AI prompts for faceless YouTube videos

The main long-form script, written to be read aloud

Write a [LENGTH]-minute faceless YouTube script about [TOPIC] for an audience of [AUDIENCE].

Rules:
- Open with a 12-second hook that states the payoff, not the topic.
- Write for the ear: short sentences, no subclauses, no words a narrator would stumble on.
- Every 45 seconds, add an open loop that is paid off later in the script.
- Never say "in this video" or "let's dive in".
- Mark each paragraph with a [VISUAL: ...] note describing the b-roll or graphic that should be on screen.
- End with one specific next-step CTA, not "like and subscribe".

In practice: Write a 9-minute faceless YouTube script about how shipping containers broke the global economy for an audience of casual business-history viewers.

Fixing the middle third where faceless videos drop off

Here is a faceless YouTube script: [PASTE SCRIPT]

Act as a retention editor. Find the three weakest 30-second stretches — the places a viewer would swipe away. For each one, tell me exactly why it sags (no new information, no tension, repeated point, or a visual that cannot be shown), then rewrite that stretch. Do not rewrite the rest. Keep the total runtime within 10% of the original.

In practice: Paste a finished 8-minute draft and ask for the three sag points before you ever record narration.

Turning words into a b-roll plan you can actually source

Take this script and produce a shot list as a table: timestamp, line of narration, visual description, and source type (stock clip, AI-generated clip, screen recording, motion graphic, or still image with a slow push-in).

Constraints:
- No shot may be longer than 6 seconds.
- Prefer specific, searchable stock terms over abstract ones — "container ship at night, aerial" not "global trade".
- Flag any line that has no filmable visual so I can rewrite it.

Script: [PASTE SCRIPT]

In practice: Run this before you buy a single stock clip — it usually kills 20% of the shots as unsourceable.

Making a synthetic voice sound like a documentary narrator

Rewrite this script for text-to-speech narration by [VOICE STYLE — e.g. calm male documentary narrator].

- Break long sentences at natural breath points.
- Insert ellipses or line breaks where the voice should pause for emphasis.
- Spell out numbers, acronyms, and any word the engine will mispronounce, phonetically in brackets.
- Mark 5-8 words for emphasis with **bold**.
- Do not change the meaning or add new claims.

Script: [PASTE SCRIPT]

In practice: The phonetic-spelling rule alone removes most re-render cycles in ElevenLabs and Murf.

Packaging the video so the script actually gets watched

Based on this script, give me 10 YouTube titles and 5 thumbnail concepts.

Titles: under 60 characters, no clickbait the script does not deliver, and at least three that lead with a number or a surprising specific.
Thumbnails: describe the single focal image, the 2-4 words of overlay text, and the emotional read. No faces — this is a faceless channel, so the image must carry the tension alone.

Script: [PASTE SCRIPT]

In practice: Ask for the thumbnail concepts before you finalise the title — the strongest pair usually emerges from the image.