How the short-form algorithm works.

The word algorithm makes it sound like a locked box with a secret inside. It is not. Instagram, TikTok, and YouTube Shorts are far more alike than the myths suggest, and once you see the shape they share, the advice stops being folklore and starts being obvious.

Here is the whole thing in one sentence: when you post, the platform shows your video to a small test audience, watches whether people stop for it, stay with it, and pass it on, and then widens the audience only if those signals are strong. You are not gaming a machine. You are passing a short attention test with real people, and the machine is just counting.

Test small, then expand.

All three platforms describe a version of the same loop. A new post goes to a small seed audience first, not straight to everyone. The system reads how that group responds within the first hours, and if the response is strong, it pushes the video to a larger group, then a larger one again. YouTube leadership has called this an explore then exploit model for Shorts, and it works the same way on Reels and the For You feed.

Two things follow from this. First, your follower count is not the gate: each video is judged on its own signals, which is exactly why a creator with a small following can still reach millions. Second, the first few hours matter enormously, because a weak early test is where most videos quietly stop spreading. The rest of this is just the four signals that test measures.

Do they stop?

The first thing the test measures is whether people stop scrolling at all. YouTube even has a name for it in Studio, viewed versus swiped away, and it is the clearest read on whether your opening earns attention. If most people flick past in the first second or two, the video fails the test before the content ever gets a vote.

This is the hook, and it is the highest-leverage thing you control. Open on the payoff, not the wind-up. Make the first frame look different from the feed. Let the first caption carry the idea on mute. In our own work, opening on the payoff instead of the setup roughly doubled how many viewers stayed past the first three seconds.

Score your opening with the free Hook Checklist, pull a stronger one from the Hook Vault, and see the numbers in what 1,000+ videos taught us about the first 3 seconds.

Do they stay?

Once someone stops, the test measures how long they stay. This is watch time, and it is the signal every platform centers on. Instagram has said watch time is the most important factor for Reels. TikTok and YouTube both weigh how much of a video people finish. Replays count too, which is why a tight, loopable video can outperform a longer one.

One useful nuance Instagram has confirmed: both the percentage watched and the total seconds watched matter. A longer video that holds people can beat a very short one that people only half-finish. So the goal is not shorter for its own sake, it is no wasted seconds. Every line should earn the next one, and the parts you love but the viewer does not need are exactly the parts to cut.

More on building a video with a shape that holds in retention is a shape.

Do they send it?

The signal that unlocks new audiences is not the like. It is the share. Instagram has been unusually direct about this: sending a Reel to someone in a DM is the strongest signal for reaching people who do not already follow you, because a share is an active recommendation, not a passive tap. TikTok and YouTube treat shares and saves as stronger signals than likes for the same reason.

This changes what you make. A like asks nothing of the viewer. A share means the video was useful enough to send to a specific person, or funny enough, or surprising enough, that passing it on felt worth it. So the quiet question behind a video that spreads is: who would send this to whom, and why? Build for that, and reach follows.

Do you give it enough tries?

Because every post gets its own fresh test, posting is how you buy more chances to pass it. Post rarely and you give the system too little to learn from and give yourself too few reps to sharpen the first three signals. Post consistently and both improve. This is why volume works, not because more is magic, but because more is more at-bats.

The trap is promising yourself a daily cadence and quitting in two weeks. Pick a number you can hold for a year and protect it. The way to hold it is not more filming, it is repurposing: one recording becomes a month of posts.

Find a cadence you can keep with the free Content Cadence Calculator, then make one shoot carry it with the Repurposing Planner.

What not to chase.

A lot of algorithm advice is noise. The exact completion percentages you see quoted are third-party estimates, not numbers the platforms publish, so do not treat them as a target to hit. Follower count does not gate a single video. Posting at a secret best time matters far less than posting something people stop for. And chasing every trend rarely beats making one genuinely good video on a topic you actually know.

One thing that is real and easy to miss: platforms generally will not recommend videos carrying another app's watermark to new audiences, so export clean from the platform you are posting to. Beyond that, the fundamentals do not change even as the details do, because they are really just measures of whether a human found your video worth their time.

Find your weakest signal first.

You do not fix all four signals at once. Read your own analytics: a high swipe-away or a cliff in the first seconds is a hook problem, a slow bleed through the middle is a retention problem, strong watch time but few shares is a value problem, and good videos that never compound is usually a volume problem. Fix the weakest one first.

If you would rather not guess, the free Content MRI points you at the single bottleneck holding your channel back, and why your videos aren't getting views walks through the fixes in order.