"The anatomy of a TikTok that sells: hooks, beats, and captions"

Watch twenty product TikToks that actually converted and they start to rhyme. Different products, different creators, same underlying machine. Here's the machine.

The hook (0–2 seconds)

The first shot has one job: make skipping feel like a loss. The reliable shapes:

• The unfinished claim. "This is why your under-eyes never improve…" — the sentence forces a second shot.
• The unusual object. Something in frame you can't name yet. Curiosity is a hook.
• The bold on-screen line. Big text making a specific promise or calling out a specific person: "POV: you've tried everything for your puffy eyes".

Notice what's not here: the product's name, a logo, a brand jingle. The hook sells the next three seconds, nothing else.

The beats (the middle)

A beat is one shot with one job. Winning videos are surprisingly disciplined about this — each beat runs a second or three, then cuts. Common sequences:

1. Context — the problem, shown not told. Tired eyes in a mirror. A dog that hates water.
2. The product enters — unboxed, picked up, put on. Handled like an object someone owns, not a hero shot.
3. Use — the product doing its thing, in real space, real light.
4. The turn — the before/after, the reaction, the payoff beat. This is usually the longest single beat, because it's the one the whole video was for.
5. The close — one line about where to get it. Short. Often just a caption.

The pacing is the invisible part people get wrong when they copy by eye. A beat that runs twice as long as it should reads as padding; the original's timing was measured against attention, and attention is unforgiving.

The caption timeline

On-screen text isn't decoration — it's the second script. Most winning videos run captions that carry the argument even with the sound off: the hook line, a claim per beat, the close. Placement is consistent (usually upper-middle or center), the wording is casual, and the CTA keyword at the end is deliberate.

When you adapt a reference, keep the caption rhythm — when text appears and how long it stays — even as you change the words to fit your product. The rhythm is part of why the original worked.

Copy the skeleton, not the skin

The wrong way to use a winning video: re-film it shot for shot and hope. The right way: extract its structure — hook type, beat list with timings, caption timeline — and rebuild each beat around your own product honestly.

That extraction is tedious to do by hand, which is why we automated it. OnlyPost's Script Generator takes a TikTok link and returns exactly this breakdown — every beat with measured timings, every caption verbatim — ready to shoot or to turn straight into a finished video.