Creator education: what the data says
The short answer
- Established creators comment most on packaging, strategy breakdowns, creator stories and creator-economy videos. Share of commenters running a 10k+ channel: packaging 13.3%, strategy 8.8%, creator stories 8.6%, creator-economy 8.0% (niche 5.3%). Fewest, outside the lanes you ruled out: AI tools for creators 2.4%, content marketing for businesses 2.5%, AI debate & experiments 2.5%, storytelling & scripting 3.0%.
- A handful of MrBeast videos pull in big creators. 32 of 269 sampled videos are about him and hold 28% of the commenters running 100k+ channels, most of them on 7 videos. Without MrBeast videos and YouTube's own uploads, only packaging & ideas, creator journey & lessons clearly stay above average; independent creator-economy essays are above average but not clearly, and strategy breakdowns and creator stories become about average.
- Your behind-the-scenes theme is about average, not a magnet. A typical build-in-public upload gets 1.01x its channel's usual (niche 1.03x); its big-creator commenters come mostly from MrBeast behind-the-scenes videos. The business topic shows no lift either (0.91x).
- Packaging is the paradox: it draws the most established creators, but a typical packaging upload sits clearly below the niche's usual (0.70x). Teach it inside a story, not as a tutorial.
- The AI anger is at AI slop, and that is your risk. 120 of 140 negative-AI phrases in the comments say "AI slop" (mostly on 3 videos). A 100% AI-made gaming channel can get lumped in.
- No title trick clearly helps. AI in the title: 0.96x the channel's usual vs a niche usual of 1.03x, and inside the same channel between 0.88x and 1.04x depending on the filter: no gain, possibly a small loss. Naming MrBeast, a new-channel premise or behind-the-scenes words show no clear gain. The story and the person carry it.
- Newcomers get through. 21 channels based in the US, UK, Canada, Australia, Ireland or New Zealand and created since 2024 have a 100k+ long-form in strategy, creator-economy, journey or behind-the-scenes; Film Booth (last long-form 2024-04) and Paddy Galloway (last long-form 2022-08) went quiet.
- Of the channels you named, the share of commenters running 100k+ channels: Colin and Samir 4.4% (13); Jon Youshaei 3.7% (11); The Futur (Chris Do) 1.4% (4); Dan Koe 0.3% (1). Differences of a few people are noise.
- The plan in one line: what 18 million subscribers taught you (packaging inside the stories), with creator-economy essays and big-creator breakdowns as the bigger swings, your operation as the proof and the GTA 6 build as the live case; AI is the method, not the headline.
Based on 17k creator-niche videos from 10k channels, the latest uploads of 6.2k channels, 50k comments and the YouTube channels of 48k commenters, plus 30 reference channels. Pulled 2026-10-27. Full numbers · The plan
1. Where do the creators you want actually comment?
For the first 100 comments on each sampled video (YouTube's relevance order, the same depth everywhere), every commenter's own channel was looked up. The bar is the share running a channel with 10,000+ subscribers; the dashed line is the niche average (5.3%). It measures commenters, not viewers, and 5-14 videos per lane, so compare bars, do not read one as "x% of the audience". Business owners without a channel cannot show up here: they write about their own business most on Content marketing for businesses (17 comments), Creator business & media company (16 comments), One-person business & personal brand (11 comments) videos (small counts). MrBeast videos are 32 of 269 sampled videos but hold 28% of the 100k+ commenters; packaging has none and is the most robust; creator-economy's bar leans on YouTube's own uploads.
Reference channels
Share of commenters running a 100k+ channel, first 100 commenters on 3 recent videos each (* = you named it; The Futur stands in for Chris Do). Left out for smaller samples: Creator Science (Jay Clouse) 7.1% (11 of 156); Sahil Bloom 0.4% (1 of 247).
2. Which kinds of video hold up?
Niche long-forms from 1.7k creator channels, 2 weeks to about 11 months old, flops included, each against its own channel's usual views, counted per channel. The dashed line is the niche's usual (1.03x). Ranges overlap for most lanes. Clearly below: Packaging & ideas (0.70x). Clearly above: none.
Most lane labels on these uploads come from a model trained on the agent labels (cross-checked accuracy 0.74). Ages are approximate (YouTube shows "3 months ago").
3. How did the niche get here?
Beginner "how to grow" tips are the base layer: one of the two biggest lanes in 7 of 13 years. What sits on top shifts, and single giant videos often carry a year: creator stories in the MrBeast era (Colin and Samir's "The Full Story of MrBeast", 22.6M views, 2021); income reveals in 2023 (62% of that lane's views that year are one video); drama in 2024 (40% one parody video); AI experiments in 2025; and in 2026 faceless, automation & money-online, AI tools for creators, creator-economy & platform news, led by essays like fern's "The Death of Educational Content on YouTube" (11.2M). YouTube's own 2020 notice on made-for-kids rules alone has 130.8M views. On YouTube search today (2026-10), "faceless youtube" scores 71 against 19 for "grow on youtube" and 9 for "creator economy" (Google Trends, same 0-100 scale).
From the search corpus (every niche video with 100k+ views, by publish year) and Google Trends; the corpus is query-biased and single giant videos swing a year. Details: history.
4. Your bets, tested
Every niche long-form from creator channels in the last ~11 months, flops included, each against its own channel's usual views, counted per channel so one big channel cannot carry a result.
Documenting a new channel
0.77x the channel's usual (likely 0.68-1.10; niche 1.03x) · 28 uploads · too few uploads to tell
Too few clean examples to call. The comps that work lead with a result and a person.
