TikTok is viewer-first, not follower-first

TikTok does not have one universal feed ranking for everyone. Each viewer gets a personalized stream shaped by what they watch, skip, replay, like, share, comment on, save, follow, and mark as not interested.

TikTok also reads video information such as captions, sounds, hashtags, and topic signals. Language, country, and device settings help with context, but TikTok says these are generally weaker signals than behavior that shows what a viewer actually chose to watch.¹

This is why a large following can help a post get an initial audience without guaranteeing wider recommendation. TikTok has said follower count and previous viral performance are not direct recommendation factors.¹

• •Watch time and completion: did the opening earn enough attention for the viewer to stay? TikTok has specifically described finishing a longer video as a strong interest signal.¹
• •Rewatching: did the idea, reveal, or loop make the viewer watch again?
• •Shares and saves: did the video feel useful, surprising, or relevant enough to keep or send?
• •Comments and follows: did the creative create a relevant reaction or a reason to see more?
• •Profile visits, clicks, and conversions: did attention become intent? These are business outcomes, not substitutes for watch quality.

No single percentage is a guaranteed pass. A high completion rate on a weak offer is not the same as a video that earns useful actions. Read the metrics together and compare them with the goal of the ad.

1. 01First 1 to 3 seconds: does the first frame and first line give someone a reason not to swipe?
2. 02Hold and average watch time: does the ad keep delivering proof, tension, or a useful next beat?
3. 03Completion and rewatches: does the structure earn the ending or invite a second look?
4. 04Saves, shares, comments, and follows: does the idea create a useful or social reason to act?
5. 05Clicks and conversions: did the attention turn into demand for the brand?

The TikTok report treats algorithm signals as creative predictions, not live account analytics. It asks whether the ad is likely to earn the viewer behaviors TikTok can observe.

The Ad Bench methodology:The points below reflect The Ad Bench's current scoring rubric and creative-review approach. They are designed to identify creative risk before media spend, not predict or guarantee campaign performance.

• •Hook scoring focuses on the first 1 to 3 seconds.
• •Completion is treated as a strong interest signal, not a fixed distribution threshold.
• •Shares, saves, comments, follows, clicks, and conversions are read as different kinds of intent.
• •Follower count is not treated as proof that the creative will travel beyond the existing audience.

A useful ad delivers one concrete insight immediately, shows proof, explains the fix, and ends with a light next step. For The Ad Bench, that could be: “Your ad can have a great hook and still fail because proof arrives too late.” Then show the example and invite the viewer to put the next ad on the bench before spending behind it.

• •Posting for “the algorithm” without a specific viewer payoff.
• •Slow branded intros that spend the opening before the point arrives.
• •Broad topics with no clear audience or reason to care.
• •Clickbait that earns a view but causes an early skip.
• •Copying a trend without adding a relevant point of view.

This guide summarizes TikTok's public explanation of its recommendation system. Policies and ranking systems change, so treat the report as a creative planning tool, not a guarantee of reach.

1. TikTok Newsroom. "How TikTok recommends videos #ForYou." Accessed August 24, 2026. newsroom.tiktok.com
2. TikTok Help Center. "How TikTok recommends content." Accessed August 24, 2026. support.tiktok.com
3. TikTok. "Introduction to the TikTok recommendation system." Accessed August 24, 2026. tiktok.com/transparency
4. The Ad Bench methodology, current scoring rubric. Reviewed August 24, 2026.