Have you ever felt overwhelmed by the volume of Wayfair's catalog? Wouldn't it be nice if there were someone to meet you at the door and help you find what you're looking for? Meet WayFinder, a digital concierge that quickly learns your preferred style and makes recommendations based on your feedback. No more browsing to page 57 to find that perfect area rug! Let WayFinder bring it to you!

What it does

WayFinder begins by presenting you with a small array of products. You click on the products you like and WayFinder uses that information to fetch a new array of products similar to the ones you previously liked. The more you use the tool, the better WayFinder does at identifying the attributes that matter and recommending products that match those attributes.

WayFinder improves upon Wayfair's existing personalization efforts in many ways

  • Improved recommendations - WayFinder can learn about users from the products they like and also from the products they don't like. Personalization on can't do the same because it doesn't know what a user has actually seen (and maybe they clicked on something but didn't like it).

  • Speed - By eliminating the act of browsing, WayFinder can learn your preferred styles in a matter of seconds. To accomplish the same degree of personalization on would take hours of browsing.

  • True personalization - Due to the rapid-fire nature of WayFinder, it doesn't need to rely on "customers like you" to generate its recommendations. WayFinder can quickly hone in on your specific needs, whether you are a typical customer or not.

  • A unique ecommerce experience - All major ecommerce sites use personalization to some degree, but nobody has taken the Pandora/Tinder/Stitchfix model and made it a viable way of exploring a furniture catalog. With rapid speed and increased personalization, WayFinder brings customers the rewarding feeling of turning a house into a home.

How we built it

  • Brad/Danny - designed the UI that allows users to mark the products they like
  • Tom/Ben - developed the algorithm to generate new/improved product recommendations based on user feedback
  • Jing - built the backend that connected the UI to Wayfair's databases and Tom/Ben's recommendation algorithm

Accomplishments that we're proud of

  • After one day, it actually works (only for table lamps).

What's next for WayFinder

  • Mobile friendly UI (swipe right on garden gnomes!)
  • Save your "liked" products to an Idea Board
  • Broaden our definition of product similarity to include things like style, price point, manufacturer, color, size, material, etc. Currently we only use visual similarity.
  • Use WayFinder results to inform/accelerate personalization elsewhere on site, as well as email/marketing campaigns

Built With

Share this project: