The Anatomy of a TikTok Video
FYPNow Research · Published August 4, 2026
Almost every piece of TikTok advice is about what to say. Much less is written about what a video is physically built like — how long it runs, how quickly the opening resolves, how often it cuts, and whether any of that tracks with how people actually respond. We measured those things across our catalog of analyzed videos. Some of what came back contradicts the standard advice.
The typical video
Medians, not averages — TikTok performance is power-law distributed, so a handful of runaway videos drag any mean upward. These describe the middle of the catalog.
Video length
29s
Half of all videos run shorter than this
Hook length
3.0s
Time before the opening premise resolves
First value delivered
5.5s
When the viewer gets the first payoff
Scene count
5
Distinct visual cuts per video
Average scene
6.5s
How long a single shot holds
Call to action starts
25.7s
Among videos that include one
Like rate
6.41%
Likes as a share of views
Save rate
0.58%
The strongest algorithmic signal
Share rate
0.38%
Shares as a share of views
Comment rate
0.14%
Comments as a share of views
Where engagement rates actually land
Engagement rate is likes, comments, shares and saves divided by views. A single number is not much use without knowing the spread, so here is the distribution across the catalog.
| Percentile | Total engagement rate |
|---|---|
| 25th | 4.97% |
| 50th (median) | 7.48% |
| 75th | 10.6% |
| 90th | 13.65% |
For the same distribution split by content niche, see our TikTok engagement rate benchmarks.
By video length
Save rate rises steadily as videos get longer, roughly doubling between the shortest and longest bands we can report on. Engagement rate moves far less. Longer videos in this corpus are saved more often than short ones — the opposite of the "keep it under 15 seconds" advice repeated across most TikTok guidance.
| Band | Engagement | Save rate |
|---|---|---|
| Under 15s | 9% | 0.63% |
| 15–30s | 9.9% | 0.78% |
| 30–60s | 8.5% | 0.79% |
| 60–120s | 10% | 1.29% |
By hook length
The same pattern shows up in the opening. Videos that take longer to resolve their hook are saved and shared more often than videos that resolve it in the first two or three seconds. The very fastest and very slowest hooks are excluded here — too few creators used them to report on.
| Band | Engagement | Save rate |
|---|---|---|
| 2–3s | 8.6% | 0.69% |
| 3–4s | 9.3% | 0.82% |
| 4–6s | 10.1% | 1.34% |
By scene count
Editing density does not move in one direction. Videos with a single held shot and videos cut into eleven or more scenes both outperform the middle of the range on engagement and saves. There is no clean "more cuts is better" line in this data.
| Band | Engagement | Save rate |
|---|---|---|
| 1–2 scenes | 9.6% | 0.98% |
| 3–5 scenes | 8.4% | 0.78% |
| 6–10 scenes | 8.9% | 0.83% |
| 11+ scenes | 11.4% | 1.15% |
By audio type
Original audio and licensed tracks perform identically on every measure we captured — engagement, saves and shares all land within a rounding error of each other. Whatever separates a video that works from one that does not, in this corpus it is not the choice between your own audio and a trending sound.
| Band | Engagement | Save rate |
|---|---|---|
| Original audio | 8.1% | 0.76% |
| Licensed track | 8.1% | 0.77% |
See these numbers for your own videos
Paste any TikTok video or profile and FYPNow runs the same analysis behind this study — hook timing, scene pacing, and how the engagement breaks down.
Prefer to explore first? Create a free account
How this was measured
Every video in the catalog runs through the same automated pipeline. Scene boundaries and hook resolution are detected from the video itself rather than self-reported; engagement, save and share rates come from the metrics attached to each post. Videos below a minimum view threshold are excluded, because engagement rate on a post with almost no reach is division by noise rather than a signal.
Bands that too few creators contributed to are not reported at all, rather than shown with a caveat. That is why the longest videos and the fastest and slowest hooks are absent from the tables above — we could compute a number for them, but not one worth printing.
What this cannot tell you
These are descriptions, not causes. The catalog is clustered — a modest number of accounts contribute a large share of the videos, and those accounts have house styles that show up in every band they appear in. When longer videos are saved more often, that may be about length, or it may be about who tends to make longer videos. Nothing here isolates one from the other, and no line above should be read as “do this and your saves will rise.”
The catalog is also not a random sample of TikTok. It is made up of creators that FYPNow users chose to track, which skews toward accounts already performing well enough to be worth watching. Read these as benchmarks for videos that found an audience, not for everything posted to the platform.
Frequently asked questions
How long should a TikTok video be?
The median video in our catalog runs 29 seconds, so half of everything analyzed is shorter than that. But length and performance do not move together the way most advice assumes: save rate climbs steadily as videos get longer, from 0.63% on videos under 15 seconds to 1.29% on videos between 60 and 120 seconds. Engagement rate barely shifts across the same range. If saves are what you are optimizing for, short is not automatically better.
How long should a TikTok hook be?
The median hook resolves at 3.0 seconds. In this catalog, videos whose hook runs 4 to 6 seconds show a higher save rate (1.34%) than those resolving in 2 to 3 seconds (0.69%). That is a description of what we observed, not a rule — the videos with slower hooks are also longer and made by a different mix of creators.
Does using original audio beat a trending sound on TikTok?
In our catalog they are indistinguishable. Original audio averages 8.1% engagement, a 0.76% save rate and a 0.46% share rate; licensed tracks average 8.1%, 0.77% and 0.48%. Every measure lands within a rounding error. Whatever separates a video that works from one that does not here, it is not that choice.
How many cuts should a TikTok video have?
The median video has 5 scenes with an average shot holding 6.5 seconds. Engagement does not rise cleanly with editing density: single-shot videos (9.6%) and videos with 11 or more scenes (11.4%) both outperform the 3-to-5-scene middle (8.4%). There is no straight line between more cuts and better performance.
What is a good TikTok engagement rate?
In this catalog the median video reaches 7.48% total engagement, the top quarter clears 10.6%, and the top 10% clear 13.65%. For a niche-by-niche breakdown, see our TikTok engagement rate benchmarks.
How was this measured?
Every video was run through the same automated analysis: scene boundaries and hook resolution are detected from the video itself, and engagement, save and share rates come from the metrics attached to each post. The catalog is made up of creators tracked on FYPNow, which skews toward accounts someone thought worth watching, so treat these as benchmarks for videos that already found an audience rather than for TikTok as a whole.