How to write four video hooks from one demo
Turn one screen recording into four video openings, each with a first shot and the proof it needs. Copy the prompt and see a tested output.
Jump to
1. List the footage you have2. Generate four openings3. Choose by proof, then test like for likeCreate four video hooks from one screen recording. Each gets a first shot and required proof; none is called a winner.
What you need: a horizontal recording, three crop previews, and the failed crop.
1. List the footage you have
List only existing footage.
Source: One horizontal screen recording
Final crops: 16:9, 1:1, and 9:16
Required proof: The same button remains visible in every final crop
Failure shot: Earlier 9:16 preview clips the button
Shared body: Show crop guides, fix 9:16, show all 3 finals
Unavailable: Face, speaker, voiceover, new recording, new assets2. Generate four openings
Paste into a fresh chat; replace the example with verified footage facts.
Task
Write four short video openings for one existing product-demo video.
Context
Available footage: one horizontal screen recording only.
Available proof: three crops made from that same recording—16:9, 1:1, and 9:16—with the same required button preserved in each final crop.
Available failure shot: an earlier 9:16 preview clips the required button at the edge.
Shared body and payoff: show the three crop guides, correct the failed vertical crop, then show all three final crops with the button visible.
No face, speaker, new recording, or new asset is available.
Output format
Return exactly four numbered options. For each include:
Opening: [12 words or fewer]
First shot: [a shot available in the context]
Required proof: [what the existing footage must show to earn the opening]
Rules
- All four options must use the same available footage, body, and payoff.
- Do not add a face, speaker, voiceover requirement, new recording, or new asset.
- Do not call any option proven, winning, viral, or high-performing.
- Make every claim visibly checkable in the existing footage.
- Say the required button is visible; do not generalize this to every control or the whole interface.
- Good: “Three crops, same button visible.” Bad: “No hidden controls.”
- If an opening mentions failure, use the failed 9:16 preview as its first shot.
- Replace the example facts above with verified facts from your own footage before using this on a real video.Actual fresh-chat test output after the narrow-claim check:
1.
Opening: One recording. Three formats. Same required button visible.
First shot: The three final crops shown together.
Required proof: The 16:9, 1:1, and 9:16 crops all show the required button visible.
2.
Opening: The vertical crop failed—then the required button came back.
First shot: The failed 9:16 preview clipping the required button at the edge.
Required proof: The footage shows the clipped preview corrected, followed by all three final crops with the required button visible.
3.
Opening: Watch one screen recording fit three formats without losing the button.
First shot: The original horizontal screen recording.
Required proof: The crop guides produce 16:9, 1:1, and 9:16 versions with the required button visible.
4.
Opening: This 9:16 crop nearly cut off the required button.
First shot: The failed 9:16 preview clipping the required button at the edge.
Required proof: The corrected 9:16 crop restores the required button, and the final three crops keep it visible.3. Choose by proof, then test like for like
Choose the candidate with the clearest existing first shot and proof. Read it once—or preview it at final overlay size—and polish its wording without changing the claim.
When comparing another opening, keep the body, payoff, edit length, audience, placement, format, and CTA stable. Review retention and qualified actions together; do not invent a pass threshold.
Finish check: the first shot exists, the body visibly earns the opening, and every variation uses the same footage and payoff.
Why this step matters
Xu and Wang’s August 2026 Frontiers in Communication study analyzed 3,392 Mandarin, speech-bearing Douyin cultural-heritage videos from a September 2024 crawl. Negative title sentiment and a title-to-transcript semantic gap were not confirmed after controls; held-out prediction was approximately zero.
Higher Title Semantic Density—an embedding-norm proxy the authors interpret as title focus—was associated with likes and comments in adjusted models. This observational, post-hoc result does not prove an effect. The study used raw counts, lacked key exposure controls, and did not analyze opening frames, music, or edits; it cannot validate all marketing hooks.
Doing this across many accounts?
EntireFeed makes short-form video at volume and posts it on real TikTok, Instagram and YouTube accounts. Tell us what you are posting and where, and we will say plainly whether we can help.