- You don't need a viral hit. Channels doing 2,000–10,000 views per video can build real communities and profitable businesses behind the content.
- Packaging comes first: title, thumbnail, and opening seconds should create a curiosity gap... a question the viewer needs answered.
- Shorts and long form play different games: shorts get algorithm pushes in the first hours; long form compounds like SEO over months and years.
- The metrics that matter: ~5% click-through rate (after enough impressions), ~33% retention at the end of long videos, and 70%+ stayed vs. swiped away on shorts if you want the next push.
- Authentic beats polished: raw edits, real voice, and non-AI thumbnails often outperform hype... use AI to brainstorm and edit, not to replace your point of view.
Jackson and I started publishing on YouTube for different reasons. He breaks down public companies on Cashew Sono. I'm running shorts on a dog channel called Nancy Decides and long-form career content on Developer Dylan. We also have access to a 14,000-subscriber family channel with years of analytics... which turned this from "post and hope" into a real experiment.
This episode is us documenting what we've learned so far. If you're a small business owner trying to figure out how attention works without burning out on every platform at once, these are the patterns worth stealing.
In case you're interested, here are the channels we're experimenting with:
- Developer Dylan: IT career related content
- Nancy Decides: Dylan's dog answers commenter questions
- Casysono: Publicy traded business breakdowns
Why we're doing this (and why one platform at a time)
The motivation wasn't "get famous on YouTube." It was closer to research.
I'm trying to understand how YouTube works the way I already understand SEO: publish, measure, iterate, let compounding do its thing. Jackson wants a long archive of business breakdowns for people who own a stock and can't explain what the company actually does. Madeline's channel started as day-in-the-life QA content and evolved with her life... and still pulls consistent views years later.
We almost spun up Twitter threads and blog posts at the same time. We didn't. We zeroed in on one thing, get good at it, then spread wings later. That discipline matters if you're already running a business and can't afford five half-built content habits.
Small audiences are a feature, not a bug
A lot of channels doing 2,000 to 10,000 views per video aren't "failing." They're building a sliver of people who identify with the creator... and that's enough to support a business, a community, or a years-long archive.
Jackson's goal on Cashew Sono isn't virality. It's 10K–99K views on company breakdowns that people revisit when they're nervous about a stock they hold. That's a ten-year game, not a weekend spike.
Treat the audience well, don't quick-buck them, and they'll stick around long enough to buy valuable stuff they actually want. The internet has infinite options now. People don't need one mega-channel. They need one person who walks them through the thing they care about.
Shorts vs. long form: two different clocks
Shorts get thrown into the feed, tested fast, and pushed again if people stay. In the first hour or two, YouTube is basically asking: did people swipe away or watch?
A run-of-the-mill short might get ~1,000 views on the first push, 500–1,000 on the second, and occasionally a third push that 5–6x the total. One Nancy Decides video hit ~6,000 views that way, then stalled. Another bombed at ~30 views. Same format. Different outcome. That's the shorts lottery.
Long form is slower. Madeline has a video at ~300,000 views that built over a year, not a month. Most of her traffic comes from search, not the home feed... very similar to a blog post ranking on Google. Better click-through when someone searches your topic → more impressions → more clicks. Compounding.
If you're choosing where to invest limited time: shorts teach you hooks and retention in fast feedback loops. Long form builds durable discovery. We're using shorts data to inform long-form packaging, not pretending they're the same sport.
The numbers we're actually watching
Early analytics lie. With 10 impressions, a 30–50% CTR often just means friends and family clicked. Wait until you have at least ~100 impressions before treating click-through rate as signal.
What we're targeting:
- Click-through rate (long form): ~5% once impressions are meaningful. Shorts are feed-native, so CTR doesn't apply the same way.
- Retention at end (long form): ~33% minimum; flatter drop-off is better. On shorts, 100%+ is possible on 10-second loops when people rewatch.
- Swipe-away vs. stayed (shorts): Aim for 70%+ stayed; 50%+ to earn the next algorithm push.
On Madeline's top QA videos, I pulled the retention graph and marked every bump up. Those bumps are either rewinds or skips to that moment. Four videos, five repeatable topics each... that's a cheat sheet for what to hit in the next upload.
Impressions matter too. More people seeing the thumbnail → more chances for the CTR KPI to mean something. YouTube's own guidance on audience retention aligns with what we're seeing in the wild: the curve shape tells you where you lost them, not just whether they left.
