Wise Songs — YouTube Video ROI Analysis

Context

We generate educational songs (Aesop's fables, mental models, STEM nursery rhymes, GRE vocabulary, cerebral/philosophical songs) via Suno and have a pipeline from lyrics → suno generation → video → YouTube publish. This document evaluates the ROI of different video production approaches and identifies where to focus effort.

1. Song Categories and Their YouTube Opportunity

2. Video Format ROI

Static Image + Audio (lowest effort)

• Build time: 30 minutes (ffmpeg automation, one-time setup)
• YouTube retention: ~15–25%
• Algorithm signal: Weak — will not get recommended
• Value: Permanent indexed asset for exact-title searches. Placeholder only.
• ROI: D — don't build as end product

Lyric Video — verse-timed text overlay (recommended v1)

• Build time: 1 day to build pipeline; ~5 min per song after
• YouTube retention: ~55–70% (people follow along with words)
• Algorithm signal: Strong — watch time signals quality to YouTube
• Value: Works as classroom tool (teachers embed it), parents use with kids to teach reading alongside listening
• ROI: B+ — viable and buildable now with ffmpeg + SRT

Animated Lyric Video — word-level highlighting (v2 target)

• Build time: 2–3 days to build pipeline (requires Whisper transcription → word timestamps → highlight animation)
• YouTube retention: ~65–80%
• Value: Kids follow along word-by-word, high educational signal, strong differentiation from static competitors
• ROI: A — best format achievable without animation budget

Full Animation (eventual)

• Build time: Weeks per song, significant budget
• ROI: A+ for children's content specifically — Cocomelon-tier engagement — but not achievable without investment
• When: After channel validates traction. Prioritize the highest-performing songs first.

3. The Critical CPM Insight

Content tagged "Made for Kids" on YouTube is subject to COPPA restrictions — personalized ads are disabled, dropping CPM to $0.25–$1.00 RPM.

Content NOT tagged "Made for Kids" targeting teens and adults (GRE vocab, mental models, cerebral songs) earns $5–$25 RPM — 10–25x more per view.

Implication: Run two tracks:

GRE vocabulary songs with near-zero competition at 10–25x CPM are the highest-ROI content we can produce right now for direct YouTube revenue. A GRE channel with 100K views/month earns ~$1,000–$2,500/month. A children's channel needs 1–2.5M views/month to earn the same.

4. What Makes a Video Valuable to Others

Teachers and educators

• Need an embeddable video they can put in a lesson plan
• A lyric video with visible words is a teaching tool; a static image is just audio
• Aesop's fables are literal K-8 curriculum content — teachers are actively searching for song versions
• STEM songs (atomic physics, DNA) align with middle school science curriculum

Parents

• Search YouTube with children looking for specific topics
• Lyric video keeps children engaged longer (reading alongside listening)
• Song compilation playlists auto-play during car trips, meals, quiet time

Students (GRE/SAT)

• Songs are mnemonics — more memorable than flashcard drilling
• Word + definition + contextual example visible on screen = the product
• No competing music channel exists in this space

The lead generation mechanism

YouTube video description → wisesongs.supernal.ai
  → "Download free lyrics PDF" (email capture)
  → Newsletter / Substack subscription
  → Notified when new songs published

The video is discovery. The website is conversion. The email list is retention.

5. Suno Technical Limits

• Extensions add ~1 minute each; can chain 2–3 extensions cleanly before quality degrades (shimmer artifacts, tonal drift)
• Practical maximum with extensions: 10–11 minutes of usable audio
• No "movie mode" exists — Suno generates songs, not long-form audio narratives
• For compilations (20–45 min YouTube videos): generate songs individually, assemble in CapCut / DaVinci Resolve

Implication: We do not need Revide or other tools to work around a Suno length limit. Our songs are 1:30–4:00, well within the 8-minute cap. For compilation videos, we build them ourselves from individual songs.

6. Recommended Production Stack

Not needed yet: YouTube Data API auto-upload. Manual upload first batch — metadata (title, description, tags) matters more than automation. Automate after validating what gets traction.

7. Recommended Priority Order

1. Build lyric video pipeline (suno_video.py): ffmpeg + verse-timed SRT → MP4. Ship the 10 songs currently in video_gen stage.
2. Publish GRE vocabulary songs first — highest CPM, zero competition, lowest production bar (kinetic typography is sufficient)
3. Publish STEM nursery rhymes — curriculum-aligned, teachers actively search these
4. Publish Aesop's Fables — high search volume, "song" format is unclaimed
5. Build Whisper-aligned word highlighting (v2) — after validating v1 traction
6. Evaluate Revid.ai — if budget allows, try their Suno-to-video pipeline on 3–5 songs
7. Build monthly compilation — once 8+ songs are live, bundle into a 20–30 min compilation

8. Success Metrics