Methodology & Data downloads

YouTube’s recommendation algorithm determines what people watch for more than 70% of the views on the platform. That's 700 million hours - or 1,000 human lifetimes - every single day. Most YouTube recommended videos appear in the “Up next” list on the right of the screen, or below the video on mobile. When autoplay is enabled, a suggested video will automatically play next.

We used a multi-step program to analyze videos recommended by YouTube's recommendation algorithm every day:

Step 1: We begin from a list of 800+ US channels. These channels have been selected to be the most recommended from top political channels. The majority of them are in the green "news" cluster seen here.

Step 2: From each of these 800+ channels, we identify the last video that was uploaded. For this video, we then gather all the recommended videos in its 'up-next' line-up.

Step 3: For each video that was observed, we count and display from how many videos it was recommended.

Please note: We perform the searches from New York City, and as a result local channels from NYC get many recommendations from the algorithm. We perform our queries without cookies or being logged in, and therefore do not collect "recommended for you" videos.

See the full list of 800+ channels here.