There have been rumblings around ecommerce for the past few years about visual search and what it can do. Millions of dollars have been invested in visual search companies around the globe. Headline after headline has come out touting the future of the web as being led by visual search.
And we’ve seen big players adopt visual search features. Google and Bing both put shoppable image-search capabilities out in the last two years, and Pinterest created a visual search function (“Lens”) that they licensed to Target.
Digiday posits that visual search is posed for a breakthrough.
The ecommerce world is finally starting to see what visual search can do for consumers and retailers. From mobile apps to websites, visual search AI is giving early adopters, from Amazon to Pinterest, an edge. Consumers are taking to the technology like a duck to water. Especially as more and more consumers shop from a mobile device, they are seeing the gaps in mobile ecommerce offerings — no wonder more companies are launching apps with a mobile camera search.
But what, exactly, does visual search AI bring to the table for ecommerce, and how can companies harness it to its fullest extent?
What Is Visual Search?
Visual search is a search process that relies on image input instead of text. If you’ve ever used the “reverse image search” feature on Google, you’ve done an image search. And while Google’s may be the most famous, there are many internet search engines out there that offer image search.
For ecommerce, visual search is a technology that “turns images of interest (from real life, the Internet, or social media screenshots) into shopping opportunities from the retailers’ collection.” Basically, sophisticated AI identifies attributes within an image, like color and shape, and uses that to figure out what is present in an image. This means that images can be matched with great accuracy, down to the length of a hemline or the metal of a necklace.
Visual search can be added as an additional feature to any website or mobile app, so a consumer who has an item of interest saved as an image can use that to search the inventory on a given ecommerce website.
Visual search’s AI can also be used outside of the search bar. It can be used in website features, such as similar-item recommendations, because it’s a superior way to index and tag items due to its granular specificity.
Perhaps most important for ecommerce, visual search is growing. eMarketer reported that monthly searches for Pinterest’s visual search tool alone jumped from 250 million in February 2017 to 600 million in February 2018. And that’s just what has been reported for one tool. If you have seen visual search pop up on popular apps and websites within the last year, you’ve seen proof that the use of the service is skyrocketing.
How Does Visual Search Affect Customers?
What visual search does as a technology is undeniably fascinating, but how it betters the customer experience is really where it makes a big play in ecommerce.
Visual search offers a superior form of discovery for customers looking for specific items. Rather than taking what is in their mind’s eye and “translating” that into text, users can skip the middleman. They see something they like, they take or save a picture of it, and then they use that picture to search.
The results that the customer gets through visual search are more accurate than the same results through normal text search. That’s because visual-search AI is matching images to other images. Take this search for “sleeveless pink gown” on the Bloomingdale’s website, which returns a variety of colors, patterns, and “sleeve” options:
And contrast it with this image-to-image search on Intu:
That’s faster, more accurate discovery — which means a customer is more likely to quickly find their moment of delight that triggers a purchase.
With visual-search AI, any image can be shoppable, including front-page images, product images on pages, images on a company blog, or anything else. Take this blog post from Atterley that allows you to “shop the edit” by hovering over any item featured in a photo:
This means that discovery is more organic because it is based on what a user sees and intuitively responds to. If they hit the front page and love the coat, they don’t have to navigate to “coats” and find that particular one — they can hover and shop the collection instantly.
Pulling in UGC
The numbers are in: 69% of young consumers are “looking for enhanced digital tools such as the ability to purchase directly via visual social platforms,” and 75% of consumers are inspired by visual media prior to making a purchase.
One of the big reasons why is that visual inspiration is everywhere. Social media is a prime example, especially the rise of influencers and user-generated content (UGC). Users can now find inspiration, save photos, pin them, screenshot, like, and save everywhere from YouTube to Instagram. Visuals are driving consumer desires more than ever before.
With a visual-search button on mobile or a website, you let shoppers pull in as much UGC as they want. If they see something they love, regardless of where it comes from, they can find out if you have anything like it.
Once a customer has clicked on an item they like, you have all you need to give them a personalized, robust set of recommendations. How? Visual search AI can pull attributes from any item that a customer has clicked on and provide product recommendations on that product page.
