Amazon.com has started generating computer-generated product photos for some categories of their e-commerce site, but it’s largely gone unnoticed until now. We first noticed this a few months ago, but just figured it was for just for a few products, or part of some temporary A/B test. The fact that it’s still prevalent on the site indicates that the images had some sort of lift on user conversion, as Amazon doesn’t do anything without data to back it up. You can see them as typically the last image in the carousel of product photos, like this one for a book:
But it’s not just books, here you can see it in action for jewelry:
As well as beauty:
And even toys:
Based on some comments on Amazon’s forum, these:
“Scale images (product images shown with a silhouette of a person to convey relative size) are added to select ASINS at Amazon’s discretion, rather than at the request of sellers”
The images seem to be primarily focused on providing a sense of scale to the products in a more visual manner than just the actual dimensions. It’s somewhat surprising, though, as there are long-standing requirements for sellers not to use mannequins in their product photos.
But what if the next generation of these photos doesn’t just convey the size of the product? Imagine not a gray silhouette, but a photo of an actual person holding the book — colorful, realistic, and personalized to your demographics. Normally, this would have to be something that would either require a professional photo shoot costing hundreds of dollars an hour, or outsourcing to a user-generated content marketplace — with variable quality and long delays. Generating this instead via software yields both low cost and immediate results.
Amazon’s already got expertise in computer vision, facial recognition, and VR; it wouldn’t be difficult to put together something more realistic that would allow products in specific categories (books, DVDs, electronics, and boxes with standard dimensions) to be showcased with avatars designed to get you to click that buy button.
Does any other e-commerce retailer leverage similar technology?
