Wayfair - Shop by Color
Senior Associate Product Manager
June 2016 - July 2017
Overview
Team
Worked closely with Suyash Sonawane, Paige Darby, Tyler Abbott, and Mike Joseph.
Goal
Help customers find desired product faster. Eliminate noise in color search results and delight customers with unique color-based experiences.
Need
With hundreds of thousands of home products, Wayfair’s web store can feel overwhelming.
The existing color search functionality, supposed to help narrow down product options, returns high level, low quality data with thousands of results.
Project
Improved Color Search
An example of the original, on-site “blue” color search results (left) vs. the finer grained color-based results (right) that my team developed as part of an internal tool. Functionality can be built around specific hues, instead of blanket color-buckets that return the likes of 17,000 results.
Process
Researched state of the art color search experiences and color extraction software. Lead brainstorm and ideation sessions for customer facing experiences. A/B tested customer facing product concepts to prove value and iterate on functionality.
Solution
My team of engineers used an open-source, color-extraction software to get color and pattern data from product images. We then built an internal tool to view the color data and tested various algorithms for returning color-data and look-a-like products recommendations that could be used in functionality across all of Wayfair. Lastly, we A/B tested the color-based product recs in a "Love this? Explore these" recommendations carousel below a selected product's detail page.
Responsibilities
Product managed the Shop by Color product initiatives.
Explored various color-based functionality, both for customer facing experiences and internal tools. A/B tested site features and, with my engineers and the Site Merch Operations team, later investigated transitioning the company from a manually-tagged color database to one that is finer-grained and auto-extracted from product images.
Testing Site Functionality
My team A/B tested color-based product recommendations in the "Love this? Explore these" rec. carousel on the product detail page. As customers switched color selections, product recommendations updated accordingly below the main product block using Ajax.
Our hypothesis was that 'look-a-like' product recs would help customers find what they want faster and could be used to replace out of stock products with alternatives.
Learnings
After running several feature tests, we weren't getting significance or delivering the experience that we wanted. I realized that we needed to improve Wayfair's color-data quality in order to achieve meaningful onsite functionality around color.
We pivoted our main focus away from site features, into building out an internal tool and color-database to better view and improve overall color-data quality.
Data Quality Iteration
Home decor product images that are indexed for color make up the Wayfair company logo. The left versus right logo show differing color algorithms and outputted data quality.
Additional Projects
SproutsIO - Overview
Mariner Exchange