TikTok Algorithm Glossary

The For You Page is the default view when you open TikTok. Unlike Instagram's follower feed, the FYP shows content from accounts you've never followed, selected by TikTok's recommendation algorithm based on your watch history, engagement patterns, and geographic signals.

For creators, FYP placement is the primary growth mechanism. A video that stays on the FYP for 48–72 hours can add tens of thousands of followers. A video that never reaches the FYP is effectively invisible to non-followers.

FYP distribution is earned sequentially: TikTok tests each video with a small initial audience (the cold start cohort), measures their engagement, and expands distribution if the signals are strong. The FYP is not a lottery — it's a scored distribution queue.

A hook is the opening of a video that either stops a viewer from scrolling or fails to. It's not a category of content — it's a function. A question, a visual surprise, a counterintuitive claim, a dramatic action: any of these can be a hook if they create enough friction to interrupt the default scroll behavior.

TikTok measures hook effectiveness through first-second retention (what percentage of viewers stayed past the 1-second mark) and 3-second retention. Videos that lose more than 50% of viewers in the first second receive sharply reduced algorithmic distribution, regardless of what happens in seconds 4–60.

Common hook types: shock/visual spectacle, open loop (pose a question the viewer needs answered), authority statement (expert claims something surprising), problem identification (names a pain the viewer has), and curiosity gap (implies information the viewer doesn't have).

Hook delay is the gap at the beginning of a video before anything happens that gives a viewer a reason to stay. This includes: walking into frame, adjusting the camera, greeting the audience ('hey guys'), setting context ('so today we're going to'), or any setup before the actual hook.

Every second of hook delay loses a measurable percentage of viewers. Studies of TikTok content at scale show that videos with 0-second hook delay (content begins at frame 1) outperform structurally identical videos with 3-second hook delay by 30–60% on retention metrics.

The VoidRuns analyzer scores hook delay as a component of the hook score. A 4-second hook delay is one of the most common fixable issues in stuck accounts — content that is genuinely good is never distributed because the algorithm makes its quality judgment before the creator has said anything.

When you post a video, TikTok doesn't immediately show it to everyone. It first distributes the video to a small audience batch — typically your existing followers plus a limited non-follower sample. This is the cold start cohort.

TikTok measures this cohort's engagement: watch-through rate, completion rate, shares, comments, and re-watches. If the metrics are above the threshold for your niche and follower count, the algorithm passes the video to a larger cohort. Each expansion multiplies reach by roughly 5–10×.

The cold start cohort is why posting time matters: if your initial batch is shown the video at 3 AM when your target audience is asleep, they can't engage. Weak early engagement due to bad timing causes the algorithm to misclassify good content as low-quality. The cohort is also why follower quality matters — a follower base that never engages hurts cold start metrics.

Retention rate is TikTok's most important quality signal. A high retention rate tells the algorithm that viewers found the content engaging enough to not scroll away. A low retention rate — especially in the first 1–3 seconds — tells the algorithm the content failed to capture interest.

TikTok measures retention at multiple checkpoints: 1 second, 3 seconds, 25%, 50%, 75%, and 100% of video length. Completion rate (100% retention) is weighted heavily — it signals that the content delivered on its premise all the way through. Videos that are watched to completion are more likely to be re-watched, which generates loop plays.

For creators: optimizing retention means making every second earn the next. Hook delay kills early retention. Dead air kills mid-video retention. Weak endings kill completion rate. Each has a different structural fix.

The VoidRuns Viral Score is a 100-point composite of five independently scored dimensions: hook quality (30 pts), timing (20 pts), trend participation (25 pts), caption SEO (15 pts), and format consistency (10 pts). Weights reflect each factor's measured contribution to FYP distribution at the content level.

A score of 80+ (A or S grade) indicates content strategy that consistently earns broad FYP distribution. A score of 60–79 (B/C) indicates specific fixable gaps. A score below 60 typically signals structural problems in hook or timing that are preventing the algorithm from ever testing content quality.

The score is not a prediction of any individual video's performance — it's a diagnostic of the structural patterns in a content strategy. Strong execution of a high-scoring strategy will consistently outperform weak execution of a high-scoring strategy.

Dead air is the enemy of retention. On TikTok, the default viewer behavior is to scroll when nothing new is happening. A pause between sentences, a shot without motion, a moment of background noise while the creator thinks — all of these are drop-off triggers.

The editing principle that solves dead air: cut every moment where nothing is happening before you would instinctively cut it. Most creators leave 1–2 seconds of dead air per cut because it feels natural in the edit. Viewers experience those same 1–2 seconds as reasons to scroll.

Dead air is different from intentional pause, which is a tool for emphasis. A deliberate one-second pause after a surprising statement ('I made $80,000 last year. [pause] Selling digital stickers.') creates tension. Unintentional silence between takes, at the end of sentences, or during camera reframes is structural dead air.

