Econsultancy: What is the most exciting use of your image recognition technology in ecommerce?
Matthias Dantone: Of course we’re excited about and confident in all of our technology’s use cases. Lately however we have witnessed a great interest in visual search – a tool by which shoppers can upload an inspirational image to an app and shop the products in the image.
Just looking at the media frenzy that surrounded the visual search announcements by Pinterest or ASOS, for instance, it’s clear that this is exactly what shoppers want. They find what they are looking for, and the process to checkout is sped up. It’s an experience that’s valuable for both the shopper and the retailer.
E: How many consumers are using visual search?
MD: While we aren’t at liberty to disclose exact numbers, we can tell you that every month there continue to be new users and stronger statistics that support that this is the direction of the future of search.
E: Uploading pictures to visual search is easiest when browsing on mobile, where conversion is typically low. Is this problematic?
MD: This is still a problem in mobile commerce, but one that we’re trying to help solve. We bridge the gap from content to commerce by streamlining the path from inspirational image – be it on Instagram or in the image library on your phone – to the checkout page.
Among the many challenges of shopping on mobile is of course the screen size, which isn’t optimized for endless scrolling. Our Visual Search eliminates that entire process: Shoppers land directly on the product they were looking for, making shopping easier than it has ever been.
E: Your tool can be used to help advertisers with content creation. Where else can you see visual search used as a back-office tool?
MD: There are many use cases for Fashwell’s tech, both on the front and backend. For instance, we work with a fashion marketplace that uses our automatic attribute tagging in the backend. Each of their products is automatically tagged with attribute labels, both for their physical and aesthetic qualities. This helps the retailer manage their catalog, as well as personalize the search results for each customer segment since they have information on their shoppers’ style.
Additionally, we help to speed up and automate the product tagging process for the curation teams at a number of technology companies who build shoppable content or distribute UGC content for publishers, brands and retailers.
E: Why is visual analysis used for product classification? Isn’t it more efficient to classify with product data?
MD: We’ve been building classifiers for product data tagging, which are a much faster and more scalable solution for product tagging within ecommerce. A big problem that retailers face is that every brand has a different set of attributes and taxonomies with which they describe their products. For example, one brand may call something “pants”, another “trousers” and yet another “long pants”.
Fashwell can standardize this by looking at a product image: Our algorithms take it one step further by adding new product information that is generally not included in any type of data – like style, neck type, or the occasion the product would be appropriate for.
E: What will it take for visual search to be widely adopted in ecommerce? Is it a matter of time?
MD: It’s definitely only a matter of time. Some of the world’s biggest companies – Amazon, eBay, Pinterest – already offer visual search, and Europe’s two biggest etailers, Zalando and ASOS, have also implemented visual search as a permanent tool for their shoppers.
It’s been predicted that 80% of all search queries are going to be either through images or speech. So with more time, more usage and more technical fine tuning, most brands and retailers will turn to visual search as an effective ecommerce tool.
Related reading:
Comments