Understanding how discoverability works on Whatnot Understanding how discoverability works on Whatnot

Understanding how discoverability works on Whatnot

Whatnot Support Whatnot Support

As our technology evolves, we want to make sure we are always providing clarity about how discoverability works on Whatnot. In this article, we’ll shed light on how these algorithms influence shows that users see in their For You, Followed Hosts, and Category feeds.

Let’s start with some context.

Shows on Whatnot

When you open the Whatnot app, you’ll see vertical feeds of show thumbnails. These feeds are based on discoverability techniques driven by a variety of signals (we’ll get more into signals below), all designed to optimize the viewing and buying experience for each individual user.

Algorithm overview

Whatnot employs a multifaceted discoverability system throughout the app to ensure users see content that aligns with their preferences and interests. Every feed on Whatnot has its own purpose, so we also tailor our discoverability methods and algorithms to each feed, instead of using a one-size-fits-all-feeds approach. We regularly test and update these methods to enhance the user experience.

Signal insights

​​We analyze user interactions with sellers, shows, listings, and other app content to generate signals. These signals help us predict the content that's most likely to drive user engagement and conversions.

The following is a list of some of the signals we use:

  • The user’s explicit and implied interests in categories, sellers, and listings on Whatnot.
    • For example, if you follow "Sports" but don't explicitly follow "Basketball Cards", we will consider "Basketball Cards" to be an implied category of interest.  
  • The quality of the experience provided to buyers by sellers, including the quality of the show, listing quality, time to drop off shipments, order fulfillment, and more.

How discoverability works in the "For You" feed

We designed the “For You” feed to help you discover shows. This feed is made up of recommendations – shows from sellers you may or may not follow. Notably, the “For You” feed only includes live shows - there are no scheduled shows in the “For You” feed.

The first step we take is defining a set of shows that are eligible to be placed in someone’s “For You” feed. To find shows you might be interested in, we look at your past activity such as shows you’ve bookmarked and categories that you’ve followed. 

Once we’ve found some shows you might be interested in, we then order them by how likely you are to engage with each one. The most important actions we predict in the “For You” feed include views, bids, and orders. The most important signals are as follows:

  • Information about the seller: We leverage signals like the seller’s main category, average order value, and how many times people have interacted with that seller in the past few weeks.
  • Your history of interacting with the seller: If you have ever watched the seller’s shows, interacted with them, or purchased / bid from them in the past, we will tend to recommend shows from this seller to you.
  • Your history of interacting with the seller’s top category: If you have ever watched content from the seller’s top category, interacted with other sellers in that category, or made bids / purchases from a category, we will tend to recommend shows from that category to you.

To ensure that buyers can get matched with reliable sellers who consistently provide positive experiences, we also prioritize shows from sellers who continue to follow Whatnot’s Community Guidelines and maintain high Trust & Safety standards.

How discoverability works in the “Followed Hosts” feed

We designed the “Followed Hosts” feed to surface popular content from the channels that you follow.

First, we define a set of eligible content. The show's channel must be one of the channels that you follow.

Once we’ve found a set of shows from channels that you follow, we order them by their historical popularity, i.e., how much activity the channel has generated in the recent past.

How discoverability works in the category feeds

We designed category feeds to surface popular content in a specific category.

To begin with, we choose which shows could go into a particular category or subcategory feed. Sometimes, one show can qualify for multiple feeds: if, for instance, someone is doing a show about football cards, it could go into the Sports Card category feed or the Football Cards subcategory feed.

Once we’ve found a set of shows that align to a category feed, we order them by their historical popularity, i.e., how much activity the channel has generated in the recent past.

Tips for sellers

For sellers who want to optimize the discoverability of their show across all our feeds, the following recommendations may help:

  • Great buyer experience:  Deliver exceptional buyer experiences. Going live frequently, shipping in a timely manner, accurately representing items, responding to reasonable inquiries, and fostering a positive community atmosphere can all contribute to improved relevance.
  • Compliance with guidelines: Abide by Whatnot's Community Guidelines and regularly monitor your Trust & Safety Metrics in Seller Hub. A commitment to integrity, honesty, and community-building enhances the relevance of your shows and listings to users.
  • Accurate categorization: Correctly categorize your shows and listings, as precise information helps identify the most relevant audience.
  •  

Add comment

Article is closed for comments.