The author’s views are entirely their own (excluding the unlikely event of hypnosis) and may not always reflect the views of Moz.
Explore the role of demographics, engagement, and personalized recommendations in how TikTok surfaces content. Discover how TikTok attempts to avoid echo chambers and gain valuable insights into its inner workings. Dive into the fascinating world of TikTok’s algorithm in this new Whiteboard Friday episode with Lidia Infante!
Click on the whiteboard image above to open a high resolution version in a new tab!
Video Transcription
Howdy, Moz fans, and welcome to a new edition of Whiteboard Fridays. Have you ever felt like TikTok is reading your mind? It’s just an algorithm. My name is Lidia Infante. I’m the Senior SEO Manager at Sanity, and I’m going to talk to you about how TikTok offers you the videos that you like.
What do you like?
The very first thing they’re trying to do is to show you videos that interest you. They’re trying to keep you in the app as long as possible. So they need to figure out what you like. How are they going to do it? The moment that you sign up, you provide some demographics data. Your age, your gender, where you’re signing up from all gives TikTok enough information to start assuming what your interests are going to be and populate your For You page.
Out of that first For You page, the topics that you engage with or don’t engage with are going to give TikTok the information about what you might enjoy seeing on your For You page. After that, they’re going to try to find related topics to the ones that you have expressed an interest in. For example, I really like cross-stitching.
It’s not difficult to assume that I also like interior design or flower arrangements, which I do. TikTok shows them to me, and I stay on the app much longer than I should. Another point of information is your followed profiles and the topics that they talk about or that they are interested about. Your Discover tab behavior is a very strong signal for TikTok too.
What that means is when you search for a hashtag, click on a hashtag, make a TikTok search, search for a user or a trend or a TikTok sound, that’s all giving strong information to TikTok about what you like because making a search or diving deeper into the content of a sound is much more complicated behavior than simply just scrolling past on your For You page.
What is the video about?
Now that they know what you like, they know that they can serve you videos about it. But first they need to understand what the videos that they have in the database of content are about. To do that, they’re going to use the visuals. So literally TikTok can “see” with the visual AI the content of your video. What are the objects that are in there?
Are there faces? What are the emotions of the faces? Is there a dog? Is there a chair? They’re going to be able to see and process all of that. Another strong signal is going to be the text overlay that you put on your TikTok video, if you have any notes or any content put on there or if you’re manually inputting the captions. The audio on your video is very relevant for TikTok to understand what the video is about, and by audio I don’t mean TikTok sounds.
We’re going to come back to that. I mean the actual voice, the words that you’re saying or the ambient noise that sets the location of your video. The captions and hashtags that you use in the video are very relevant and help TikTok understand what it’s about as well. When you are done with creating your video and you start typing it out, when you create the cover for your video using the TikTok tools, that gives TikTok that level of information.
Then we’ve got the TikTok sound. If you’re using a TikTok sound that’s related to a trend or to a topic, it also gives TikTok the information about what your content is, what the emotional tone of your content is, and so forth. All of these elements, as a hardcore SEO, are what we would treat as on-page SEO but on the TikTok side of things.
Is the video any good?
Now they know what you like, they know what their database of videos are about, but they also know that they have a limited amount of time for your attention. So they’re going to want to show you only the best of the best of their content. They’re going to have to understand if the videos they’ve got are any good. How are they going to do that?
They’re going to be using, amongst other metrics, engagement. So the engagement of users that have expressed similar interest to yours are going to help TikTok predict your potential engagement and how much you’re going to like the video. They’re also going be favoring native content creation. So any video that’s being created directly from TikTok using the TikTok tools is going to get a little bit of a boost.
The language that the content is in is very relevant too. When signing up for TikTok and creating your profile, you state what are the languages that you speak and what is your preferred language. TikTok can also understand this based on your behavior. Then the device that you’re using and the strength of your connection also help TikTok serve you lighter videos and shorter videos when you’re maybe on a train or don’t have as good connection to actually have a good viewing session.
What isn’t eligible “For You” page content?
There is some stuff that TikTok does not want to see on their For You page. Any content created by a user that has stated that their date of birth is under 16 is not going to be eligible for the For You page. So if you’re trying to be cute and giving your brand account a date of birth of when it was founded, don’t because nobody is going to see it.
They’re also quite strict with QR codes. They don’t want to see any QR codes in there because of safety reasons. You don’t know if that QR code is going to take your users to a spam website or a dangerous website showing porn or violent content. But also they don’t want to miss out on the affiliate marketing payout that they get when you shop directly on the TikTok Shop.
Any dangerous or violent behavior is also not accepted or eligible for the For You page. So unless it’s tagged as done by specialists, don’t try this at home and so forth, it’s not going to show up. Engagement bait content is also not eligible for the For You page. So content that tries to trick users into engaging with it without them actually naturally wanting to engage with it is going to be demoted from the page.
That type of content is like when on Instagram we used to see posts that said tap twice to see some magic, and the only magic is that you had liked the post and increased the engagement of the brand. That is the type of engagement bait that’s not acceptable on TikTok. Another thing that they don’t accept on their For You page is any content that shows tobacco products.
Is TikTok capable of breaking bubbles
TikTok wants to show you videos that interest you and keep you in their platform, but they’re also trying to not generate an echo chamber for you. Echo chambers are a very real risk of any recommendation algorithm that’s based on interest and engagement, and we have seen a rise in radicalization across different populations because of this.
The way that TikTok is trying to address this problem is by offering random content of random interests, random creators that you’ve never seen before on your For You page. Initially, they said they were doing this about 10% of the time, but there aren’t any current numbers about this activity. This is not enough to remove echo chambers because of the psychological effects of familiarity and their exposure effect.
Once you see something that does not quite fit right with your For You page, you’re very quick to scroll. It is a similar phenomenon to banner blindness when we are going around the website. We kind of know what an ad looks like, and we don’t even look in that direction. It’s a similar effect.
So it is not enough, but one must appreciate that they’re trying. If you have found this interesting, you can catch up with me and my content on Twitter @LidiaInfanteM, on LinkedIn looking for Lidia Infante, or on my website lidia-infante.com. Thank you for watching.
