TikTok secret algorithm is a big 'Duh!'

TikTok’s powerful algorithm steering the social media activity of millions of daily users is no longer a mystery.

In a new New York Times report that outlines information gleaned from what’s been verified as an official document, TikTok’s algorithm is laid bare. Wait for it: It prioritizes retention and time spent, and is insanely obvious.

Boiled down to its essence, the TikTok algorithm looks for video likes, comments, and how much of the video you watch (a fully watched video is obviously worth more than one you stop or scroll past before it’s complete).

The system understands, based on tags and other details, what each video is about. If you like a certain kind of video and watch it all the way through, or even pause it repeatedly and then still complete playing it, those are positive signals for the algorithm.

Since TikTok’s goal is to bring you back to the platform as often as possible, and retain your attention once you’re there, the algorithm will then feed you more of the kind of videos you’ve indicated that you like/prefer through the TikTok algorithm

It’s an addictive, virtuous circle, right?

Videos that string together a story – and get you to watch all of them – can supercharge the influence of that topic in the algorithm to give you more of that kind of meaty content.

A path to content addiction

When I started watching TikTok videos a few years ago, virtually all of them were 15 seconds or less. My feed was full of magic tricks, DIY, and people doing dances. None of this, even back then, was random. I was interested in learning new magic tricks (I’ve been an amateur magician almost all my life), I love home hacks, and couldn’t get enough of the dances because I appreciated the skill, and wondered how people of all ages (including mine!) had the skill and energy to learn and do them. What I watched fed all the videos I saw. To this day, my youngest child often comments that I see a very different TikTok than they do.

Nowadays, I watch longer, storytelling videos, like those from Elise Meyers, whom I stumbled on a few months ago when she told this lengthy and hilarious tale about a blind date who picked her up, drove them to Taco Bell, and then bought 100 tacos.

She’s a brilliant storyteller. I know this because now I’ve consumed dozens of her videos thanks to the algorithm, which is feeding me more and more of them (along with other long-form story-telling videos).

These three-minute TikToks are obviously a boon for TikTok’s core goal of collecting and retaining more users. Time spent as a metric isn't unique to TikTok. Anyone who runs a content website knows the value of more time (and pages) consumed, which usually translates into more served ads.

Down the rabbit hole

What’s notable, though, is that TikTok’s algorithm still allows for discovery. Yes, there’s a lot of showing you more of what you clearly like or want (to the possible detriment of those who may be in a dark place and are gravitating to depressing/angry/harmful videos).

On the other hand, TikTok retains a bit of serendipity. Every once in a while, I see a video that has nothing to do with my likes or interests (at least as I express them on TikTok), but I get hooked. That’s how Elise Meyers happened. These random videos are usually a product of extreme popularity elsewhere on TikTok, which then drives that content into your feed, for you to feed the algorithm with fresh attention info.

The downside of this fairly simple algorithm is that it can appear to get stuck. Suddenly, I have five or six Elise Meyers in one feed session, and even I can’t slog through that many minutes of her crazy stories. I usually take a break and then come back to them.

I’ve also found that, once you understand the TikTok algorithm, you can untrain and retrain it. If I find too much of one kind of video in my feed. I do some hashtag searches and then rabbit-hole down a few fresh topics.

This usually works until I gravitate back to my old haunts (magic, DIY, FX, Elise Meyers) – and then I’m back to where I started.

Post a Comment

0 Comments