New technology shops photos for comparable outfits
Milwaukee Journal Sentinel
Published 5:48 p.m. UTC Jul 26, 2018
It might be a dress worn by a friend in a photo on Facebook, a blazer on a colleague at work or a blouse shown on Instagram … You want it. Now.
With visual search technology, driven by artificial intelligence and deep learning, shoppers can see now and buy (at least something similar) right now. The apps put out by retailers and technology firms allow customers to upload inspiration photos and be shown fashions similar to those in the image that are available to buy.
Joy Tang calls it “frictionless shopping.”
Tang, the CEO of the Madison-based provider of visual search technology Markable, wants to make the fashion industry more efficient. It’s no surprise — Tang used to be a high-frequency financial trader.
She grew frustrated by wanting an outfit she saw on Instagram and not being able to find a similar look.
“Sometimes the blogger may tag the brand link,” Tang said. “But if you ask about a product that’s two or three months old, the link is dead. Fashion products go out every six to nine weeks. From a consumer perspective, you don’t need the exact one — sometimes you want something to be a little different.”
Now, when Tang sees an outfit she likes, she can just use Markable’s technology to find items to buy.
The software put in place by Markable and its competitors are like advanced versions of a reverse image search on Google. The technology goes far beyond that basic search to find the actual items — hats, dresses, shoes — a person in a photo may be wearing.
Already, Markable’s technology can find similar fashions to what’s shown in any photo. The programming identifies what a person is wearing — a jacket, skirt and handbag. Then it determines the garments’ attributes such as color, pattern and fit. The information from the inspiration photo is then cross-referenced with the clothing sold by Markable’s customers to give the user suggestions for similar clothes.
The chances of Markable’s technology finding the product you searched in its top two results is close to 80 percent, Tang said. Markable has around 800 brands in its database. Users report receiving relevant search results more than 90 percent of the time, Tang said.
Deborah Weinswig, founder and CEO of the retail think tank Coresight Research, said the technology and its function could evolve into a “Shazam for fashion.”
Just like a click on the Shazam app will identify a song playing on the radio, the emerging technology can find clothing spotted on an Instagram post or a stranger walking down the street. It’ll even provide a link to buy it.
“Consumers of the future are likely to be taking images of friends or colleagues or even strangers in the street and running those through visual search engines to find what they are wearing,” Weinswig said.
Tapping into a global industry
The underlying motivation for retailers is to make it easier for customers to find items to buy.
One visual search company, Switzerland-based Fashwell, said conversions are up 10 percent with a 5 percent increase in average order value at retailers that use the technology.
Text searches on a retailers’ app find a product the customer wants less than 50 percent of the time, according to Fashwell’s research.
Markable sells its visual search technology to retailers to use on their platforms. Most of their work is under non-disclosure agreements, but Tang said three of their clients are e-commerce shops with more than $1 billion in annual sales and another is a social media giant in China.
Akira, a women’s clothing retailer, started working with Markable to use visual search on its website in November 2017.
“Having the ability to match a single dress that a customer is interested in, with the click of a button, to all similar style dresses that we carry on site, is phenomenal,” said Akira Marketing Manager Sara Berke. “When— we launched, none of our competitors offered this type of technology, and still to this day are lacking. It gives us an edge.”
Akira is redeveloping the visual search technology on its site now that will launch next month.
Markable is about to release an extension to use in Google’s Chrome browser and developing a consumer-facing app, Tang said. The 20-member team, split between Madison and New York City, is getting ready to add a few more employees in the coming year and open an office in Tang’s native China.
Markable is projecting taking in $7 million this year from its two dozen clients, Tang said. She wants the company to grow three or four times that in the next year.
“I will hope for $30 million (in revenue) next year,” Tang said.
Markable also is closing in on a multi-million-dollar fundraising round, Tang said. Markable raised $1.9 million in 2016, according to filings with the Securities and Exchange Commission. The company, which started in 2016, pivoted last year from a consumer- to business-facing model.
ARCHIVE: Markable Inc. raises $1.9 million in equity funding
Many retailers are entering the game, including Menomonee-Falls based Kohl’s Corp. On the store’s app, customers can upload any photo to see similar products available at Kohl’s.
Pinterest added a visual search option to its platform last year where users can upload images of fashion and design inspiration seen offline to get recommendations online. Users perform more than 600 million visual searches each month, according to Pinterest.
“So far, a lot of discussion on visual search has focused around identifying products shown online, and much of this has been driven by image-focused social-media firms seeking ways to monetize their platforms,” Weinswig said. “The benefit to consumers seems to lie as much in discovering products in the physical world as online.”
Sarah Hauer can be reached at [email protected] or on Instagram @HauerSarah and Twitter @SarahHauer.