YouTube shares new details about how its recommendation algorithm works.
New information about how various factors influence YouTube’s video recommendation algorithm is revealed by members of the team responsible for working on it.
Having only been implemented in 2016, we still have a rudimentary of how YouTube’s machine learning algorithm works.
We know video recommendations are influenced by factors such as clicks, watch time, likes/dislikes, comments, freshness, and upload frequency.
We do not know, for example, whether external traffic has any impact on video recommendations.
It’s also not known whether underperforming videos will affect the likelihood of future videos being recommended.
The impact of other potentially negative factors, such as inactive subscribers or too-frequent uploads, is not known either.
Those are the factors YouTube’s team discusses in a new Q&A video about the recommendation algorithm. Here is a summary of all questions and answers.
Underperforming Videos
If one of my videos under-performs, is that going to hurt my channel? Could a few poor videos pull down better videos in the future?
YouTube doesn’t make assessments about a channel as a whole based on the performance of a few videos.
YouTube only cares about how people are responding to a given video when deciding whether to recommend it to others.
The recommendation algorithm is always going to be “following the audience.”
If a video is attracting an audience then it will show up in users’ recommendations regardless of how the channel’s previous videos performed.
Channels shouldn’t be concerned about some kind of algorithmic demotion based on a dip in viewership.
It’s normal for the performance of videos to fluctuate in terms of views and other metrics. So YouTube is careful not to have all of its recommendations driven by those metrics.
Too Many Uploads Per Day
Is there a point at which the number of videos per day/week on each channel is so high that the algorithm is overwhelmed and videos slip through?
There is no limit to how many videos can be recommended to a given viewer from a channel in a single day.
Channels can upload as much as they want. How many views each video receives comes down to viewer preferences.
YouTube’s recommendation system will continue to recommend videos as long as viewers continue to watch them.
If a channel is uploading more videos than usual, and each video is getting progressively fewer views, that may be a sign the audience is getting burned out.
While there is no limit to how many videos YouTube will recommend from a channel in a single day, there is a limit to how many notifications will be sent out.
