The Trillion Dollar Potential of Visual Search All Boils Down to Getting the Full Picture

Published on July 26, 2019

The digital transformation movement continues to drive customer expectations to new and seemingly insurmountable heights. In an effort to stay competitive, brands often scramble to incorporate the latest and greatest technology into their customer journey to somewhat diminishing returns.

A couple of years back, numerous software companies clamored to introduce “artificial intelligence” into their solutions as a response to a  growing trend that promised exponential ROI. However, many of the products that resulted couldn’t truly deliver the capabilities of other fully realized solutions and did very little to help the customer journey for those over-zealous companies.

One can only imagine the wasted budgets and lost time.

Visual search is quickly becoming the latest technology to be approached haphazardly by retail businesses looking to stay competitive. With major players like Amazon investing in the technology, it’s only a matter of time before visual search becomes the standard and naturally, retailers are hustling to incorporate it into their online experience. However, like so many game changing technologies before it, not all visual search is created equal.

Accuracy is Key

It’s not enough to have visual search on a retail site, it must be accurate in order to be effective. If it’s not, a retailer stands to lose thousands, if not millions in potential sales.

“The way we look at it, visual search is not worth having if the probability of getting accurate results is like flipping a coin,” says Lihi Pinto Fryman, the CMO of Syte. “The value of visual search lies in its ability to give shoppers an experience where they are understood.”

Syte is a startup out of Tel Aviv that is developing visual search technology for major brands such as Tommy Hilfiger, Forever 21, and Marks and Spencer. Testing by both Microsoft and Samsung has revealed that Syte’s image search capabilities generate 95 percent accuracy, while the nearest competitor could only yield 50 percent accuracy – a distinction that is causing Syte to dominate the market.

One of the major ways Syte has improved the accuracy of its feature, is by allowing customers to shop every item in the picture they snap, rather than the item the technology “guesses” what the user is interested in.

Getting the Full Picture. Literally.

According to Pinto Fryman, “If a user uploads an image of a full outfit looking for the earrings and gets results only of the pants, trust in the technology is lost. By powering visual search that exceeds expectations for accuracy, speed, and user experience, our goal is to instil that trust in users that there is technology out there that can help you discover any product that inspires you.”

As seen in this demo of Syte’s technology, customers can snap a picture of an outfit and be served multiple options to search by clothing type (example, the jacket, the belt, the pants, etc). The results populate quickly and are visible very similar to the picture. Many competitor technology only offers results for one article of clothing, missing numerous opportunities for increased sales volume.

And this technology can also maximize social influencer marketing.

Say a social influencer posts about a certain blouse. Her Instagram would link to that specific product on the retailer’s page and, assuming a visitor wants that blouse, she’ll buy it. But what if she likes the earrings, shorts, sandals and handbag that are not being promoted by the Instagrammer? That’s where visual AI comes in. One picture snap of the whole outfit and the customer can search for the items that give her the complete look (and maximize the ROI for the retailer).

A Picture is Worth a Thousand Words

Syte’s technology enables multi object detection, automatic gender detection, and there is absolutely no manual work involved. The company’s proprietary algorithm has been trained to recognize the most minute details in fashion and home decor and retrieve the most visually similar item available within the inventory, without any manual interference.

Syte automatically breaks apart each component of an image, understanding the most minute details of a garment and retrieving the most visually similar product available within the retailer’s inventory.

As explained by Susan Aubrey-Cound, Syte’s Business Development Director: “Images are processed 60,000 times faster by the human brain than words or text.” This noticeable lag in processing time has elevated the impact of images in social media and given it a clear advantage over text. It has also fundamentally reshaped the acquisition funnel of new customers.

The Results

A recent Gartner study found that early adopters who build in visual search will see digital commerce revenue increase by 30 percent and Syte’s results certainly support that stat.

BooHoo, a UK ecommerce retailer considered in some circles a potential challenger to ASOS, saw its overall page visits increase by 125 percent for those using Camera Search function (which in turn draws on Syte’s image search capability) and also enjoyed an 85 percent higher conversion rate for those using Camera Search, as compared to those who did not.

Pretty Little Thing also put Syte’s technology to work. Though the results were somewhat dependent on region, the retailer saw conversion rates increase between 2.0 percent and 11.7 percent.

The Future of Digital Search

In the end, the future belongs to those who can deliver results faster, and more effectively. While Syte’s technology has proven itself faster than its competitors, and more accurate as well, the potential represented by delivering the best in visual search will ensure this highly competitive market continues to innovate at an exponential rate.

Visual search will prove to be a dynamic market, full of new entrants eager to take their own shot at the brass ring of market share and entrenched presences just as eager to hold the ground they’ve already taken. No matter how the ensuing battle for the market goes, however, one thing is clear: seeing the full picture with 20/20 accuracy will pay dividends for retailers.


Molly St. Louis is a News Columnist at Grit Daily.

Read more

More GD News