AI in the title
0.96x the channel's usual (likely 0.92-1.00; niche 1.03x) · 2.5k uploads · no clear difference from the niche's usual
Inside the same channel: 0.88x (0.79-0.98) counting every channel with both kinds, 1.04x with 2+ of each. No gain, possibly a small loss; the bigger risk is the comment section.
One-person business in the title
1.13x the channel's usual (likely 0.94-1.85; niche 1.03x) · 66 uploads · no clear difference from the niche's usual
Unclear, and the words belong to AI side-hustle titles ("Laziest One-Person Business Model To Start in 2026 ($100/day+)").
Behind-the-scenes words in the title
0.77x the channel's usual (likely 0.38-1.32; niche 1.03x) · 17 uploads · too few uploads to tell
Too few clean examples; the build-in-public lane itself is about average.
Naming MrBeast
1.16x the channel's usual (likely 0.97-1.52; niche 1.03x) · 125 uploads · no clear difference from the niche's usual
No clear gain; the best ones are critical takes ("Why MrBeast Is Getting Scary").
5. How fast do new channels grow here?
21 channels based in the US, UK, Canada, Australia, Ireland or New Zealand and created since 2024 have a 100k+ long-form in strategy, creator-economy, journey or behind-the-scenes lanes. An established creator's new channel can start fast: jakkuh (ex-Linus Tech Tips) reached 795k subscribers about 14 months after creating his; MKBHD's behind-the-scenes channel passed about 175k subscribers a day after launch (2021).
6. What viewers reward
From reading the top comments of 275 videos across all lanes (quotes verbatim).
The video proves its own method
The video uses the method it teaches (meta-proof)
Real numbers, up front
Money numbers shown first, no stalling, creator volunteers the uncomfortable detail
The real operation, unpolished
Unedited, unscripted real-person footage beats the polished main channel
Human-made is the trust badge
Human-made as the trust and prestige badge
7. Who is out there
YouTube does not show other channels' viewer ages or jobs; these groups come from the comments and the commenter-channel check.
Working creators and channel operators
Views collapsing despite strong metrics, packaging that stays honest, team pressure and burnout, AI copycats.
For you: SERVE, core. Make each video its own case study: his numbers, team, costs and failed tests, packaged as legitbait. Open with the 18M credential to answer 'why aren't you bigger'. Judge success by who comments, not how many.
Business owners, agency owners and marketers
Pricing, clients, getting customers, paid versus gifted deals, team tooling.
For you: SERVE, the money layer. Frame the one-person media company in business terms: headcount, cost per video, margins, systems, what a brand gets. Never funnel-first. Measure them off YouTube (newsletter, inbound), since comments will not show them.
AI workflow builders
Setup details (CLAUDE.md files, voice tools), real cost of the tools, whether demos hold up off-camera.
For you: BORROW, carefully. Show the real stack and its monthly cost inside creator-business stories, and open-source files for loyalty, but never become a tool-tutorial channel: that pulls freelancers and side-hustlers, not 100k+ creators.
Human-craft loyalists (anti-slop)
Trust, human-made work, undisclosed AI, AI-looking thumbnails, platforms pushing slop, real work flagged as AI.
For you: EARN, do not provoke. They include his target peers. Keep face, voice and judgment his, disclose where AI sits, never pitch 'AI runs my channel', no AI-looking thumbnails. Lead with proof of quality, then show the pipeline.
Creator-culture spectators
The person, the scale of the machine, unedited moments.
For you: BORROW the format. Behind-the-scenes of his machine pulls this crowd and his 18M fan base, so use it as the reach layer, but make the payoff operator-level and skip drama and milestone content.
Beginners and small creators
First 1,000 subs, monetization thresholds, views after 48 hours, company.
For you: AVOID. No 'start a channel', first-1,000-subs, settings or monetization titles; they flood comments, teach the algorithm the wrong viewer, and cannot become the 100k+ audience.
Money seekers (MMO, faceless, side hustle)
Any income fast, scams, blocked cash-outs.
For you: AVOID. No income-promise titles, no '$X a month with AI', no faceless recipes. Borrow only the honesty format: real P&L, failures shown, nothing to sell.
8. What is winning right now
English long-forms from the last 6 months in the lanes you would enter (skip-list lanes left out).
9. Money and rules
- Sponsors pay, ads do not. Colin and Samir: about 95% of show revenue from sponsors, about $1 per 1,000 views from ads (Feb 2022).
- Products: the best documented paid community is The Futur's (900+ members at $449 a quarter, reported $1M+ a year, about 2023). Course businesses have documented drops: Ali Abdaal (2022), Creator Science digital products -74% (2025).
- AI rules: AI help with scripts, thumbnails, titles, captions or ideas, and your own cloned voice for voice-overs or dubs, need no disclosure; realistic AI scenes do. Mass-produced template AI content cannot earn.
- The bar rises Feb 1 2027: channels applying from then need 8,000 watch hours (not 4,000); channels already in the Partner Program are not affected.
What this means for your channel
- Lead with your own lessons (journey), with packaging taught inside the stories: the two lanes where established creators comment even without MrBeast videos or YouTube's own uploads. Creator-economy essays and breakdowns of the biggest creators are the bigger swings.
- Make your operation the proof: real numbers, the team, the AI pipeline on screen. Nobody in the strategist lanes runs an 18M-subscriber channel.
- Keep AI as the method you show, not the claim you argue; show your face and decisions in the GTA channel so it never reads as slop.
- Package with people and results, not with "one-person business" or "AI agents": those words pull the side-hustle crowd.
- Skip beginner tips, faceless and money content, AI tool tutorials and AI debate as formats.
Data from YouTube Data API, yt-dlp search, TubeLab, Google Trends, pulled 2026-10-27. Data figures are computed from the project's data files by the page builder; web facts are quoted from linked sources; planning estimates are labeled as assumptions.