How the TikTok algorithm works
TikTok has published more about its recommendation system than most platforms. The mechanics are simpler than the folklore — and more indifferent to who you are.
8 minute read
What TikTok says it measures
TikTok has described the For You feed publicly on several occasions, and the stated inputs fall into three buckets: user interactions (what a viewer watches, finishes, rewatches, likes, shares, and skips), video information (captions, sounds, effects — the metadata that tells the system what a clip is about), and device or account settings, which carry the least weight. The strongest signal in the first bucket is completion — watching a video to the end, or watching it twice, tells the system far more than a like does. None of this is secret; the folklore mostly comes from creators reasoning backwards from single data points.
What TikTok says it does NOT measure
Two absences matter more than anything on the list. TikTok has stated that follower count is not a direct ranking factor, and neither is whether your previous videos performed well. Each video is evaluated on its own behaviour in the feed. This is why brand-new accounts can land a widely-viewed video in their first week, and why accounts with large followings routinely post videos that go nowhere. Your history earns you almost nothing per-post — which is bruising if you've built an audience, and liberating if you haven't started yet.
The testing pool, concretely
A new video is shown to a small batch of viewers chosen partly by the video's metadata and partly by availability — not primarily your followers. If that batch watches, finishes, rewatches, or shares at a healthy rate, the video graduates to a larger batch, and the process repeats. Most videos fail an early round and settle at modest numbers; a few keep passing rounds and compound. The visible symptom is the characteristic TikTok view pattern: a burst in the first hours, a plateau, and occasionally a second wave days later when the system re-tests against a new audience segment.
The interest graph, and why niches are visible to the system
TikTok's feed is built on interests rather than social connections — it groups viewers by what they watch, not who they know. Practically, this means the system is always trying to answer one question about your video: which cluster of viewers is this for? Videos that answer it clearly — consistent subject, consistent format, spoken keywords and on-screen captions that match the content — get tested against the right audience and hold better. Videos that could be for anyone get tested against a general audience and usually die there. Legibility to the interest graph is one of the few durable levers a creator has.
What you can actually control
Strip the mechanics to what a creator can act on and you get a short list. Hook hard: the skip decision happens in the first second, so open with the most interesting frame and line you have. Make it finishable: shorter clips complete more easily, and completion is the strongest signal — don't pad. Make it legible: say and show what the video is about early, caption it, and stay on a consistent subject so the interest graph knows where you belong. And publish at a rate that gives the testing loop data — per-video variance is so high that a handful of posts tells you nothing about whether your approach works.
FAQ
Does follower count affect TikTok reach?
Not directly — TikTok has said follower count isn't a ranking factor for the For You feed, and each video is tested on its own behaviour. Followers still matter for the follower feed, lives, and the credibility of your profile once someone lands on it.
Is there a TikTok shadowban?
TikTok acknowledges reducing distribution for content that violates or borders on violating its guidelines. Outside those cases, most 'shadowban' experiences are the normal variance of per-video testing — a few clips failing early rounds in a row.
Do trending sounds help videos go viral?
They help the system categorise your video and can surface it to viewers browsing that sound, which is a modest boost. They don't compensate for a weak hook or poor completion rate — retention still decides the outcome.
Related pages
Keep reading
More Creator Growth →The best time to post on Instagram, according to four studies
Buffer says Thursday 9am. Later says 5am. Sprout says early afternoon. They analysed billions of posts and still disagree — here is why, and what to do about it.
8 minute readThe best time to post on TikTok, according to five studies
Buffer analysed 7.1 million posts and rates Saturday the best day. Sprout Social analysed 2 billion engagements and says avoid the weekend entirely. Both are right, for different accounts.
8 minute readThe best time to post on LinkedIn, according to five studies
Buffer's 4.8 million posts point to Wednesday at 4pm and rank Tuesday among the worst days. Sprout Social's 2 billion engagements call Tuesday the best day of the week. Here is why.
8 minute readTurn one long video into a week of posts
300 credits for 3 days · no card.
Start free