01
The first second is not a hook, it is a filter
On a feed where the next video is one thumb-flick away, the opening frame is doing a single job: convincing someone that scrolling would cost them something. That is a lower bar than "be interesting" and a higher bar than "start the video".
The most common failure is a video that starts with a person walking into frame, adjusting the camera, or saying hello. None of those are the video. Start on the most arresting frame you have — visually or verbally — and let the context catch up.
- Open on movement, a visual anomaly, or a sentence that cannot be ignored.
- State the subject explicitly within the first couple of seconds. Ambiguity reads as irrelevance.
- Text on the opening frame works, but only if it can be read in the time it is on screen.
- Do not front-load your logo. Nobody has decided they care yet.
02
Design for the loop
TikTok replays automatically. A video whose end flows into its beginning gets watched more than once by the same person, and watch time counted that way is real watch time. This is a genuine structural lever that most people ignore.
The simplest version: end on the same visual you opened on. The stronger version: end on a line that makes the opening line mean something different, so a second viewing is actually rewarding rather than merely accidental.
- Avoid a trailing dead beat at the end. Cut on the last useful frame.
- Do not end with a long call-to-action; it is the part people scroll away from and it breaks the loop.
- A hard cut back to the opening composition is the cheapest loop you can build.
03
Build a format, do not chase a trend
Trends are a discovery mechanism with a short half-life. An account built entirely on them is rebuilt from scratch every few weeks and never accumulates anything. A format — a repeatable structure that is recognisably yours — accumulates, because returning viewers know what they are getting and new viewers can immediately see there is more.
The strongest accounts usually have one format they run relentlessly and a small amount of variation around it. The format can be extremely simple. What it cannot be is different every time.
- A format is a structure, not a topic. "Cooking videos" is a topic. "One ingredient, three ways, in sixty seconds" is a format.
- Use trends inside your format rather than instead of it.
- Recognisability compounds: same framing, same font, same rhythm.
- If a post does well, make more of that exact thing before you move on. Most people abandon a working format far too early.
04
Sound is half the platform
TikTok is an audio-first feed in a way the other short-form surfaces are not. Sound choice affects discovery, and a video cut to the beat of its audio simply feels better than one laid over it.
That said, muted viewing is common enough that anything essential must also exist visually. Captions are not an accessibility afterthought here; they are a retention mechanism.
- Cut on the beat. It is the cheapest production upgrade available.
- Burn captions in rather than relying on auto-captions, so you control placement and timing.
- Silence immediately before a reveal does more work than louder music during it.
05
The comments are a content pipeline
A question in the comments of a video that performed is a pre-validated idea for the next one. Replying with a video is a native feature specifically because this loop works, and it is close to free content that is guaranteed to interest at least the people who already engaged.
It also has a compounding effect on the original video, which keeps receiving engagement while the reply circulates.
- Answer the most-liked question with a video, not a text reply.
- A mildly controversial but honest opinion generates more useful comments than a safe one.
- Pin the comment that best frames the video for a new viewer.
06
What does not work as well as people claim
Posting volume for its own sake, hashtag stacking, and "engagement bait" that does not survive contact with the actual video are all overrated. Volume without a working format just produces a larger pile of posts that did not land.
The one thing genuinely worth doing at volume is testing hooks — the same underlying video with a materially different opening — because the opening is the variable with the widest outcome range.
- Hashtags help categorise; they are not a growth strategy.
- Deleting posts that underperform does not help and loses you the data.
- Reposting the identical file rarely works; a genuinely different hook on the same body sometimes does.