What 18 million subscribers taught me, and where the creator economy is going, told from inside a working operation, with a new channel built live as the proof.
Packaging and creator-journey videos are the only lanes that clearly stay above average for established creators without MrBeast videos or YouTube-official uploads; creator-economy essays are above average but not clearly; strategy breakdowns and behind the scenes are about average on their own. Your 18M operation, with its numbers, is the proof inside the lessons and essays; breakdowns of the biggest creators are the bigger swing; the GTA 6 build is the live case. AI is the method on screen, not the headline.
What it is not: beginner tips, faceless or money-online, AI tool tutorials, or a one-person-business guru channel. Comps below are best cases for their channels, not typical uploads.
Your own lessons, mistakes and packaging teardowns from HafuGo (thumbnails that flopped, the edit that fixed retention).
Creator journey & lessons: typical upload 0.98x its channel's usual (likely 0.81-1.06; niche 1.03x); 6.4% of commenters run 10k+ channels, 3.5% run 100k+ (niche 5.3% and 2.9%). Packaging draws the most established creators of any lane (13.3%) but a typical packaging upload sits clearly below the niche's usual (0.70x): teach packaging inside a story, not as a tutorial. An established creator's own story on a new channel can break out fast (jakkuh, ex-LTT).
An essay on a shift in the platform or the business (AI, Shorts, private equity, the Feb 2027 Partner Program change), with your inside read.
Creator-economy & platform news: typical upload 1.08x its channel's usual (likely 0.99-1.16; niche 1.03x); 8.0% of commenters run 10k+ channels, 3.1% run 100k+ (niche 5.3% and 2.9%). 8 of its 13 sampled videos are YouTube's own uploads (mostly Creator Roundups); the independent essays alone draw 6.6% (average without MrBeast and YouTube's own videos 4.4%): above average, not clearly. This is also where the 'AI slop' anger lives: take the human side of the AI story here.
Behind the scenes of a working 18M-subscriber media company: the team, the costs, the AI pipeline, the real numbers on screen.
Your theme. Build in public & BTS: typical upload 1.01x its channel's usual (likely 0.95-1.19; niche 1.03x); 5.8% of commenters run 10k+ channels, 4.0% run 100k+ (niche 5.3% and 2.9%). About average: its big-creator commenters came mostly for MrBeast behind-the-scenes videos (without them 3.9% vs an average of 4.4% without MrBeast and YouTube's own videos). Viewers reward inside access with real numbers shown up front (comment mining). Inference: an 18M operation is closer to that than a small creator's vlog, so use it as proof inside the other formats.
Why a named creator, video or format works, told with your own data and production experience as the proof.
Strategy breakdowns: typical upload 0.98x its channel's usual (likely 0.85-1.16; niche 1.03x); 8.8% of commenters run 10k+ channels, 5.1% run 100k+ (niche 5.3% and 2.9%). 10 of its 14 sampled videos are about MrBeast; without them it reads 7.1% (average 4.4% without MrBeast and YouTube's own videos), but one video carries most of that (Paddy Galloway); without it 3.4%. So the big creators come for the biggest subjects. Two of the best-known strategist channels have stopped posting long-form, and a 33.9k-subscriber channel's Veritasium breakdown has 2.7M views. Use it as the occasional big swing, not the base.
The GTA 6 channel as a running experiment: what the system does, what you still do, the numbers each week. Stakes and a result in every title.
