I Scraped 1,300 Videos About "Faceless YouTube Channel." Here's What the Data Says.
I've been building YouTube tools at TubeAlfred and wanted to understand the faceless YouTube channel niche properly. Not from watching a few videos. From data.
So I built a scraper, collected everything I could, and analyzed it.
This post covers the full process: what I scraped, how I scraped it, what I found, and what I think it means.
Why This Niche
Every other video on YouTube promises you can make $10K/month without showing your face, using AI to do everything. The faceless YouTube space is one of the fastest growing topics on the platform right now.
I wanted to know:
- What tools do these creators actually use vs. what they promote?
- How do they make money?
- What video formats perform best?
- What are the viewers actually asking for?
Not opinions. Data.
The Scraping Process
I started with 75 search keywords. All variations of terms people actually search when they're trying to start a faceless YouTube channel.
The keywords fall into a few clusters:
Channel ideas (33 keywords): faceless youtube channel ideas, faceless youtube channel ideas 2025, faceless youtube channel ideas for girls, faceless youtube channel ideas aesthetic, faceless youtube channel ideas motivation, faceless youtube channel ideas to make money, faceless youtube channel ideas using ai, best faceless youtube channel ideas 2026, 10 faceless youtube channel ideas 2026, faceless youtube channel niches, faceless youtube shorts ideas, faceless youtube shorts niches, and more.
Automation (12 keywords): faceless youtube automation, faceless youtube automation 2025, faceless youtube automation ai, faceless youtube automation channel ideas, faceless youtube automation course, faceless youtube automation full course, faceless youtube automation n8n, faceless youtube automation niches, faceless youtube automation shorts, faceless youtube automation step by step, faceless youtube shorts automation channel, and more.
AI tools (12 keywords): faceless youtube channel ai, faceless youtube channel ai agent, faceless youtube channel ai automation, faceless youtube channel ai free, faceless youtube channel ai guide, faceless youtube channel ai shorts, faceless youtube channel ai tools, faceless youtube channel ai video, faceless youtube channel ai voice, faceless youtube channel using ai, faceless youtube channel with ai 2025, faceless youtube channel with free ai tools.
Monetization (10 keywords): faceless ai youtube channel monetization, faceless voiceless youtube channel monetization, faceless youtube channel get monetized, faceless youtube channel monetization, faceless youtube channel monetization 2025, faceless youtube channel monetization tamil, faceless youtube channel not monetized, faceless youtube channels that make money, faceless youtube shorts monetization, no face youtube channel monetization.
Shorts (5 keywords): faceless youtube shorts, faceless youtube shorts ai, faceless youtube shorts channel, faceless youtube shorts channel ideas, faceless youtube shorts channel using ai.
Other (3 keywords): faceless youtube, faceless youtube channel, faceless youtube channel tutorial.
For each keyword, I collected up to 100 videos from YouTube search results. Each keyword returned roughly 98-100 videos.
The Raw Dataset
After deduplication (many videos rank for multiple keywords), I ended up with:
| Metric | Count |
|---|---|
| Unique videos | 1,421 |
| Unique channels | 839 |
| Comments | 116,214 |
| Unique commenters | 100,608 |
| Transcripts | 1,338 |
| Total transcript words | 3,659,646 |
| Transcript segments | 519,237 |
| Video descriptions w/ metadata | 1,421 |
| Video-to-keyword relationships | 7,136 |
Upload dates range from December 2020 to January 2026. Roughly 5 years of content.
716 videos appeared for only one keyword. 705 videos appeared for two or more. One video ranked for 71 out of 75 keywords.
