How to Go Viral on TikTok (What Actually Drives Distribution)

TikTok shows a new video to a small test audience and watches what they do. The strongest signal by a distance is watch time — specifically, what proportion of the video people complete, and whether they rewatch it. Shares come next, then comments, then likes, which are the weakest of the four despite being the most visible.

This ordering explains a lot of otherwise confusing behaviour. It is why a video with 400 likes can out-travel one with 4,000: if the first held people to the end and the second lost them at second three, the first is the better bet and the system treats it that way.

It also explains why follower count is a poor predictor of reach. The For You feed is not primarily serving your followers. A brand new account and one with 200,000 followers get tested on broadly similar logic, which is why accounts with nothing behind them go viral constantly.

Almost every retention problem is a hook problem. If a viewer swipes at second two, nothing else in the video was ever evaluated — the editing, the payoff, the information you spent an hour on were never seen by the system in any meaningful way.

The hooks that work share a shape: they establish a reason to keep watching before they establish context. Most creators do the reverse, opening with an introduction, a logo, or a slow build to the interesting part. On a feed where the alternative is one thumb movement away, that ordering is fatal.

Because completion rate matters, a common conclusion is that videos should be as short as possible. That is half right. A seven-second video is easier to complete, but completions of very short videos carry less weight than sustained watch time on a longer one, and a loop that nobody rewatches is not the win it appears to be.

The more useful target is a video that is exactly as long as its idea and no longer. Trim to the point where removing another second would remove something a viewer wanted. That usually lands well short of where most creators stop cutting.

Of the engagement actions available, a share is the strongest, because it takes the video outside your existing audience graph and puts it in front of people who have never encountered your account. That is functionally what going viral means.

Shares come from a specific emotional trigger: the video says something the viewer wants to be seen agreeing with, or gives them something useful enough to pass on. Content that is merely enjoyable gets liked. Content that says something on the viewer's behalf gets sent to a group chat, and that is a different order of distribution.

Virality is not reliably repeatable on any individual video. Even accounts that go viral often are producing a lot of posts that do not, and the ones that break through are frequently not the ones the creator expected.

The practical consequence is that posting frequency is a genuine lever, not a platitude. Each post is an independent draw. An account posting five times a week is taking substantially more draws than one posting twice, and over a quarter that difference dominates almost every optimisation discussed above.

This is also why chasing a single viral video is a poor strategy even when it works. One spike brings a wave of viewers who have no reason to expect a second good video, which is why so many creators watch a hundred thousand views convert into a few dozen follows.

Entering a trend early gives you access to an audience already primed for that format, and the sound itself carries some distribution. Entering late does close to nothing — by the time a sound is obviously everywhere, the surplus attention has been consumed.

The stronger play is to use a trending format as a vehicle for something specific to your niche rather than reproducing it as-is. A generic entry competes with thousands of identical videos on execution alone. A version that only your account could have made competes with far fewer.

No, and it is worth being blunt about it. Purchased views raise the number under a video; they do not make people watch it to the end, and completion rate is the thing that drives distribution. A video with poor retention and fifty thousand bought views is still a video with poor retention.

What views can do is clear the cold-start problem — get a post past the point where it has too little data to be evaluated, so a genuinely good video gets tested properly instead of dying in a hundred-view sample. That is a narrow, real use. Anyone selling it as a route to virality is selling you something else.