YouTube Discovery Prototype

Background info

YouTube, the most popular video sharing platform, has over 1.5 billion monthly users. Every minute, 500 hours of videos are uploaded. On average, each user spend at least 1 hour on YouTube.

This project is from DubsTech protothon. Above is a picture of the prompt

Problem Statement

Massive amount of videos are uploaded every minutes, but only a few are discovered by the users. How to help new channels grow subscribers? How can new content be discovered?

Stakeholders:

Viewers:
  • Not aware of new channels
  • Not satisfied with the known channels
  • Want to watch more new videos
  • Too lazy to discover new channels
YouTubers:
  • Not known by the users
  • Users don’t know there’s a new video if they didn’t subscribe
  • Hard to get feedbacks from users
  • Might lose confidence in making videos
YouTube:
  • Want to make more profit by having more popular YouTube channels
  • Have a hard time to suggest channels for users
  • Want more users to use YouTube

User Research:

  • 40% of the users uses mobile devices
  • Video uploaded in 2017 are mostly "people & blogs", "gaming", "film & animation", and "entertainment"
  • At least 50% of the users are between age 18-45
  • Top earning users have over 3 millions of subscribers
  • Average video length of a video is less than 30 minutes (unless it’s a movie) and 50% of them are under 10 minutes
  • Most popular searched terms are "music", "songs", "Minecraft", and "happy"

Existing Solutions:

YouTube uses a method called “co-visitation”. First, it determines a set of related videos after users watched a “seed” video. Then it computes a “relatedness score” based on how often these videos are watched by the same user and how many were liked by the users. Later, YouTube can use this score to recommend videos for new users. This score can also vary by video quality, comments, and similarity to the seed video. For example, if a video talked about the same thing as the seed video, its score will be changed so it won't show again.

Ways YouTubers can increase popularity:

  1. Provide links to other videos. 
  2. Make playlists. 
  3. Ask users to subscribe to the channel. 
  4. Use YouTube Creator Academy(basically teaches you how to be a YouTuber)
  5. Upload mind-blowing, unprecedented content.

Ideation

I had many ideas for this project. 

  • I thought about adding a discovery page that only posts new channels and new YouTubers. But I don't think people will be that interested. It's too similar to trending page.
  • A ranking chart that shows the top 10 fast growing new YouTubers by genre. But I also thought the ranking chart doesn't show the content. Even if users looked at the name of the channel, the topic of each channel could be ambiguous. 
  • Another idea is to hide the number of views for each video, so users wouldn't judge a channel by it's viewer. Many times I followed a YouTuber because he/she has a lot of followers. Reduces a priming effect of better the channel the more viewers are there. But this solution could prevent popular channels from growing.
  • My final solution is to spoon feed the users new videos.

The picture below shows the home page I designed initially. The existing home page always suggests the videos by subscribed channels. Most times, these channels don't post videos on a daily basis(personal experience) I suggest the home page to have a slideshow on the top. It has a different collection each week and each collection contains both newer and older videos. It's like a combo of everything. Users can explore and expand their interests. The section below the slideshow contains video by category or by relevance.

home page

Personally speaking, I have never opened trending page on the YouTube app. After talking to people in the Protothon(basically hackathon for UX Designers), I realize a lot of people are like me. We're all too lazy to explore channels, and we just look at the feeds on the home page. Sometimes we will search things we're interested in, like "people eating tide pods" and start watching random videos.

My hypothesis is that people will not use the separate discovery channel because it's too much clicking around, so discovery should be embedded into the existing YouTube. The picture below shows the original trending page. It will suggest users with new popular videos and related videos that users have watched before.

To test my hypothesis, I would do an A/B test. A test would be putting this suggested area in the home page and the B test would be putting this area in the trending page.

trending page

Prototyping

I used Adobe XD  to create high fidelity prototypes so that users can have a better understanding of my design.

Home Page

Home Page

This is the main page of the YouTube app. On the top of the pager, there is  a slideshow of each week's collections of videos. The collections are put together based on the news, or season or holidays.

Beneath the collection are the suggested trending videos and suggested categories. In the trending part, users can find the  most viewed videos of the day. In the suggested channels, users will find videos similar to their watch history content. They're grouped by topics. The number of views is hidden for the suggested videos, so people won't only view popular videos.

Subscription page

Subscriptions Page

This is subscriptions page which contains all the subscribed videos. All the videos are blended together, but if the user wants to, they can select channel name.

Each videos shows the title, channel, the length of the video. Notice that the number of views and likes are also hidden.

Library

 Library Page

Usually, the videos with the same topic are put into playlists. On this page, user can see the playlists they've watched in the past. They can go to any playlist to rewatch the video and they can also make their own playlists.

On this page, users can also find the video history, his/her videos and subscribed videos.

Search Bar

Search Bar

Even before the users start typing, the search bar with suggest new videos for users to watch. These search key words are generated based on user's watch history.

Once user typed words, the search bar will generate trendy topics related to the words.

 

Link to the prototype :) feel free to try the app and leave me any comments! Thank you!