Video categories (as tagged by creators):
| Category | Videos | % of Total |
|---|---|---|
| Education | 412 | 29% |
| People & Blogs | 373 | 26% |
| Howto & Style | 184 | 13% |
| Science & Technology | 169 | 12% |
| Uncategorized | 172 | 12% |
| Entertainment | 77 | 5% |
| Other (Film, Gaming, Comedy, Sports) | 27 | 2% |
Top channels by video count:
| Channel | Videos | % of Dataset |
|---|---|---|
| InVideo For Content Creators | 26 | 1.8% |
| vidIQ | 23 | 1.6% |
| Make Money Matt | 19 | 1.3% |
| Skills Maker TV | 18 | 1.3% |
| Money Degree | 16 | 1.1% |
| Romayroh | 13 | 0.9% |
| The Zinny Studio | 11 | 0.8% |
| Carl Faceless | 11 | 0.8% |
| Alex Christian | 11 | 0.8% |
| Steffen Miro | 10 | 0.7% |
The top 10 channels account for about 11% of all videos. The remaining 829 channels make up the rest. This is a long-tail niche with lots of small creators.
What I Stored
The database has 8 tables:
| Table | Rows | What It Stores |
|---|---|---|
| keywords | 75 | Search keyword, parent keyword, video count |
| videos | 1,421 | Title, description snippet, channel, views, duration, publish date |
| video_keywords | 7,136 | Video-to-keyword mapping with search rank position |
| video_details | 1,421 | Full description, tags, category, subs, likes, comments, Short/live flag |
| comments | 116,214 | Text, author, channel ID, likes, replies, pinned/hearted status |
| transcripts | 1,338 | Full transcript text and language |
| transcript_segments | 519,237 | Individual segments with start/end timestamps (ms) |
How I Counted
Two decisions that affected every number in this analysis:
1. I excluded viral outliers.
121 videos had over 1M views. One had 115M. These viral hits skew averages badly. A single 115M-view video would make the average look 80x higher than what a normal creator gets.
The remaining 1,300 videos show what normal creators actually experience.
2. I counted unique videos, not mentions.
If ChatGPT appears 50 times in one transcript, that's 1 video mentioning ChatGPT, not 50 mentions. Without this, a single repetitive video could make a tool look 50x more popular than it actually is.
Same for income claims. If "$10K/month" appears 10 times in one video, that's 1 video making that promise.
Finding 1: The Tool Gap
I used regex to search for tool names across three data sources: video descriptions, transcripts, and comments.
| Tool | In Descriptions | In Transcripts | Gap |
|---|---|---|---|
| ChatGPT | 163 | 284 | Genuine use + affiliate |
| ElevenLabs | 150 | 2 | Almost pure affiliate |
| Canva | 63 | 157 | Genuine use, weak affiliate |
| VidIQ | 105 | 18 | Mostly affiliate |
| CapCut | 92 | 43 | Mixed |
| N8N | 81 | 30 | Mixed |
| Leonardo AI | 12 | 57 | Genuine use |
| Sora | 8 | 52 | Genuine use |
| Claude | 5 | 50 | Genuine use |
The gap between descriptions and transcripts tells you everything.
ElevenLabs: 150 descriptions, 2 transcripts. Creators link it in every description but almost never talk about it in the actual video. This is pure affiliate play. They get paid per signup, so they drop the link, but the tool isn't central to their workflow.
Canva: 63 descriptions, 157 transcripts. The opposite. Creators talk about Canva constantly in their videos but don't link it as much. Why? Canva's affiliate program is less lucrative. It's what they actually use, not what they get paid to promote.
ChatGPT is the only tool that's high in both descriptions (163) and transcripts (284). It's genuinely central to the workflow AND has an affiliate incentive.
The takeaway: if you see a tool linked in every description but never mentioned in the video, the creator is being paid to promote it. Follow the transcripts for what people actually use.
Finding 2: How These Channels Make Money
I scanned ~1,100 video descriptions for links to affiliate programs, course platforms, and sponsorship indicators.
