The 4 Claude Prompts Running an $8K Faceless YouTube Channel (Day 16)
✳
Mike Means Business
Day 16 · AI Hacks · The Prompt Pack

The 4 Claude Prompts Running an $8K Faceless YouTube Channel (Day 16)

You commented HACKS. Here's the pack. Four prompts, one lazy-mode one-shot that chains all of them. The reel showed you what's possible. This post shows you how it actually works.

▸ The honest version of the math

Faceless YouTube in 2026 is real. It's also tighter than the gurus claim. 600K views at a realistic $8–$12 RPM lands you $4,800–$7,200 from AdSense. The $8K figure only stacks once you've layered in affiliates, one sponsor deal, and a small digital product. That takes three to four months minimum, not three weeks. The system works. The timeline is just longer than TikTok makes it look.

The 4 Prompts

Prompt 01 Niche Picker

Runs a data-driven niche audit across up to five candidates. Kills the vibe-pick problem before it kills your channel.

claude — niche-picker.txt
You are a YouTube niche analyst helping me pick a faceless channel niche.

INPUT
- Candidate niches: [LIST 3–5 NICHES]
- Target audience region: [US / UK / global]
- Hours per week I can commit: [N]
- Existing audience I can port (any platform): [DESCRIBE OR "NONE"]

For each candidate niche, return:
1. RPM band (low / mid / high) with one-line reasoning
2. Competition level (low / medium / high) with one signal (e.g. saturated keyword density)
3. Content repeatability score 1–10 — how many videos on the same template before it goes stale
4. Audience size potential — back this with one search-trend or platform stat
5. Monetisation paths beyond AdSense realistic in year 1 (affiliate / sponsor / digital product / membership)
6. One untapped content angle no one is doing in this niche

THEN
- Recommended niche given my time + region + existing audience
- Why this one wins over the others
- First 5 video titles I should publish in week 1
Tweak If the analyst hedges, add "Pick one. Don't tell me 'it depends'." at the end.
Prompt 02 Script Writer

Replaces a $200 scriptwriter. Forces the pattern-interrupt hook, opens loops early, writes for the ear. Includes five thumbnail concepts and three title variations.

claude — script-writer.txt
You are writing a faceless YouTube script for a busy, skeptical viewer.

INPUT
- Topic: [TOPIC]
- Niche: [NICHE]
- Target length: [5 / 8 / 12 minutes]
- Tone: [educational / story-driven / list-based]
- One proof point I can reference (real data, real number, real source): [INSERT]

REQUIREMENTS
- Hook in the first 20 seconds. Pattern interrupt. No greeting.
- Open the loop early — tease the payoff but hold it back.
- Spoken English. Short sentences. Written for the ear, not the page.
- Insert [PAUSE] markers at natural breath points.
- Insert [VISUAL: short description] tags at every scene change.
- End with a soft CTA. No begging. No "smash that like button."

DELIVER
1. Full word-for-word script
2. Estimated runtime at 150 wpm
3. 5 thumbnail concepts (image idea + text overlay for each)
4. 3 title variations: one curiosity-led, one keyword-led, one emotional
Tweak Paste 1–2 transcripts from your favourite creators in the niche before running. The script will calibrate to that voice.
Prompt 03 Pipeline Brief

Converts your finished script into a production brief: shot list, voiceover notes, music brief, caption style. Tells you exactly how long the edit will take.

claude — pipeline-brief.txt
You are turning a finished YouTube script into a production brief.

