If a picture is worth 1,000 words, how much more valuable is a reverse image search than a text search?
Google Image Search was created in the wake of the 2000 Grammys after a rash of search queries for a photo of Jennifer Lopez led Google to realize people wanted search engines to find the exact image they had in mind, not an approximation.
Image search has come a long way in the 21 years since then, partly thanks to the improvement of content-based image retrieval (CBIR) technology, which allows for image-to-image search, and is having a huge impact on eCommerce. The ability to find products via a visual search is hugely attractive for consumers, which makes sense seeing as about 90% of information transmitted to the human brain is visual.
More than 85% of respondents to a Pinterest survey said visual information is more important than text when shopping online for clothes and furniture, while in the same survey 61% of consumers reported that visual search elevated their experience while in-store browsing.
More and more customers are turning to visual searches for their online shopping needs, so how can you use this trend to your eCommerce store’s advantage?
First things first, it helps to have a rough understanding of how visual search technology works in general before we dive into how it can work for your eCommerce business.
Just like at a crime scene, fingerprints are crucial to visual searches. However, instead of using ink to make a copy of a suspect’s fingerprints, visual searches use pixels to create a virtual fingerprint for images. As the smallest unit of a digital image, pixels are the ideal element on which to base an image’s fingerprint, as the odds of two images having the exact same pixel fingerprint are astronomical.
To create this fingerprint, different elements of an image are measured such as color, textures, gradients, shapes, and more, depending on the visual search engine’s specific algorithm. These elements of the image are then translated and stored in a database as letters and numbers, making them easy to compare against each other.
So whenever an image is uploaded to a visual search engine, first the fingerprint is taken and broken down into letters and numbers, then compared against the visual search engine’s existing database to find exact matches or similar pictures. As you can probably guess, this means visual search engines need huge amounts of data to be effective. Pioneering visual search engine TinEye compares image searches to almost 50 billion images in its database, for example.
Now for some inspiration, let’s take a look at how visual search is used to sell everything from couches to dungarees online.
The IKEA Place app allows users to image search and to leverage AR technology to see how furniture looks in their home.
The start-up offered a pretty simple app that took some tech wizardry to pull off: point your phone camera at a piece of furniture, and it will search for similar items for sale.
IKEA decided they wanted this functionality on their app to make it that much easier for customers to discover IKEA products.
“To make AR work, that’s where you really need tech like visual search,” said GrokStyle co-founder Kavita Bala. “It lets you find things, cool designs and furniture, all in situ and visualize it in place.”
GrokStyle has since been swallowed up by Facebook as part of the social media giant’s foray into computer vision, so you know they were on to something good.
Fast fashion eCommerce and visual search are a match made in heaven and British fashion and cosmetic retailer Asos has gone all-in on the technology.
Style Match is a feature of the Asos app that allows customers to take a photo of an item of clothing, and then the app will “help you find the product in the picture, or recommend something similar” according to Asos. With some 5,000 items added to the Asos catalog each week, the visual search functionality is a much more practical option for shoppers rather than combing through the site or using keyword searches.
Style Match received rave reviews when it first rolled out and has stayed popular since, judging by Asos’ continued revenue increases amid the pandemic.
“Overall, the feature's great if you want to emulate the style of your idols or use a photo for inspiration. It's also fab for finding the exact items you want quickly, if it's something that ASOS sell,” wrote “Glamour” magazine.
Of course, the godfather of online search has gotten involved in visual search eCommerce, and in a big way too.
Using Google Lens image recognition software, Google is enabling shopping from within Google Photos. If you take a photo of a product, Google Photos will suggest searching for it with Google Lens, essentially allowing you to shop in-app.
Of course, Google isn’t selling these products itself. Instead, online retailers can list their wares on Google, allowing their items to show up in regular text-based Google searches, and in Google Lens recommendations.
Seeing is believing, so let’s take a look at what visual search can do for your eCommerce business.
A key part of an effective omnichannel strategy is providing a seamless online-to-offline (O2O) shopping experience, and visual search is one way to meld the real and virtual worlds.
The IKEA, Asos, and Google examples all show how the O2O experience can be enhanced using visual search technology. If, for example, you visit a neighbor and see he has a lamp that is simply to die for, all you have to do is take a photo using the IKEA Place app, and voila, options for buying it will be presented. The customer demand was generated offline, the transition to online was seamless, and the brand is now in a great position to satisfy that demand.
SmartOSC has been preaching the benefits of offering a solid O2O shopping experience for a while now, but in case you needed reminding, try this statistic on for size:
“Roughly 81% of all US adults (90% of internet users) conduct online research before buying things locally,” according to Search Engine Land.
If you make that research as easy to perform as taking a photo, you’ll be doing your customers a big favor.
It should be easier to close sales with customers who use visual search.
A customer who uses visual search tends to know what they want. Think of it as the equivalent of walking into a store in the 1980s with a newspaper or magazine cutting; these are customers with a strong intent to buy. So straight off the bat, you should have an advantage when it comes to conversions, especially if you can get the physical product in front of them quickly.
Of course, some people use visual search to find items similar to the one in their picture or they just use it as a reference, but again, you should have a good chance of converting these folks into sales if you can provide them with a relevant recommendation.
Plenty of companies of all stripes have used visual search to boost their conversions, including Central European footwear retailer CCC Group, which improved its conversion rate by four times compared to keyword search. Similarly, fashion retailer BooHoo “reported an 85% conversion rate for customers that used Camera Search versus those who did not” according to Forbes.
Google can be a cruel mistress and getting a lot of attention for your brand via text-based search isn’t always easy, but visual search can help you cut through the noise. Fewer brands than you’d think are taking advantage of the technology. Trust me, it wasn’t easy finding many companies outside of the fashion industry to use as examples above!
On the other hand, uptake among consumers has been impressively fast. Within just a year of launching, Pinterest’s Lens search engine had received “more than 600 million visual searches every month”, which goes to show the demand is out there.
Beauty is in the eye of the beholder, so let’s take a warts-and-all look at visual search to see if it’s right for your business.
While the accuracy of visual search results has increased significantly over the years, the truth is that it’s still in its infancy, especially compared to traditional text searches.
The need for complex AI programs and machine learning can also put knowhow and cost barriers between some businesses and visual searches.
Google has an advanced “Did you mean?” algorithm for when users make typos with text searches that can make an educated guess as to what they were looking for. But with visual searches, this functionality is nowhere near as advanced.
If you take a bad photo your image search will be less accurate, and similarly, it can be hard to specify that you’re looking for a particular object within an image, instead of the image as a whole.
Visual search works best when you have something visually appealing to sell, obviously, and when the aesthetics of what you’re selling are important.
This explains why the fashion industry has been most heavily involved in visual search, whereas it’s hard to see the value of a SaaS company investing a lot of resources in image search marketing.
Text-based search is still king, but visual search is here to stay. Research has found that 62% of Millennials want to visually search over using any other new technology, showing the high potential it has to drive eCommerce sales, now and in the future.
To find out how SmartOSC can help grow your business, get in touch with one of our eCommerce experts now.