Affiliate programs (by unique videos linking them):
| Program | Videos |
|---|---|
| VidIQ | 92 |
| ElevenLabs | 88 |
| Amazon | 60 |
| N8N | 31 |
| InVideo | 22 |
Course platforms:
| Platform | Videos |
|---|---|
| Skool | 109 |
| Whop | 28 |
| Gumroad | 6 |
Skool is the dominant course platform in this niche. Nearly 4x more videos link to Skool than Whop. If you're selling a course about how to start a faceless YouTube channel, Skool is where the market has settled.
Sponsored vs. Affiliate performance:
| Type | Videos | Avg Views |
|---|---|---|
| Sponsored | 70 | 131,682 |
| Affiliate | 208 | 98,647 |
Sponsored videos get 33% more views. This makes sense. Sponsors vet creators before paying them, so sponsored content tends to be higher quality. Anyone can paste affiliate links.
Finding 3: Video Length
I parsed video durations and matched them with transcript word counts.
| Duration | Videos | Avg Views | Avg Words |
|---|---|---|---|
| 0-60s (Shorts) | 370 | 172,250 | 93 |
| 1-3 min | 53 | 57,902 | 291 |
| 3-5 min | 49 | 23,937 | 644 |
| 5-10 min | 208 | 81,173 | 1,464 |
| 10-20 min | 352 | 101,415 | 2,692 |
| 20-60 min | 163 | 113,968 | 5,984 |
Three formats work:
- Shorts (under 60s): 172K avg views
- Long-form (20-60 min): 114K avg views
- Comprehensive (60+ min): 123K avg views
The 3-5 minute range is death. 24K average views. Not short enough to be snackable, not long enough to be a proper tutorial.
The common advice of "make 8-12 minute videos" doesn't hold up here. The data says go extreme. Either ultra-short or ultra-long.
Only about 13% of videos in the dataset are 20+ minutes. But they average 114K views. This is underserved.
Finding 4: Income and Timeline Promises
I used regex to find specific dollar amounts and timeframes mentioned in transcripts.
Income claims (unique videos):
| Amount | Videos |
|---|---|
| $10,000/month | 61 |
| $5,000/month | 33 |
| $20,000/month | 31 |
| $3,000/month | 24 |
| $100,000/month | 21 |
$10K/month appears in 61 videos. Nearly double any other figure. It's the "quit your job" number. High enough to be meaningful, low enough to seem possible.
Timeline promises (unique videos):
| Timeframe | Videos |
|---|---|
| 30 days | 194 |
| 90 days | 153 |
| 60 days | 126 |
| 14 days | 94 |
| 1 year | 81 |
194 videos promise results in 30 days. Only 81 mention 1 year, which is far more realistic.
The winning formula in this niche is clear: promise $10K/month in 30-90 days. That combination outperforms everything else.
Finding 5: Buyer Intent in Comments
This is the finding I'm most interested in, since we're building tools around YouTube comments at TubeAlfred.
I scanned all 116,214 comments using regex patterns for buyer intent signals: price questions, budget concerns, tool comparisons, "where can I buy" type language, and "I'm ready to start" signals.
Results:
| Signal Type | Videos | % of 1,300 |
|---|---|---|
| Any buyer intent | 540 | 41% |
| Price/budget questions | 366 | 28% |
| Alternative seeking | 315 | 24% |
| Ready to start | 155 | 12% |
| Tool recommendations | 109 | 8% |
540 out of 1,300 videos (41%) have at least one comment showing buyer intent.
Real examples pulled from the database:
[80 likes] "as a beginner, what is the initial cost for all tools on monthly basis?" on "I Built An AI Influencer Automation in N8n"
[48 likes] "And the big question is. How much money does it cost to run all those API's?" on "Automating Faceless Shorts with AI"
[29 likes] "Great tips and instructions, then I got priced out of the idea. Not everyone has $100 to throw at a service every month starting out. Can you do another video with cheaper alternatives?" on "I Copied a YouTube Channel That Makes $55K/Month"
[434 likes] "guys those alternatives are not free yes they do a free trail, but it will not give you even to do 10 minutes video" on "3 FREE Elevenlabs AI Alternatives"
[65 likes] "The website is very expensive for video creation and honestly disappointing. I subscribed 120$ to make a video, and in the end, I was shocked that I couldn't create more than 5 minutes!" on "How To Create Kids Songs For YouTube with AI"
These are real people asking for help, comparing alternatives, and revealing their budgets. The consistent theme: cost is the #1 objection. 366 videos have comments specifically about pricing concerns.
