Recently some YouTubers posted some videos about how YouTube has changed. YouTube mostly promoted some contents that they subjectively think are not great content. A controlled system like YouTube controls how the user gets the content. For now, it’s based on the virality and popularity of the content. If you check your YouTube home page, you will find those contents are recommendation from YouTube. It’s all about the algorithm.
Talking about the algorithm model, on YouTube, you can shape those recommendations, but it needs to click each content and set that you’re not interested in that content. The control of content to appear still really on the system, not on the users. There are two models of fundamental algorithms for content feed that I know:
- following/subscribing model (Twitter, Instagram)
- explore model (YouTube, TikTok)
It doesn’t mean that an app doesn’t allow you to use another model, but their homepage basically will use that one as a default. To use another model, you should click on the menu. How do you decide which one to use as a default? Both have pro and con.
Following or Subscribing Model
This model let the user control the content they want to see. They believe their taste or standard is perfect for themself. I think this model only for the users who know their interest, because for new users, finding a channel or user to follow will need some effort except you know the users or the content already from outside the ecosystem. I saw some people find that Twitter is not for them because they’re not enjoying it. I believe this is because they can’t build their feed to fit with their taste.
A product mantra will tell you to listen to your users, but what if the users don’t know what they want. Then you should help and guide them to find it. On Twitter, there is a recommendation whom to follow, but then this is mostly based on the geographical popularity/virality, not on user interest. That’s why recently, Twitter promote topics to follow instead of a user to follow.
I believe this model is perfect for a creator-focused platform, not the content. Because we believe the creator will create good content for us. That’s why we follow them. This model is maybe more for musician, artist, and other creators whom we believe in having a similar taste, standard, or genre with us. Who always rely on their fans.
Explore or recommendation model
This model more focuses on popularity/virality of the contents, and mostly it’s based on geographical or age. Because this is the essential information that you can collect easily from an early user, then you can build their interest model once they start using your product by collecting their behaviour or types of contents they’re consuming.
Starting with this model in the early phase of the product will help you to gain the users. Tiktok uses this model by default, and that’s why content easy to get viral on Tiktok because the recommendation is based on the popularity/virality. Then later, you need to build a topic interest model to provide a better recommendation because the users will stay if they keep interested in consuming the content. The bad thing, the creators who are focusing on niche content, will struggle to get their content discovered because the model relies on a number of users that like the content. The niche content consumer still lacks in numbers if compared with the mainstream one. So that’s why this model will kill the niche creators.
So that’s why I believe this is not about taste, genre, or standard. We just enjoy the content no matter who the creators. But once we realized that a user keeps producing good content, then we start to become a fan of that creator and care more about the content that they produce.
Mixing the two models
There is no perfect model between these two. But you can only choose one as default and provide another one as an option. As we know Twitter use the following models, but they kind like mix a bit for recommendation model by showing up content that your following was like, and it more potentially showing up on your timeline if that content also has a lot likes and retweets.
Recommendation model is perfect if we start a new content platform because we don’t have a user to recommend yet. The user is not having enough content to show if we focus on the following model. But once our platform is getting bigger, we need to see how people consume the contents. Is it more about a creator or the content? We can see which model is it by using the data and answer the question, is the user consume the content because of the content itself or because it created by the creators they like.
Everyone said that the product must listen to the users need. I believe it’s true, but a product owner you have a right and power to choose your users. Maybe this is sound like an idealist vision, and in reality, you could end up dying because you too idealist like Vimeo that I like as an excellent platform that provides non-mainstream good quality contents. They’re great, but they end up getting sold a couple of times because the tractions were not that big compare with other platforms. So you want to become an indie maker or mainstream one?