INPUT
- Full script (with [VISUAL] tags): [PASTE]
- Voice (ElevenLabs ID or description): [INPUT]
- Editor I'm using: [CapCut / Premiere / DaVinci]

DELIVER
1. Shot list — every [VISUAL] tag from the script with a Pexels-ready stock search term
2. Voiceover notes — pace, energy level, emphasis points (mark per section)
3. Music brief — genre, BPM range, where it ducks for VO
4. Caption style — font, position, animation pattern
5. Estimated total assembly time at this scope (be honest, not optimistic)
Tweak If you're new, ask it to add "production difficulty 1–5" next to each shot. Kill the hard ones until you have practice.
Prompt 04 Metadata Writer

Writes the full upload package. Title, description, 15 tags, chapter markers, three pinned-comment options. Run TubeBuddy or VidIQ after for keyword sanity-check.

claude — metadata-writer.txt
You are a YouTube SEO specialist writing the upload metadata.

INPUT
- Topic: [TOPIC]
- Niche: [NICHE]
- 2-sentence script summary: [SUMMARY]
- Target keywords (if known): [LIST OR "FIND THEM"]

DELIVER
1. Title under 60 characters — main keyword + curiosity trigger
2. Description, 180–220 words, SEO-optimised, includes timestamps + 1 CTA + 3 internal links to my other videos (placeholders ok)
3. 15 tags — mix of broad + long-tail + branded
4. 5 chapter titles with suggested timestamps
5. 3 pinned-comment options that boost engagement

Optimise for click-through rate and watch time. Not virality theatre.
Tweak Check the title in TubeBuddy or VidIQ before uploading. If the search volume is zero, the metadata won't save the video.

★ The Hero Asset
The Lazy-Mode One-Shot
If you only paste one thing into Claude, paste this. Chains all four phases in a single call. Niche check, full script, production brief, upload metadata. One prompt, one session.
claude — one-shot-pipeline.txt
You are running the full faceless YouTube production pipeline in one call. Output sequentially with clear section headers. Do not stop until all four phases are complete.

INPUT
- Channel niche or candidate niches: [LIST OR ONE NICHE]
- Audience region: [US / UK / global]
- Video topic (or "PICK" and you choose based on the niche): [TOPIC OR "PICK"]
- Video length: [MINUTES]
- Tone: [educational / story / list]
- ElevenLabs voice: [VOICE NAME OR "PICK"]
- One proof point I can reference: [INSERT OR "INVENT A REALISTIC PLACEHOLDER"]

PHASE 1 — NICHE CHECK
If I gave you multiple niches, pick the best one using RPM band, competition, content repeatability, monetisation paths beyond AdSense, and one untapped angle. Name the winner + one-line why. If I gave one niche, confirm it and flag any obvious risk.

PHASE 2 — SCRIPT
Write the full faceless YouTube script. Pattern-interrupt hook in the first 20 seconds, open loop, spoken English, [PAUSE] markers, [VISUAL] tags per scene, soft CTA. Include 5 thumbnail concepts and 3 title variations after the script.

PHASE 3 — PRODUCTION BRIEF
Convert every [VISUAL] tag from Phase 2 into a Pexels-ready shot list. Add voiceover notes (pace, emphasis) and a one-line music brief.

PHASE 4 — METADATA
Write the title (under 60 chars), description (180–220 words with timestamps + CTA), 15 tags, 5 chapter titles, 3 pinned-comment options.

FORMAT
Use ## H2 section headers for each phase. Put every final deliverable in a fenced code block so I can copy without reformatting. Be specific, not generic. If a section needs a decision I haven't given you, make the decision and flag it with a 🟡 so I can override later.
Power move Paste this prompt once into a Claude Project and set it as the system prompt. Every new chat in that project becomes a one-shot pipeline. You never paste it again.

What to Do Next

1
Run the lazy-mode prompt now. Pick your top two niche candidates. Let Claude make the call. Don't spend three hours researching what the prompt does in 60 seconds.
2
Publish video one. Don't optimise it. Volume is the teacher here, not strategy. The algorithm tells you what's working once you've given it data. You need five videos minimum before you have anything worth reading.
3
Reply to this email with your channel URL when you've shipped video five. Mike will give you one real piece of feedback. Not a generic response. One specific thing to fix next.
Automate the unscalable.