Amazon Visual Fashion Finder

 

A new way for customers to shop visually for fashion on Amazon. My approach was focused on visual filtering, visual search and applied machine learning. Customers can use visual cues and photos instead of words to describe what they are looking for. Like a personal shopper: find it on Amazon by describing it visually.

My role was to solve for a complex problem in 12 weeks: Irrelevant Search Results. I focused on a highly visual and hard-to-describe category: dresses, while considering scaling as well. Many of the most important aspects for shopping for garments and accessories including cut, material, and pattern cannot be specified through filters. My challenge was to improve the experience for fashion on Amazon by providing comprehensive tools for filtering by the visual aspects of a product, including a friendly interface that makes shopping for clothes on Amazon more fun and creative. I was paired with other interns, an Android developer and PM to work on this solution, with the support of the Amazon fashion design, engineering and PM team.

The impact was that I inspired new filtering approaches and the feature went on to launch as, "Dress Finder," based on my concept, within the next year.


Screen Shot 2018-07-24 at 1.11.50 PM
Screen Shot 2018-07-24 at 1.09.32 PM
Screen Shot 2018-07-24 at 1.12.31 PM

Taking inspiration from my concept brief, my plan was to create a mobile experience that is:
1. A convenient and efficient solution for customers to, "Shopby" rather than "Shopfor"
2. Scalable, reliable, a long-term solution that reduces customer effort
3. Fun, creative, interactive, innovative, unique; game-like
4. In line with the Amazon brand and its best practices


Visual Search

Shop using a photo

Results Feed

View results on Amazon

Recommendations

Filtering

Interaction

        © Andrea Alam 2019