A pattern interrupt is anything unexpected: a zoom cut, a text overlay appearing suddenly, a sound change, a visual cut to a different scene, a direct-to-camera break in voice-over. The purpose is to interrupt the default scroll behavior by introducing novelty into the visual stream.

TikTok's best-performing creators use pattern interrupts every 2–4 seconds. This isn't editing for style — it's editing for the physiological fact that human attention resets to stimuli changes. Regular pattern interrupts tell the brain 'something new is happening' repeatedly, which is neurologically equivalent to beginning the video multiple times.

Pattern interrupts are particularly valuable at predictable drop-off points: the 10-second mark (where viewers who stayed past the hook but aren't hooked often drop), the midpoint, and the 80% mark (where viewers considering whether to finish decide to scroll).

An open loop is a promise that hasn't been fulfilled yet. 'I'm going to show you the number one mistake that kills fitness results — and it's not what you think.' The viewer now has an open loop: a question that needs an answer before they can stop watching. Until the loop closes, leaving feels like leaving a conversation mid-sentence.

Strong open loops are specific and personal. 'You're making a mistake' is weak — it could apply to anyone. 'If you've ever wondered why your videos never get over 10,000 views, I found the reason' is strong — it speaks to a specific, recognizable pain that a specific audience has.

Open loops compound with watch time: the longer the loop stays open, the further into the video the viewer has traveled. Well-structured long-form content (60-second+ TikToks) uses multiple nested open loops — a primary loop that runs the whole video, and smaller loops that open and close every 15–20 seconds to maintain mid-video retention.

TikTok doesn't officially use the term 'shadowban,' but account-level distribution suppression is documented and measurable. Affected creators see videos drop from 50K+ average views to under 2K, with analytics showing FYP distribution near zero while follower-source views remain.

Common triggers: sudden posting frequency spikes (spam signal), niche pivots (account repurposing signal), use of flagged hashtags, sharp engagement ratio changes (bot detection), and community guideline violations (even minor or borderline content).

Recovery pattern: 3–7 day posting pause, followed by gradual return at 1 video/day, posting content structurally identical to pre-penalty content, no new hashtag experiments for 2–3 weeks. Most accounts following this protocol see measurable recovery in 2–6 weeks.

A trending sound is any audio clip (song, dialogue, original audio) experiencing rapid usage growth on TikTok. The algorithm uses sound as a content cluster signal — videos using the same sound are grouped and recommended together. When a sound is trending, its content cluster has active algorithmic promotion.

Sound trends follow a curve: discovery (low usage, no algorithmic boost), acceleration (rapid growth, strong boost, low competition), peak (maximum usage, maximum competition), saturation (declining boost, overcrowded pool), and death (no algorithmic signal). The acceleration phase is the optimal posting window.

Using a sound at peak or in saturation means competing in a pool of thousands of videos for the same algorithmic boost. Using the same sound 3–5 days before peak means entering a smaller pool while the boost is already active. Identifying sounds in acceleration requires monitoring the trending audio list and checking usage growth rate, not just current popularity.

Tier 1 — Peak Trend: A sound at maximum usage and maximum competition. The algorithmic boost is real but spread across millions of competing videos. Unless your content is exceptional, Tier 1 sounds produce average results for non-viral accounts.

Tier 2 — Acceleration Phase: A sound growing rapidly, not yet saturated, with active algorithmic promotion. This is the optimal posting window. The boost is comparable to Tier 1 but the competition pool is 10–100× smaller. Tier 2 sound use is the highest-leverage timing decision a creator can make.

Tier 3 — Declining or Niche-Specific: A sound past peak saturation, or an evergreen sound with low trend velocity. No active boost. Using Tier 3 sounds provides no algorithmic advantage but doesn't penalize content — the video competes on other signals (hook, retention, engagement).

Short TikTok videos (under 15 seconds) loop automatically. When a video loops, TikTok counts each loop as a play. A viewer who watches a 12-second video three times in a row generates three plays and signals to the algorithm that the content was compelling enough to re-watch.

Videos designed to loop — where the end transitions smoothly back to the beginning — accumulate loop play signals more aggressively. This is particularly common in satisfying process videos, satisfying visual sequences, and videos where the ending recontextualizes the beginning (making viewers want to rewatch).

Loop play data is one of TikTok's positive engagement signals. High loop rate → high re-watch value → algorithmic boost. Creators building short-form content (under 20 seconds) should specifically consider whether their ending naturally transitions back to their beginning.

A niche pivot is when a creator changes the primary subject matter of their content. A fitness creator starting to post cooking videos. A lifestyle creator switching to business content. A comedian pivoting to educational content.

The algorithm doesn't update its category profile for an account instantly. It built that profile over months of consistent content. A niche change means: removal from the existing content pool (losing algorithmic equity) before entering the new content pool at zero priority. Most creators see 60–90% reach reduction after a hard pivot.

