Why this lesson: before we tour Instagram, YouTube and LinkedIn one by one, you need the single idea that runs under all three. Learn it once and each platform stops feeling like a mystery box — you'll already know what it's trying to do.
First, recall
Without scrolling back: name the 4 building blocks of any piece of content, in order. (Answer at the bottom of Lesson 1.) Hold them in mind — we're about to connect every one to how the feed works.
The one idea: a feed is a prediction engine
You don't choose what you see on Instagram Reels, the YouTube home page, or the LinkedIn feed. An algorithm chooses for you. And it is trying to answer exactly one question about every post it could show you:
"How likely is this person to engage with this — right now?"
It's a bet. The platform predicts your reaction, shows the posts it's most confident about, then watches what you actually do to correct its next bet. That feedback loop is the whole machine. Your job as a creator is simply to give it strong, honest signals to bet on [YouTube Creators — how recommendations work].
The 3 signals every platform reads
Feeds differ in the details, but all three watch the same three things — in this order:
1Stop?
2Stay?
3React?
Will they stop? — Did the post earn the scroll-stop / the click? On video this is the click-through rate (CTR). If almost no one stops, the platform stops showing it. This is your Hook, measured.
Will they stay? — Once stopped, how long did they watch or read? On video this is average view duration (AVD) / retention. High retention is the strongest signal a feed has that content is good [MrBeast production memo]. This is your Value + Payoff, measured.
Will they react? — Did they like, comment, save, or share? Shares and saves count most — they signal "worth passing on." This is your CTA, measured.
The click that should just happen
The 4 blocks you learned aren't a writing trick — they're the levers for these three signals. Hook → Stop. Value + Payoff → Stay. CTA → React. Make good content and you're already feeding the algorithm exactly what it measures. There is no separate "beat the algorithm" skill to learn first.
Worked example · one Git Reel, three signals
Hook "You've been undoing Git the dangerous way." → viewers stop scrolling. Good CTR.
Value + Payoff 15-sec screen recording of git revert saving the day. → viewers stay to the end. High retention.
CTA "Save this for your next bad merge." → viewers save it. Strong react signal.
Same faceless Reel from Lesson 1 — now you can see why it would spread. Three green signals in a row is what "the algorithm liked it" actually means.
Quick self-check
No clues in the formatting — pick from memory, then click.
1. A feed algorithm is best described as a machine that tries to…
The feed is a prediction engine. It bets on your likely reaction, shows its best guesses, then learns from what you actually do — recency and volume are minor next to predicted engagement.
2. Your Reel gets tons of scroll-stops but almost everyone leaves in two seconds. Which signal is weak?
People stopped (good hook) but didn't stay — that's low retention / average view duration. The fix lives in your Value and Payoff, not the hook.
3. Which viewer action gives the feed the strongest "show this to more people" signal?
Saves and shares are the loudest react signals — they mean "worth keeping / passing on." That's why a good CTA asks for exactly one of them.
Primary source · read this
The clearest distillation of stop → stay → react is the annotated MrBeast production memo — obsess over click-through, average view duration, and the first 60 seconds. Read it for the mental model; skip the hustle-culture bits. More vetted sources in RESOURCES.md.
💬 Your teacher is here. Want to see how these three signals look different on LinkedIn (where "stay" is reading, not watching)? Or unsure how a platform measures a "stop" on a text post? Just ask — that's the next lesson, but I can preview it for you.