One of the key elements in any social network is the search and recommendation algorithm: the one in charge of determining what content will be relevant and what will not. It is responsible for choosing those publications worthy of being seen by certain users, based on their relevance, tastes and preferences, but also on their profile, contacts and more. The algorithm allows the connection with other users and the discovery of new content, a fundamental element to achieve a full experience on the social network and for brands to stand out and reach their target audience.
It sounds easy, and it can be if you keep in mind that TikTok’s algorithm differs from other algorithms. We have already talked about or how it works Today we will explain how the social network of the moment works.
Factors influencing content ranking in the TikTok algorithm
The experience on TikTok is different from other networks: as soon as you open the social network and find your feed, you can see a quick sequence of short videos, generally with selected themes based on content that you might like. These recommendations powered by the TikTok algorithm make each feed special and unique, fully adapting to each user and creating a totally personalized experience.
Among the points that the TikTok algorithm uses to classify its content, user interactions stand out above all, such as:
- the videos you like
- The videos you share
- The accounts they follow
- The comments you post
- The content that the same user creates
The TikTok algorithm also takes into account other factors in its classification, such as:
- The video informationwhich can include details such as subtitles, sounds and hashtags
- Device and account settings, such as your language preference, country setting, and device type. These latter factors are included to ensure that the system is optimized for performance, but they receive a lower weight in the TikTok algorithm relative to other data, mainly because users do not actively express them as preferences.
“all these factors are processed by our recommendation system and weighted according to their value for the user”. Thus, a strong indicator of interest such as the fact that the user finishes watching a longer video from beginning to end, will receive more weight than a weak indicator, such as the fact that the user and the creator of the content are in the same country.
On the other hand, TikTok also penalizes in some way the duplicity of content: “To keep your feed interesting and varied, our recommendation system works to intersperse various types of content alongside content it already knows you like. For exampleyour feed will generally not show two videos in a row made with the same sound or by the same creator. We also do not recommend duplicate content, content that you have seen before, or any content that is considered spam. However, they may recommend a video that has been well received by other users who share similar interests.
How the feed is set on TikTok
The micro-video social network ensures that while a video is likely to receive more views if it is posted by an account that has more followers, It won’t carry much weight for your ranking, follower base, or whether the account has had high-performing videos in the past.: the key is hyper-personalization of the results offered.
But… what happens when a user is new or has not interacted with the content? The social network ensures that to help users get started, they are invited to select some categories of interestsuch as pets, or travel-related issues.
With this information, the platform can create an initial feed and from there begin to refine recommendations based on interactions with a first set of videos. If any user does not want to select favorite categories, the platform will start by offering a generalized source of popular videos.
Thus, each new interaction helps the system learn about your interests and suggest content– Following new accounts, exploring hashtags, sounds, effects and trending topics in the “Discover” tab will further tailor the user experience.
But the TikTok algorithm can also work in reverse: when a user finds a video they don’t like, they can long press it and tap “not interested” to dismiss it, or you can also choose to hide videos from a certain creator, a sound, or report when a video doesn’t seem to conform to the platform’s policies.
TikTok wants to avoid the “bubble effect”
One more action that the social network has taken to keep its users’ feed interesting and varied is insert various types of content along with those that the user has selected: The platform ensures that it does not recommend duplicate content, content that has already been seen before, or any content considered spam.
In addition, its recommendation system is also designed based on security: videos that have just been uploaded, that are under review, and those videos that seek to artificially increase traffic may not be eligible for recommendation in the users feed.
As explained from the social network itself: “Our recommendation system also It is designed with safety in mind. Content depicting things like medical procedures or legal consumption of regulated products, for example, which may be shocking if it appears as a recommended video to a general audience who has not opted in to such content, may not be eligible for recommendation. Similarly, videos that have just been uploaded or are being reviewed, and spammy content, such as videos that seek to artificially increase traffic, may also not be eligible for recommendation from anyone’s feed.”
Stay informed of the most relevant news on our news channel