How the Shorts Feed Decides What to Push

The Shorts feed does not work like the rest of YouTube, and the biggest mistake creators make is assuming it does. There is no thumbnail to win the click, no title doing half the persuading, no suggested column on the side. There is just a vertical stack of videos that autoplay one after another, and a viewer whose thumb is already hovering over the next swipe. The algorithm behind that feed is reading a different set of signals than the one behind your long-form videos, and once you see which ones, the strategy stops being guesswork.

This post is about what the Shorts feed actually measures and how it decides whether to keep showing your video to more people. We will stick to what YouTube has stated and label the rest as opinion, because the Shorts space is thick with confident claims that have no source behind them.

#The feed is the product

Shorts surface mainly through the swipeable, never-ending feed you reach by tapping the Shorts tab. They also appear in search, on the home feed, on your channel page, in subscriptions, and through notifications, but the endless feed is the engine. That format matters because it changes the unit of competition. On the watch page, you are competing for a click. In the feed, you are competing for the half-second decision to keep watching instead of flicking up.

So the algorithm is not trying to predict whether someone will click your video. It is trying to predict whether, the instant your Short starts playing, that person will stay. Everything downstream follows from that single bet.

#Engaged views are the signal that matters

Here is the distinction that trips people up. Since March 31, 2025, YouTube counts a "view" each time a Short starts to play or replay, with no minimum watch time. That number is the one you see most prominently, and it is generous on purpose. The older, stricter metric did not disappear, though. It was preserved as "engaged views," which counts only viewers who chose to continue watching past that first moment.

Engaged views are the per-viewer retention signal, and they are what actually drive distribution. A Short can rack up a huge raw view count and still go nowhere if almost everyone swipes away immediately, because the engaged-view rate, the share of people who stuck around, stayed low. When you are reading your own Shorts analytics, the raw view count tells you how many feeds you reached; the engaged views tell you whether you earned the next push.

#It judges the video, not the channel

This is the part that should change how you feel about posting Shorts at all. YouTube evaluates Shorts per-video and per-topic, not channel-wide. A Short that flops does not drag down the rest of your channel, and it does not suppress your next long-form video. The two systems are scored separately.

"Someone might enjoy your Shorts without wanting to watch your 20-minute videos, and vice versa."

Rene Ritchie, YouTube Creator Liaison

Ritchie has made this point repeatedly: Shorts and long-form audiences are treated as distinct, and YouTube tries to surface each video to the people most likely to want that specific thing. That is liberating in one direction, because a string of underperforming Shorts will not poison your main content. It is sobering in the other, because a subscriber who loves your Shorts is not automatically going to watch your long-form, and the feed will not force them to. We dig into that gap in turning Shorts viewers into long-form subscribers.

#The real metrics, and the fake thresholds

YouTube Studio does give you genuine feed-level data. "Viewed vs Swiped Away" is a real metric: it shows the share of feed impressions that were watched versus scrolled past. It is a useful gut check on whether your opening is working, and it pairs naturally with the hook and loop tactics in Shorts hooks and loops.

What is not real are the specific numeric thresholds that get passed around as gospel. You will read that "a 70% swipe-away rate kills your Short" or that you need to clear some exact retention percentage in the first three seconds. Those figures are SEO-blog fabrications, not anything YouTube has published. Watch your "Viewed vs Swiped Away" trend over time and compare your Shorts against each other, but do not chase a magic number that nobody at YouTube ever stated.

#Length changed, so the math did too

Since October 15, 2024, the maximum Shorts length is three minutes, up from sixty seconds. Any video up to three minutes with a square or vertical aspect ratio, uploaded on or after that date, is classified as a Short. That gives you more room, but it does not change the core dynamic: a longer Short still has to survive the first second in the feed, and now it has more runway over which to lose people. More length is more rope.

Context for scale: YouTube CEO Neal Mohan has said roughly 70% of YouTube channels now upload Shorts, and in his 2026 letter put Shorts at around 200 billion daily views. The feed is enormous and crowded, which is exactly why the per-swipe retention signal is the one the algorithm leans on hardest.

#Reading the feed by watching your niche

Because the Shorts algorithm is so opinionated about topic and retention, the fastest way to learn what it is rewarding in your niche right now is to watch what is actually breaking out around you. Which formats are competitors leaning into, which hooks keep reappearing, how often are they posting Shorts versus long-form. Those moves are published results, and they shift faster on Shorts than anywhere else on the platform.

#The honest summary

• The feed competes for the next swipe, not the click. Your opening second is the whole pitch.
• Engaged views, not raw views, drive distribution. Judge a Short on who kept watching.
• Shorts are scored per-video and per-topic. A flop does not hurt your channel or your long-form.
• "Viewed vs Swiped Away" is a real Studio metric. Specific kill-thresholds are invented.
• Three-minute Shorts give more runway and more chances to lose the viewer. Earn the length.