TikTok Algorithm Signals: What the System Actually Measures

How the algorithm reads signals

The algorithm operates on two levels simultaneously: it builds a profile for every user based on their behaviour, and a profile for every piece of content based on the signals it generates. Its sole task is to match these two profiles — finding the right viewer for each video as quickly as possible.

Every signal you send is added to a probabilistic model that determines how many additional viewers your video will be shown to. Positive signals expand the circle; negative signals narrow it or stop it entirely. A healthy average completion rate on a successful account sits between 15% and 22% — if your account is in that range, you are in the safe zone and growth is on track.

Completion rate — the highest weight

The percentage of viewers who watch to the end is the strongest signal the system measures. The algorithm treats the first three seconds as a "gateway" and video completion as the "green light" for expansion.

The difference between two numbers can create entirely different worlds:

• A 22-second video that achieved a 42% completion rate jumped directly to the FYP and reached 1.8 million views.
• A 45-second video whose completion rate dropped to 4.5% — because of a long, dull introduction — had its support cut entirely by the algorithm and stopped at 2,300 views, most of them from existing followers.

This is why a strong hook and a consistently engaging pace are not optional extras — they are the conditions for a video staying inside the distribution cycle. The minimum worth targeting is a 30% completion rate — below that, the algorithm will cut support early.

Shares — the distribution multiplier

Shares are the most powerful positive signal for a specific reason: when someone shares your video on WhatsApp, Instagram, or Telegram, they are bringing a new user into the app from outside — which is precisely what the algorithm wants to reward.

The numbers show the true scale of this effect: a video about "secret AI tools" reached 500,000 views with 25,000 shares — a 5% share rate. But the critical moment was in the first hour: the moment the share rate crossed 3%, the algorithm multiplied the video's distribution velocity by four times, because the "shared value" signal was so strong.

The practical takeaway: design content that makes the viewer think "someone I know needs to see this" — useful information for people they know, or a situation that describes someone in their life.

Comments — extending a video's lifespan

Comments have an effect that most creators are unaware of: when viewers read comments while the video plays in the background, it increases "dwell time" inside the video — giving the algorithm a continuous activity signal.

The difference in a video's lifespan is striking: a career advice video that ended with the contentious question "Do you think resigning without a backup plan is ever justified?" received 1,400 comments compared to the account's average of 50. The result? While the typical video's lifespan on TikTok is around 48 hours, the algorithm kept pushing this video for 14 consecutive days — because every new comment was reviving it in the algorithm's eyes.

To provoke comments: ask a question with two clearly opposing answers, or leave a deliberate gap in your content that invites people to weigh in.

Replays — the addiction signal

Replaying a video sends an exceptional signal: a retention rate above 100% — meaning the video was watched more than once per viewer. The algorithm classifies this content as "addictive" and pushes it with unusual force.

A real example: a 7-second "visual illusion" video achieved a retention rate of 145% — because people replayed it twice to understand the trick or read the fast-moving text. The result: 3.2 million views in just 24 hours, despite the video being too short to be considered "full" content by traditional standards.

To encourage replays: include a subtle detail that is hard to catch on the first watch, information that appears and disappears quickly, or an unexpected ending that makes the viewer go back to confirm what they saw.

Negative signals — the sudden brake

Negative signals do not just slow distribution — they stop it almost immediately, and can sometimes penalise the next video too.

On a video that strayed outside the account's usual niche, 8% of viewers in the first test wave tapped "Not interested," and the immediate skip rate was 90%. The result: distribution stopped entirely within 40 minutes, the video never passed 450 views — and the account's ranking was temporarily suppressed, which affected the very next video published.

The three negative signals that trigger the brake:

• Tapping "Not interested": the most damaging — it tells the algorithm the content does not suit this viewer and recalibrates targeting negatively
• Immediate exit in the first second: its impact on the model is similar to tapping "Not interested"
• Reporting content: an exceptional signal that affects distribution immediately and may trigger a full account review

User profile and content profile

The algorithm builds a detailed profile for every user based on: the topics they watch, the time they spend on each content type, the sounds that recur in their history, and the accounts they engage with. This profile determines who your video is shown to in its first test phase.

In parallel, the algorithm reads every video through: the caption text and spoken words inside the video, hashtags, the sound used, and computer vision techniques that analyse what appears on screen. The sum of these signals forms the video's "identity card" that determines who it will be shown to. On a typical, average-performing video, 75% of views come from the FYP, 18% from followers, and 7% from search and profile — a distribution that confirms precise targeting matters more than follower count.

The practical equation

Based on these numbers, the algorithm's decision can be distilled into a single equation:

Completion rate above 30% + share rate exceeding 3% = near-certain FYP explosion.

Every strategic decision you make in designing a video — the hook, the pacing, the ending, the topic choice — is at its core an attempt to reach or exceed this equation.

• Design every video so watching to the end is the natural path — remove weak spots in the middle that push the viewer to leave
• Create content that gets shared naturally: useful information, a funny moment, or a claim the viewer wants to verify with someone they know
• Ask a contentious question at the end to provoke comments and extend the video's lifespan
• Add a subtle detail or fast-moving information to encourage replays
• Stay within your niche — straying outside it activates negative signals even among your existing audience

For the broader picture of how the algorithm builds its decisions and distributes content through progressive phases, read TikTok algorithm & going viral. And for the complete guide to the platform, read The complete TikTok guide.