A new-channel premise on its own has too few clean examples to call (28 uploads), and AI in a title runs slightly below the niche's usual (0.96x vs 1.03x). The comps that work lead with a result and a dare ('to prove it is not luck'). AI in a title is a wash across the niche; Jack Craig's AI version of the same experiment did better than his non-AI one (both shown). Judgment: channel-experiment creators draw many aspiring creators; your 18M credential is what changes who shows up.
| When | Working title | Format | Note | Closest comp |
|---|---|---|---|---|
| Film Nov 6 (2-hour block), out ~Nov 8 | I'm Starting a New Channel From Zero. Here's the Plan. | The live case study: building a new channel | The GTA 6 channel, a week after its first videos. Timely only before the launch. | I made a new YouTube Channel to prove Ludwig wrong, only to go viral. (JHXC64, 73k subs, 3.0M views, 2023-07) |
| Film Nov 6, out ~Nov 13 (scheduled, you are traveling) | What 18 Million Subscribers Taught Me That Nobody Says Out Loud | What 18 million subscribers taught me | Your own lessons; the robust lane for established creators. | Why I left Linus Tech Tips (jakkuh, 795k subs, 4.8M views, 2026-01, 5.8x its channel's usual) |
| Film after you are home, out ~Nov 24 | I Launched a Gaming Channel the Night GTA 6 Came Out. Here's What Happened. | The live case study: building a new channel | Launch night (Nov 18, LA) plus the first days of numbers. | I BLEW UP a YouTube Channel in 7 Days to Prove it's NOT Luck (Jack Craig, 260k subs, 2.0M views, 2024-12) |
| Out ~Nov 27 | YouTube Is Raising the Bar for New Channels on Feb 1. What Changes. | Where the creator economy is going | Date-driven essay on the Partner Program change. | The Relatable Influencer Is Dying… and Sophie Silva Proves It ( Vera, 20k subs, 1.1M views, 2026-08, 62.2x its channel's usual) |
| Out ~Nov 30 | What It Actually Costs to Run an 18M-Subscriber Channel | Inside the operation | Number-led; the inside access has to carry it. | I Documented My Life Until I Made $1M (Dean Johnson, 28k subs, 893k views, 2025-05, 9.4x its channel's usual) |
| Out ~Dec 2 | My 10 Worst Thumbnails at 18M Subscribers (and What Fixed Them) | What 18 million subscribers taught me | Packaging inside a story. | How I Actually Make Viral Thumbnails (Isaac, 230k subs, 1.1M views, 2024-01) |
| Pre-film before China, out Dec | Why Veritasium Still Gets Views (From Someone Running an 18M Channel) | Strategy breakdown, from someone running one | The big swing: a named giant; pick the creator in the ideation run. | Veritasium - why he still gets views (The Internet Stamp, 34k subs, 2.7M views, 2026-07, 134.2x its channel's usual) |
| Pre-film before China, out Dec | 30 Days of Building a New Channel: The Honest Numbers | The live case study: building a new channel | The result video; publish it even if the result is bad. | I Got 100,000 Subscribers In 4 Days To Prove It’s Not Luck (MoMike Chamberlin, 97k subs, 814k views, 2024-10) |
| Pre-film before China, out Dec | The Edit That Saved My Channel's Retention | What 18 million subscribers taught me | Your own data as the proof. | How I Grew from 0 to 100K Subscribers in 5 Months (Nobody Teaches This) (Kallaway, 479k subs, 757k views, 2025-03) |
| Pre-film before China, out Dec | AI Didn't Replace My Team. It Replaced the Busywork. | Where the creator economy is going | The AI take, human-first, only after trust is built. | The Real Reason Veritasium Sold His Channel (Jon Youshaei, 818k subs, 1.8M views, 2026-08, 3.1x its channel's usual) |
From the GTA 6 launch outline and your Nov 30 outcomes; travel is tentative there too.