The safer alternative: gradual transition with explicit bridges. Post 80% original niche, 20% new niche, for 4–6 weeks. Use bridge content ('GRWM for the gym' → 'my fitness routine GRWM' → 'my fitness routine') that maintains existing audience context while introducing new keywords. Hard pivots cost months of recovery time; gradual pivots cost weeks.

TikTok doesn't publish a 'trust score' officially, but its effects are measurable: accounts with stable, consistent posting history, clear niche definition, clean community guideline record, and organic engagement patterns receive broad FYP distribution. Accounts with erratic behavior, policy violations, or anomalous engagement patterns receive restricted distribution.

Trust score is built slowly and damaged quickly. An account that posts consistently for six months and never triggers a flag can see its trust score destroyed in 3–5 days by a sudden posting spike, a policy-adjacent video, or a hard niche pivot. Recovery to previous distribution levels takes weeks to months.

The practical implication: consistency is more algorithmically valuable than peak performance. An account that posts three times per week, every week, in the same niche, with the same format, will compound trust score advantages over time. An account that posts twelve times in three days, disappears for three weeks, and pivots niches is building nothing.

TikTok rewards consistency over volume. Posting one video per day, every day, builds a distribution profile that the algorithm learns to predict and prioritize. Posting seven videos in one day and nothing for a week generates spam signals and trust score degradation.

The algorithm builds a content cadence expectation based on your historical posting pattern. When you post on schedule, it has an active distribution profile ready. When you go silent, that profile degrades. The longer the gap, the more the next video is treated like a new account's first video.

Recommended minimum: 3–5 videos per week, on predictable days and times. The specific time matters less than the consistency. A creator who posts at noon every Tuesday, Thursday, and Saturday outperforms a creator who posts at peak time but unpredictably.

TikTok Search has become a significant discovery channel, particularly for Gen Z users who search TikTok instead of Google for tutorials, recommendations, and reviews. Videos with captions that match search queries surface in TikTok Search results, independently of the FYP algorithm.

Caption SEO is distinct from FYP SEO (which uses trending sounds, strong hooks, and high retention). Caption SEO is about choosing words that match what your target viewer types into TikTok's search bar. 'Leg day workout' is SEO-weak. 'Why your legs aren't growing despite working out' is SEO-strong — it matches a real search query.

The structure that works: lead with a search-intent sentence (the question or problem your viewer searches for), add 3–5 specific hashtags (not 20 generic ones), and avoid keyword stuffing. Keyword-dense captions without sentence structure are lower-quality signals than natural language that happens to contain high-value keywords.

Watch time is the aggregate measure of how much of your video viewers actually consume. TikTok weighs watch time per viewer (did they watch 10% or 90%?) and total watch time (how many total seconds has this video accumulated) as primary signals for distribution decisions.

Watch time and retention rate are related but different. Retention rate is the percentage who reached a given point. Watch time is the total seconds consumed across all views. A 30-second video watched by 1,000 people at 90% completion has 27,000 seconds of watch time. That signal tells the algorithm the content is high-quality and should reach more people.

Maximizing watch time: hook retention (keep viewers past the first 3 seconds), eliminate dead air (every second of nothing costs watch time), and build toward an ending that rewards completion (viewers who reach the end are more likely to loop or re-watch).

A Duet shows your video alongside the original in a split-screen format. A Stitch clips up to 5 seconds of another video and allows you to respond directly after it. Both formats are algorithmically advantaged because they inherit distribution context from the original video's audience and content pool.

The strategic use case: react to content that is already going viral, or to content in your niche that has established distribution. Your duet or stitch enters the same algorithmic neighborhood as the original — it gets shown to people who engaged with similar content. This is traffic arbitrage: you're borrowing distribution infrastructure that someone else built.

Duets and stitches also serve an authority-building function. Reacting to a viral question or misconception in your area of expertise ('let me respond to this as someone who actually works in the field') delivers credibility signals efficiently. The original creator did the audience-capture; you deliver the expert layer.

B-roll is any footage that isn't the main speaker — product close-ups, process footage, environment shots, screen recordings, text animations, or any visual that supports the narration. In TikTok context, B-roll serves as a pattern interrupt that resets attention and prevents drop-off.

Static talking-head video (one camera angle, creator speaks directly to camera for 30–60 seconds) is the lowest-retention format on TikTok. Adding B-roll every 2–3 seconds — even phone footage, screen recording, or stock video — transforms the attention experience from 'someone talking' to 'information delivery with evidence.'

The practical rule: if your voice is explaining something, your visual should be showing it (or showing something related). If you can't show it, cut to a text overlay. If you can't do either, insert a fast B-roll shot of anything visually interesting at that moment. The visual track should never stay static for more than 3 seconds.