The curiosity gap (packaging before you record)
Before you write, record, or design anything, ask: why would someone care to click?
Mr. Beast's line on this is useful: be clickbaity, but truthful. Bait the click, then pay it off in the first five seconds or you lose trust forever.
For my first Developer Dylan day-in-the-life, the thumbnail was me with a backpack and an arrow: "college dropout." Title: day in the life of a software engineer. The gap: how did he get here? Opening line: "This is a day in my life as a software engineer." Then deliver.
Same pattern works on LinkedIn. One post hit 56,000 views because the first line, second line, and image created a gap you had to close. Software engineer + college dropout. Where's the missing middle?
For Jackson's "strawberries in space" business video, the thumbnail already had absurdity built in. The next layer is stacking two true absurd facts with a gap between them... not AI-slop visuals people swipe past.
Jackson's experimenting with simpler, obviously non-AI thumbnails next week. Smart. Authentic packaging matches authentic footage.
Authenticity vs. strategy (and where AI fits)
Madeline never chased trends. She vlogged what she wanted to vlog. Over time that became AdSense that covered bills... not life-changing money, but real income from doing something she'd do anyway.
Jackson's best Cashew Sono videos are raw. I tried aggressive cuts on his filler phrases. The top performers kept the "ums" and transition buffers. That's what people came for.
I'm using less AI on this stuff, not more. ChatGPT captions a thumbnail when I'm lazy. Fine. But headlines, hooks, and curiosity gaps? I work those out myself because I know the viewer... I've mentored them, interviewed them, sat in meetings with them. If you're talking to a specific person every day, outside input can dilute the native voice.
Strategy still matters. E-commerce brain meets authentic creator brain is a good pairing: test packaging, read retention, tweak thumbnails the next day. The algorithm isn't out to get you. It's dumb. Psychology wins. Build something people would actually choose to watch.
Distribution first (what a dog channel taught me)
Nancy Decides is stupid simple: two sheets of paper, yes and no, treat on each, ask my dog a question. 10–12 second shorts. People comment questions. I reply as Nancy. One comment asked if God created humans because loneliness was the only thing an omnipotent being couldn't solve. Another: "Will I get a girlfriend?"
Latest video: ~1.2K views on one upload. Channel total: ~45,000 views in about a month. Average swipe-away rate is brutal... often only ~30% stay. I need 50–70% for reliable pushes. Same exact format as a 6,000-view winner sometimes dies at 30 views. Still learning.
But here's the business angle: if you sell dog treats or kibble, mid-roll shoutouts in a 20-second short are 47,000 impressions for five minutes a day. Distribution first. Build attention with something straightforward, then inject the brand once you know people stay through a gap → answer structure. Even a 10-second intro + 10-second sponsor beat + answer can hold if the middle earns it.
We're going to experiment with tracked landing pages next. Small business takeaway: you don't need a perfect channel. You need a format people watch and a place to send them.
Buddy up and play the long game
Share analytics with someone who'll text you the next day: "change the thumbnail." Jackson does that for me. I'm doing it for him. Krista's coming back into the orbit with video editing skills and a huge newsletter audience from mission work in East Africa. Four people sharing retention graphs beats one person guessing.
Jackson has ~6,000 publicly traded US stocks to cover. At a couple videos a week, that's a multi-year archive. We're committed to at least a year of learning out loud.
If your brain loves numbers like mine, pair with someone who's in it for the craft, not the dashboard. Madeline's lesson: don't let the KPI obsession override the reason you're publishing.
What to do this week
- Pick one platform. Not five. Get one rhythm working.
- Write the thumbnail and title before you record. What gap are you opening?
- Publish one short and one long piece (or pick one format) and wait for 100+ impressions before judging CTR.
- Pull the retention graph. Mark bumps. That's your outline for the next video.
- Find one accountability partner who'll roast your packaging lovingly.
If you want help turning raw stories into marketing-ready narratives, the Content Creation Hub walks you through guided prompts. For showing up consistently while you experiment, the free social media content calendar beats winging it every week.
Conclusion
YouTube isn't viral-or-bust. It's packaging, psychology, and patience... with different rules for shorts and search-driven long form.
Create a truthful curiosity gap. Pay it off immediately. Measure retention, not vanity. Document what you'd do anyway. And remember: 1,200 people watched a dog pick yes or no on a single video... while most swiped away. Both things can be true. Keep iterating.
Want prompt starters for content, hooks, and everyday business tasks? Browse the AI Prompt Library.