Though it’s rare to find visual search implemented on an ecommerce site (Zappos had it but took it down), visual search and image recognition technology has a lot of potential in online shopping. I’ve been asked in the past what I think they can do for ecommerce. This post is a brain dump of a bunch of ways I could see retailers benefiting from image recognition and visual search.
1. Retailers with large product catalogs
Like.com is an affiliate shopping site that aggregates fashion items from a large number of retailers, and is an example of how helpful a visual search tool can be as an alternative to other forms of filtered navigation. Clicking on any product’s Visual Search button re-sorts a page to show items that most resemble the item clicked. This helps customers find items they like faster. It can also improve conversion. Say a customer likes a handbag that appears on the first page of your “Handbags” category, but the price is too high for her. A simple visual search may expose hidden gems that fit her budget, but may have appeared on page 14 of your category results. EyeAlike is a tool already available for retailers who want to use visual search on their sites.
I blogged before about EyeBuyDirect’s Wall of Frame which has become a big hit with its customers. The Wall of Frame is a collection of photos customers submit of themselves using the eyewear retailer’s try-before-you-buy tool.
Imagine uploading a picture of yourself and allowing face recognition to filter the Wall of Frame to people with your face shape. You could find a pair that flatters your face without trying random styles on with the tool, then try them on your own photo after you found one that you liked on another’s photo. The less time customers have to spend searching for a good pair of glasses, the more likely they are to convert.
3. Apparel retailers
MyShape is a site that offers “custom shops” to members based on their style preferences and their actual body measurements. This takes the guesswork out of finding brands and styles that fit, because the whole catalog is filtered based on what will look good on you. You can fill in a full profile including all of your body measurements, but if you’re lazy you can just filter the shop by SHAPE (each letter applies to a different body shape). You could upload a photo of yourself dressed in black and have image recognition determine what shape you are.
4. Home hardware
Imagine you could locate replacement parts for appliances and home tools. Upload a photo of a drill and let the system detect the make and model, taking you to a page with the right accessories/bits for that tool.
5. Mobile phones
Like home hardware, you could upload a photo of your cellular phone and be matched with compatible accessories. (Think of a customer who buys a second-hand phone and doesn’t know the exact make or model).
Say you have to buy something for your mother in law. Snap a picture of her good china and you can find and purchase an item from that exact set.
Amazon’s already using image recognition in their mobile application, just snap a photo of any product at home, on the street or in brick-and-mortar stores with your device’s camera and match it to products on Amazon.com. There’s no reason other online retailers couldn’t benefit from a similar tool.
Truly the possibilities are exciting and endless, but we still don’t see image recognition used much in practice today. Perhaps in 5 years it will be standard across online stores?