In a recent side project, I was able to use some emerging artificial intelligence capabilities to distinguish with a high degree of accuracy the significant items in a vintage photograph. I was able to create a title by comparing the keywords dissected from the photograph and ranking them as search terms in Google. Not only was I able to get a title for my eBay listings, but also a keyword rich title that improved the search results and click-through rates. It also is able to answer all of the item specifics questions for the vintage photograph category. This allows us to create a complete listing for our vintage photographs as quick as we could scan them. Paired with our high-speed scanner, we are able to complete 30 photographs ready to be listed on eBay within one minute. The rapid advancement in artificial intelligence and deep machine learning made it possible.
Not far in the future you will use the same technologies to create eBay listings with nothing more than a photograph. This will cure two current problems for eBay. First, it will allow sellers to list items much easier. Imagine walking through your house taking photographs of things you no longer use with your smartphone. Having those items identified and compared against previously sold items on eBay instantaneously and creating a listing. Secondly, it allows eBay to easily create a product catalog and a single listing for the same product. No longer will we have 457 listings for Super Mario Brothers, but one listing with 457 sellers competing for the buyer.
Some will say this is no different than Amazon. But, it is because you did it all with a single photograph. No searching to make sure that the item is in the correct place. No more multiple listings for the same item. If you have a blue and green plaid Ralph Lauren shirt, it'll find all of the other plaid blue and green Ralph Lauren shirts of the same exact pattern and place them in a single category arranged by size and condition.
Now that we have the catalog sorted out, let's take this a step further. Imagine you are a hipster sitting on the subway. You see a shirt that you really like on a person standing across the aisle. You snap a picture with your eBay app. The photograph is inspected for the key elements and determines that the shirt is a vintage 1996 Dallas Cowboys Logo Athletic T-shirt. The app then returns the listing for that exact shirt; however, it also shows you shirts that people who liked this shirt also liked or purchased.
Perhaps you want to find replacements for your grandmother's worn flatware, but you don't know the pattern. You simply take a quick photo with your smartphone and eBay will find all the flatware in that pattern for you, or if there is none currently listed it will send you a push notification as soon as it is listed.
This is why I think eBay should skip the idea of using a UPC as the key to their current process of creating a catalog of products. Product identifiers like UPC and ISBN are hard to find for used products and do not exist for products made before the 1970's. If eBay wants to remain a leading ecommerce platform, it needs to lead the way with something innovative. Being a leader with this type of technology would open the door to many different business opportunities. With all the photos eBay has for products sold on its site over the last 20 years, it could have a viable product in 18 months.
That’s some serious technology to think about. Or, maybe some bytes to bite into the future.
No comments:
Post a Comment