Nobody is systematically responding to these people. That's the gap we're building for with TubeHarvest.
Finding 6: Top vs. Bottom Performers
I compared word frequency in titles of the top 100 videos (by views) against the bottom 100.
| Word | Top 100 | Bottom 100 | Difference |
|---|---|---|---|
| "secret" | 11 | 7 | +57% |
| "easy" | 23 | 17 | +35% |
| "money" | 32 | 26 | +23% |
| "automation" | 20 | 26 | -23% |
| "chatgpt" | 2 | 3 | -33% |
| "ai" | 32 | 50 | -36% |
The word "AI" appears 36% less in top-performing titles. It's become associated with low-effort content. Top performers talk about specific outcomes ("easy money," "secret method") rather than generic technology labels.
What I Built This With
The scraping tool is something we built internally at TubeAlfred. It pulls video metadata, descriptions, comments, and transcripts from YouTube via their API and stores everything in a SQLite database.
The analysis was done with Python, mostly regex pattern matching and SQL queries. Nothing fancy. The value isn't in the technology. It's in asking the right questions and actually looking at the data.
The database schema is straightforward: keywords, videos, video_keywords (many-to-many), video_details, comments, transcripts, and transcript_segments. About 850K total rows across all tables.
What I'd Do Differently
A few things I'd improve if I ran this again:
More keywords. 75 keywords is decent but I'd expand to related niches like "youtube automation," "ai video generator," "make money youtube" to see how this niche compares.
Track over time. This is a snapshot. I'd want to run the same scrape monthly to see trends. Are Shorts growing? Is the "AI" fatigue getting worse? Are new tools replacing old ones?
Sentiment analysis on comments. The regex buyer intent detection works, but proper sentiment analysis would catch more nuanced signals.
Channel-level analysis. I grouped by video, but grouping by channel would reveal which creators consistently perform vs. which got lucky with one viral hit.
The Takeaways
After looking at 1,421 videos, 116K comments, and 3.6 million words of transcripts:
On content:
- Go extreme on video length. Shorts or 20+ minute deep dives. The middle range underperforms badly.
- Stop overusing "AI" in titles. It correlates with 36% fewer views in top performers.
- Promise 30-day transformations. That's the timeframe people respond to.
On monetization:
- Skool dominates course platforms (109 videos, 4x more than Whop).
- Sponsored videos outperform affiliate videos by 33%.
- The description-transcript gap reveals which tool recommendations are genuine vs. paid.
On opportunity:
- 41% of videos have buyer intent in comments. Nobody's responding systematically.
- Cost is the #1 concern. 366 videos have pricing questions. There's massive demand for affordable alternatives.
- Long-form is underserved. Only 13% of videos are 20+ minutes, but they average 114K views.
Tool Recommendation (Affiliate Link)
I've been experimenting with AITuber by @imdhiva and planning to run a full 3-month faceless YouTube experiment with it. Will share the raw numbers once I have enough data.
It handles the full pipeline: script, voice, visuals, captions, auto-publish. One tool instead of stitching together five subscriptions (which is exactly what the comments in this dataset keep complaining about).
aituber.app (affiliate link)
What's Next
I'm planning to run similar analyses on other niches. If there's a keyword or topic you want me to scrape and break down, let me know.
The tools I used for this are part of what we're building at TubeAlfred. ChapterFast generates SEO-optimized chapters for your videos. TubeHarvest scrapes and categorizes comments to find content gaps and buyer intent. Both free.
All stats from real data. 1,300 videos analyzed (121 viral outliers excluded). Unique video counting throughout.