This is great for sold-out pages, comparison shopping, and presenting the customer with exciting options that keep them on your website. Instead of getting frustrated trying to sort through hundreds of items to find a few to choose between, you give them that spread right away. Check out Atterley’s full lineup of checked coats from a product page for buffalo-plaid coats:
Transforming ecommerce marketplaces
If you are a vendor that sells many brands, rather than just your own stock, automatic attribute tagging and visual search is critical. If you rely on subpar product descriptions from third-party vendors, you won’t be able to offer the search granularity that consumers want. They’ll get frustrated when they can’t find what they want and will search elsewhere.
With visual search, you can easily match customers’ queries even without having item attributes you need to build a truly accurate text search. With visual search AI, you can build that accuracy by running any product image through the tech and getting back the needed tags.
There’s growing evidence that, with increased use, visual search helps boost conversions. Search Engine Journal recently reported that more visual searches than ever are being offered up by Google, and retailers are seeing their visual searches grow as well.
As retailers have started to get enough data to calculate changes in user behavior around visual-search AI, the results have been strong. Boohoo, who works with Syte to implement visual search on their website, reports, “We have seen that visitors who engage with ‘View Similar’ on the PDP have a conversion rate over 100% higher.”
Who Is Leading the Pack?
Companies that have successfully implemented visual search include retailers across industries. In 2017, Gartner reported that 8% of retail brands had visual search for ecommerce, but it’s not hard to see that is growing, from Google’s and Bing’s shoppable options to Target. Here are a few others that are leading the pack, and what they’re doing right.
Boohoo has seen a big boost on their ecommerce site when users take advantage of visual-search AI. They report: “We have seen that visitors who engage with ‘View Similar’ on the PDP have a conversion rate over 100% higher than those who do not and the AOV is 12% higher. Lastly, pages per session are over 135% higher.”
Take their selection of similar items from a plum roll-neck tunic turtleneck:
They’re providing similar items, similarly styled, with price and color options easily available. If a user sees something they like, they can hit “quick buy”:
And instantly purchase without even navigating to a different page:
It’s the easiest continuation of a customer journey on a product page yet.
Home Depot’s mobile app and its visual AI has really hit home with their demographic. That’s probably because home-decor projects are difficult, multistep processes that rely heavily on visuals, both for inspiration and as test samples. With the Home Depot app, users can easily find items similar to their inspiration:
They can even check out how paint samples “look” on their wall. In-store associates and shoppers alike use the app to find the perfect items to finish off a DIY project.
Visual-search AI has revolutionized the back-end workings of Intu, especially because it is a company that indexes other vendors’ products and often doesn’t have robust product information to accurately tag them. They say:
“Organizing data and tagging items for search is particularly hard when you’re an aggregator, like us, because we don’t control any of our data. We don’t control the imagery — we receive data feeds from our retailers and often that data isn’t good enough to be able to create any kind of discoverability on our site for consumers.”
They continue: “[Now, on our site] you can scan through pages and find a style that you like, and all of a sudden, similar styles are there for you immediately, and Syte works particularly well because of the breadth and depth of product that we have.”
For Pinterest, a company so steeped in visuals, it’s no surprise that their users would respond favorably to their visual AI that renders any image shoppable. They say that there are now “more than 600 million visual searches every month across Lens, our browser extension and the visual search tool inside Pins.”
Further, they report that 90% of Pinners will use the platform to make purchase decisions, and 70% use Pinterest to find new products. That’s a strong indication that visuals are prompting spending and that Pinterest will continue to use Lens and their visual search. Pinterest responded directly to consumer need and existing user behavior on mobile and on the web, and that’s given them a huge opportunity to build new modes of discovery for users.
Implement visual search today
If you are a retailer interested in putting visual search on your website or using any other visual-search AI features you’ve read about, contact us today.
Today, we’re happy to share that Syte raised $21.5 million in Series B Funding. Viola Ventures lead the investment, alongside high-profile investors Storm Ventures, Commerce Ventures, Axess Ventures, and Lyra Ventures. The total funding to date is $30 million.