基础分发模式
我们先讲一下抖音最基本的推荐是什么运作的。这里需要先说一下流量池,在抖音的内部有许多大大小小的流量池,每一个新的视频都会把你放入最小的流量池,在这个流量池里,看这个视频的完播率、点赞数、转发量、评论量,根据这些数据看你是否有资格进入下一个流量池,给你更多的播放量。
正常的视频应该是:完播率>点赞量>评论量>转发量
标签分发模式
我们刷抖音的时候会有一种感觉一开始什么视频都有,慢慢的我们视频开始慢慢集中到某一类型。比如我以前经常玩游戏就经常有一些吃鸡,王者荣耀的视频,后来慢慢游戏玩的少了,现在就主要是一些科普类型,和情景剧。这就是抖音给我们每一个人的标签不同,给我们推荐的也不同。在我们刷抖音和用头条系产品的时候,甚至手机别的应用的时候,就会给你打上不同的标签。
抖音用这些标签来画出用户画像,根据每一个用户画像来分发不同的内容。
Basic distribution mode
Let's first talk about how the most basic recommendation of Douyin works. Here we need to talk about the traffic pool first. There are many large and small traffic pools inside Douyin. Every new video will put you into the smallest traffic pool. In this traffic pool, the completion rate, number of likes, number of reposts, and number of comments of this video are looked at. Based on these data, whether you are qualified to enter the next traffic pool and give you more playback volume.
A normal video should be: completion rate > number of likes > number of comments > number of reposts
Label distribution mode
When we browse Douyin, we will have a feeling that there are all kinds of videos at the beginning, and slowly our videos begin to focus on a certain type. For example, I used to play games frequently, and I often had some videos of eating chicken and King of Glory. Later, I played fewer games, and now it is mainly some popular science types and sitcoms. This is why Douyin gives each of us different labels and recommends us differently. When we use Douyin and Toutiao products, or even other mobile applications, you will be labeled with different tags.
Douyin uses these tags to draw user portraits and distribute different content based on each user portrait.