| When | What is happening (GTA 6 outline, Nov 30 outcomes) | Creator channel |
|---|---|---|
| Oct 26 - Nov 1 | Locked month at home. Gaming channel set up by Oct 31; gaming videos 1-2 out Oct 30 and Nov 1; Life Hacks sponsor shoot Oct 29-30 (an open collision in the GTA outline) | No talking-head sittings. Film the gaming channel build as it happens (GTA outline item 12): phone and screen captures that videos 1, 3 and 8 need. |
| Nov 2 - 8 | Still at home and full (GTA outline): GTA videos 3-4 filmed Nov 2, 5-6 filmed Nov 4; streams Nov 1, 3 and 5 (Nov 6-7 if hours are short); Partner Program application and rehearsal Nov 7; launch-ready check Nov 8 | One 2-hour block on Nov 6, the lightest day, for videos 1-2. If extra streams take Nov 6, start Nov 21 instead. |
| Nov 9 - 18 | Travel (tentative in the GTA outline): Miami ~Nov 9-11, LA from Nov 11/12, 3D printing shoot Nov 14, Clio Awards Nov 18 | Video 2 goes out on schedule. Phone behind-the-scenes only. |
| Nov 18, 9 PM PT | GTA 6 unlocks for US/CA (PS Store listing) while you are in LA; where you play launch night is still open in the GTA outline | Capture the night for video 3. |
| Nov 19 - 20 | Fly home (outline says Nov 19; Nov 20 was proposed, not decided) | Nothing. |
| Nov 21 - Dec 2 | Rest of the locked month | Four sittings of about 1 hour (two a week): videos 3-6. |
| Dec | Toronto lease Dec 1, China early December (working plan) | Two 2-hour blocks before China for videos 7-10; the AI pipeline edits them. |
| Lane | Why | Number |
|---|---|---|
| Beginner growth tips | The audience you ruled out; the biggest lane by uploads, about half of typical uploads get under 10k views. | lane table |
| Faceless, automation, make-money-online | The audience you ruled out; 2.1% of commenters run 10k+ channels, the fewest of any lane. | lane table |
| AI debate as a format | Few established creators (2.5%) and typical uploads run 0.86x their channel's usual; the comments are mostly about AI slop. | lane table |
| AI tool tutorials | Few established creators (2.4%), typical uploads 0.93x; a crowded tutorial lane where tools change monthly. | lane table |
| One-person business as the label | Established creators about average or below (3.5%, likely 1.7-5.6%), and the phrase belongs to AI side-hustle titles. Use the idea, not the words. | lane table |
| Creator drama | Big views, but its creator-commenter share is carried by one parody video, and the drama register works against an authority brand. | lane table |
| Content marketing for businesses | Few established creators (2.5%); business owners do show up here, so reach them through the 'inside the operation' episodes instead. | lane table |
Your Brand System has creator education 'on hold until 100 posts' and your Oct 17 plan makes it the first cut if you fall behind; today you said you want to document this month.
Suggested: one 2-hour block on Nov 6 (the lightest day in the GTA outline) for videos 1-2, only if the gaming channel's first videos are out and no extra stream needs that day; otherwise start Nov 21 with the launch-night video.
The share of established creators clearly stays above average on packaging and creator-journey videos; strategy, creator-story and creator-economy videos are not clearly above average once MrBeast videos and YouTube's own uploads are removed. Business owners write about their own business most on content marketing for businesses (1.9%, 17 comments), creator business & media company (0.7%, 16 comments), one-person business & personal brand (0.6%, 11 comments), interviews & podcasts (0.5%, 12 comments) videos (2-17 such comments per lane, so treat it as a hint).
Suggested: creators first (your credential's natural audience); reach operators with 'the business of an 18M channel' episodes, and keep one-person-business words out of titles.
Working creators in the comments treat human-made as a badge and aim their anger at AI slop; AI in a title runs slightly below the niche's usual.
Suggested: show the AI doing the labor and you doing the taste; save the 'I love AI' argument for video 10.
It gives a live result every week; the new-channel comps that work lead with a result and a person.
Suggested: yes, as one of two recurring formats, not every video.
Course businesses in this niche have documented drops (Ali Abdaal 2022, Creator Science 2025); the best documented paid community is The Futur's (about 2023).
Suggested: a free newsletter now; decide community vs AI product at 10,000 subscribers.
You picked a new channel under your name.
Suggested: your name, with 'HafuGo' in the about text for the credential.
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.