I'm 38. Project manager during the day. Last March I started a YouTube channel I never appear on. Eleven months in, it makes about three grand a month, mostly from ads, some from sponsors.
I wasn't planning a side hustle that worked. I had three dead ones behind me. This one stuck and I don't fully understand why.
How I picked the niche
I wrote 10 ideas down one Sunday. Threw out the ones I knew I'd get bored of by video 15. Threw out the ones where the existing channels were already great. What was left was forgotten history. Failed inventions, vanished towns, almost-businesses. I'd been reading about that stuff since I was a teenager. So if nothing else I'd entertain myself.
Ones I almost picked and didn't:
- AI tool reviews. Already cooked by 2025.
- Reddit narration. Saturated and a bit copyright-shaped.
- Productivity. Everyone face-shown does it better.
- Crypto. I just don't believe in it.
The tools
- ChatGPT and Claude for scripts
- ElevenLabs cloning my own voice
- Midjourney for stills, Pexels for stock
- Runway, Pika or Veo when I need clips that don't exist
- CapCut for the edit
- Ideogram for thumbnail text
- Suno for music
About $100 a month. The first month I tried to do it entirely on free tiers. The output was bad enough that I gave up after three videos.
My Sunday
- 20 min picking a topic
- 30 min in Claude on the script
- 10 min through ElevenLabs
- 30 min stitching the visuals together
- 25 min in CapCut
- 5 min uploading
Some Sundays it bleeds to three hours. Some Sundays I cheat and do half on Saturday morning when my partner is at the gym.
What happened, in order
Month one. Four videos. The third one pulled 47 views over five days. I almost quit at week three.
Month four. A video about a 1962 small-town transit experiment caught fire. 12k views in five days. Crossed monetisation.
Month nine. First sponsor. I quoted them way too high. They paid it anyway, which still confuses me.
Month eleven. Steady around three grand.
I keep waiting for it to stop working.
The all-in-one tools
Around month six I tried one of those "one prompt, full video" platforms. Won't name it. The thing promised to write the script, generate the voice, pick the b-roll, and stitch the whole video together from a single input box.
The output was technically watchable. It was also lifeless in a way I couldn't quite put my finger on. The pacing was off. The voice picked the wrong word to emphasise. The b-roll cut at the wrong moment. My retention on those uploads dropped almost a third compared to the ones I'd stitched manually.
I went back to the multi-tool workflow after three videos. The tools will get there. They're not there yet. Picking your own voice, your own b-roll, your own cuts still matters more than the speed.
Things I'd skip if I started over
The default ElevenLabs voices. Other YouTubers can hear them a mile away. Clone your own.
The all-in-one auto-pipeline tools, as above. Speed at the cost of retention is a bad trade.
The easy-looking niche. It looks easy because everyone else already picked it.
Treating thumbnails like the last item on the list. The thumbnail is the click.
Quitting at month three. I almost did. The algorithm hadn't figured out my channel yet.
What I'd skip
Six weeks of bouncing between YouTube tutorials trying to figure out which AI tools actually mattered.
I ended up doing a course on Mindwand once I'd burned through enough of those tutorials. Could have done a Skool community or a Coursera thing instead, plenty of those around. The Mindwand one was short and cheap and I finished it, which is more than I can say for the two Coursera courses still sitting unfinished in my account.
That's the whole story. I'm still uploading. Maybe by Christmas this falls apart. Probably not. The thing I share with every other survivor I know is that we didn't skip weeks.