inspiration will have a greater impact on the ROI.
Hi Eric. I am from My Virtual Model. Very interesting article. I agree that the problem of online fit is still unsolved. Partly because of the difficulty in getting accurate body measurements. It is very difficult to measure oneself at home, even with the help of a friend. But knowing what size to buy is not the only question shoppers have. At My Virtual Model we focus on outfitting – the “Look”. By mixing & matching tops and bottoms with shoes and accessories we not only create a very fun activity, but also empower people to see outfits on their body shape. With face mapping you can optionally add a photo of your face to your Virtual Model. This is similar to the augmented reality solutions like Fashionista in that you see “yourself” wearing the outfit, but the advantage with My Virtual Model is that you can create complete outfits with tops, bottoms, jackets, shoes, hand bags that actually fit on your body shape like they would in real life.
I also agree with the challenge of virtualizing physical garments. Similar to Mimicme we have brought down the cost of digitizing garments from weeks to only a few minutes. So stay tuned for a come back of virtual dressing rooms that provide fantastic outfitting capabilities.
I am a children’s boutique clothing designer and I am working on my website. I would love to have a virtual manniquin that customers can change and combine outfits on. If anyone can give me specifics on where to find such a program to intergrate into my website, please send me info. Thank you so much.