TikTok’s story is one of acceleration into the mainstream, with occasional speedbumps over privacy concerns and national security warnings. A lot of its momentum builds through its ‘For You’ feed—a content recommendation system that TikTok says is where most users spend their time.

Now, TikTok has offered a glimpse at the algorithm that dictates what appears in that feed. It’s out first fleeting look at the engine that powers the app’s virality.

Three Main Factors

TikTok says the videos that get recommended to any particular user depend on three main factors: user interactions, video information, plus device and account settings.

The first two factors clearly relate to defining the content a user might be attracted to. The latter, well, we’re not exactly sure…

User Interactions

Engagement with posts and accounts on the TikTok platform play a major part in the algorithm. This involves a history of the user’s:

• likes
• shares
• comments, and
• posted videos.

Of course, accounts that a user follows are also a big influence. Users will be recommended content not only from directly followed accounts, but from similar content too.

Viewing history is not explicitly listed by TikTok as a decisive factor, but it surely is. This is attested by the boast that the system even takes into account whether a user watches a video through to completion, or if they skip out halfway through.

Video Information

Video details that TikTok uses to connect the verticals include:

• songs
• captions, and
• hashtags.

Markers like these are used as indicators to group alike-content.

Device and Account Settings

Finally, TikTok factors in data drawn directly from users’ device and account details, including:

• language preference
• country setting, and
• device type.

TikTok says it is ‘settings like’ these that are collected. It does not say these are the only such settings that are collected; the actual datapoints are potentially much more varied.

TikTok claims these factors ‘make sure the system is optimized for performance’, though it’s not immediately evident how. TikTok concedes that ‘they receive lower weight in the recommendation system’. So, why mention them at all?

Perhaps these datapoints got a mention here to at least partially explain away the oft-voiced concerns about the type and extent of data that the TikTok app collects from its users.

What Doesn’t Matter

TikTok explicitly states:

"…neither follower count nor whether the account has had previous high-performing videos are direct factors in the recommendation system"

So, influencers looking to monetise their profiles will have to prove consistency in their content rather than rely purely on their historical metrics.

Take-Aways

Julia Alexander of The Verge saw TikTok’s revelations as an instructional guide for users to ‘tune’ their personal feed, observing:

"The app might not surface videos from the Black Lives Matter protests or may not recommend disabled or queer creators, if a user doesn’t specifically go out of their way to tune the algorithm in that direction."

Molly McGlew at Later reckons brands can use the new info to "get more engagement on their TikTok content". She recommends you hone your hashtag strategy, pick trending songs and sounds, post when your audience is active and keep your captions short and on-topic.

TikTok’s revelations are far from a Rosetta Stone to hack the ‘For You’ feed. Many of the details were easy enough to confidently infer even before TikTok PR confirmed them. There may be much greater detail to come before brands and influencers can formulate effective ‘For You’ penetration plans.

Will more info on how TikTok manages the feed make you change the way you use the platform?

Copy Transmission is a Melbourne-based agency :: Better Brands. Loud & Clear.