TikTokUser Psychology6 min readA feed with no navigation left to do
One video, full screen, and a swipe that costs nothing.
- The surface
- The For You feed on open: full screen video, vertical swipe, and the side rail of like, comment, share and sound.
- What the user wants
- I have a few minutes and I want to be entertained without picking anything.
Playing before you ask
The app opens directly into a playing full screen video rather than a menu, a grid or a following feed.
Zero-decision entry
Every screen between opening and being entertained is a place to lose someone. Starting mid-experience means the user never has to form an intent. It also means the first item has to be strong, which is why the opening slot is the most heavily optimised position in the product.
One item at a time
Each video fills the screen with no visible neighbour, and the only navigation is a vertical swipe.
Removing comparison
A grid asks the viewer to choose, which is work and invites rejection of everything on offer. A single item removes comparison entirely: the only question is stay or swipe. That binary is fast, so the system gets a clean signal on every item.
Swipe as the main signal
Watch duration, rewatches and swipe-away timing are captured on every video, and the viewer does not have to like anything for the feed to respond.
High-frequency implicit feedback
Most recommenders are starved of data because rating is rare. A feed of short items generates a judgement every few seconds, and the swipe timing carries more information than a like would. Model quality here is a function of interaction frequency, not of clever inference.
Short items, variable payoff
Videos are short and quality varies sharply from one to the next, with no ordering the viewer can predict.
Variable-ratio reinforcement
Unpredictable reward on a short interval is the schedule that produces the most persistent behaviour, which is why slot machines use it. Short items make the cost of a miss trivial, so the viewer keeps pulling. The product's stickiness comes from the variance, not from the average quality.
The sound and the template
Every video carries a tappable sound that opens a feed of other videos using it, and the app surfaces formats that many creators have reused.
Participation scaffolding
A blank creation tool is intimidating, and most people never use one. A sound with an established format tells a newcomer exactly what to make, which converts viewers into creators at a rate an open canvas never would. It also gives the recommender a structural handle on trends.
No end
The feed does not terminate, has no read state and no session summary.
Absence of a stopping cue
People leave experiences at boundaries: the end of an article, the end of an episode, the bottom of a page. Removing every boundary means the session ends only when something outside the app intervenes. This is the single most consequential design decision in the product.
Whether a feed can be finished
What not to copy
- The combination of variable reward: An unpredictable payoff, which sustains a behaviour longer than a predictable one., zero interaction cost and no stopping cue is engineered compulsion. Screen time tools inside the app do not offset a mechanic built to defeat them.
- Because the feed responds to watch time rather than stated preference, a viewer who lingers on something they feel bad about gets more of it, and has no clean way to say so.
- Full screen single-item presentation removes context. A claim in a video arrives with no neighbouring material to argue with it, which makes the format unusually good at carrying misinformation.
- The creator side is a tournament: format reuse rewards volume and speed, and the distribution any given creator receives is unstable by design.
The takeaway
If your feed has no end, no read state and no summary, you have made stopping the user's problem, and you should say so out loud in the room.
Finished the teardown? Bank it and the day counts toward your run.
Where the principles come from
- Hooked: How to Build Habit-Forming Products, Nir Eyal
- Indistractable, Nir Eyal
- Schedules of Reinforcement, B. F. Skinner and C. B. Ferster
- Thinking, Fast and Slow, Daniel Kahneman
Written from public behaviour of the product, not from inside it. Interfaces change often, so treat the flow described here as of the time of writing and check the live product before quoting it.