How-To Guide

How does the YouTube Shorts algorithm work?

YouTube names what ranks a Short: choosing to watch, staying, and enjoying it. What is confirmed, what is myth, and how to read your own Shorts data.

8 min read

By the Superdirector editorial team. YouTube Help, the YouTube Blog and vendor pages read November 7, 2026.

At a glance

How does the YouTube Shorts algorithm work?

The YouTube Shorts algorithm matches each Short to individual viewers and ranks it on how they respond when it is shown: whether they choose to watch or swipe away, how long they stay, and whether they enjoyed it. YouTube names these as appeal, engagement and satisfaction, and adds the viewer's watch and search history, topic interest, competition and seasonality. It publishes no weights, no pass mark for swipe-away and no posting-frequency rule.

  • YouTube names three signal groups: appeal, engagement and satisfaction, then personalizes by viewer and by surface.
  • Views are counted from the first frame since August 24, 2026; engaged views are the stricter count.
  • Percentage thresholds you read online are vendor hypotheses; YouTube publishes none.
One Short runs three numbered checks with a single viewer, chose to watch, stayed and was glad, and the result loops back as more or fewer viewers are shown the Short.

YouTube ranks a Short on how each viewer responds, in the order they meet it: appeal, engagement, satisfaction. Original diagram; source basis is YouTube Help's Shorts search and discovery page and its performance signals page, read November 2026.

Superdirector

Make it work for your brand

Find it. Remake it. Post it. Superdirector turns the viral formats in your niche into your brand's carousels, X posts and video scripts, and keeps the calendar full.

How YouTube ranks a Short, in YouTube's own words

The YouTube Shorts algorithm matches each Short to individual viewers and ranks it on how they respond when it is shown: whether they choose to watch or swipe away, how long they stay, and whether they enjoyed it. YouTube names these as appeal, engagement and satisfaction, and adds the viewer's watch and search history, topic interest, competition and seasonality. It publishes no weights, no pass mark for swipe-away and no posting-frequency rule.

That answer comes from YouTube's Shorts search and discovery page and its performance signals page. YouTube says its system "has no opinion about what type of Shorts you make" and that Shorts are ranked on performance and viewer personalization. Performance is read in the order a viewer meets a Short: first the choice to watch, then whether they stay, then whether they liked it.

Signal groupWhat YouTube says it looks atWhere you see it in YouTube Studio
AppealWhen a Short is recommended, whether the viewer chooses to watch it, ignores it or taps "not interested"Shown in feed, and How many chose to view (the viewed versus swiped away report)
EngagementWhether viewers stick around: the percentage who chose to view, average view duration and average percentage viewedWatch time and audience retention in the Shorts analytics
SatisfactionWhether viewers enjoyed the Short, from likes and post-watch survey resultsLikes in the key metrics card. The Shorts analytics page lists no survey report

Three outside factors change how many people see a Short even when its numbers are good. Topic interest is how many people worldwide are watching Shorts on the subject. Competition means a Short can get fewer impressions if other channels' Shorts are performing even better. Seasonality moves traffic around holidays and through the year. On top of that, each viewer's own history shapes the feed: Shorts and channels they enjoyed, topics they watch, and sounds they engaged with.

Ranking also depends on where the viewer meets the Short. YouTube's discovery page lists six places. The Shorts tab ranks by performance and by relevance to that one viewer. Home shows Shorts YouTube believes a specific viewer will enjoy. The Subscriptions tab shows the most recent uploads from channels the viewer is subscribed to. Search ranks on how well the metadata matches the query and on whether viewers click and watch. The sounds and hashtags surface groups Shorts by trending sound, and YouTube says using sounds from the Audio Library raises a creator's chance of appearing there. The Trends page lists Shorts by popularity.

What the pages do not contain is a number. They list the signals; none gives a weight for each, a swipe-away percentage that unlocks reach or a retention bar. Any such figure you read is someone's hypothesis, and a later section sorts them from the rules YouTube does state.

Make your own contribution unmistakable

The newest change to the ranking is about whose work it is. On October 1, 2026, a post from YouTube's support team (TeamYouTube) in the YouTube Community told creators the company is updating its recommendation systems to further prioritize original content and reduce the reach of content re-uploaded from other creators without adding anything of their own. It says channels that mainly aggregate or re-upload other creators' videos without significant changes will likely see less distribution in the Shorts feed, and that creators using someone else's clip should add their own commentary, edits or storytelling. TOS.gg's write-up reports the same change.

The same line runs through the monetization rules. YouTube's Shorts monetization policy lists reuploaded content from other creators and compilations with no original content added among the views that do not count toward revenue sharing.

For a brand, the practical reading is narrow. Your own footage, voice, numbers and customers are original by definition. A Short rebuilt from another account's hook and structure, filmed and written by you, adds your own work. A trimmed copy of someone else's clip does not. YouTube's notice is about content re-uploaded from other creators; it says nothing about your own videos appearing on more than one platform.

Your view count changed twice, so read the right column

Two definition changes affect how a Shorts dashboard reads, and neither changed recommendations. Compare views over time only if you know which definition each month used.

DateWhat changedWhat YouTube says it did not change
March 31, 2025A Short's view counts when it starts or replays, with no minimum watch time (Shorts help). The older metric stays in Analytics as "engaged views".Partner Program eligibility and Shorts ad revenue sharing stayed on engaged Shorts views
August 24, 2026Views across YouTube, Shorts included, are counted from the first frame. An engaged view is clicking to watch or watching past the initial seconds (YouTube Blog)."No change to recommendations"; the vast majority of analytics stay anchored on engaged views

An engaged Shorts view, per YouTube's August 12, 2026 post on qualified views, means the viewer stayed past the initial seconds, and it does not include loops. So a rising headline count can come from starts and replays. Read engaged views, average view duration and the share who chose to view before you call a Short a hit.

If monetization is part of your plan, the thresholds count qualified views, not the headline. YouTube's August 10, 2026 Partner Program post says that from February 1, 2027 a channel needs 10 million qualified Shorts views over 90 days to share in Shorts ad and subscription revenue, and that new Partner Program applicants need 8,000 qualified watch hours in 365 days or 20 million qualified Shorts views in 90 days. Check the dates against YouTube's page before you plan around them.

Rules you will read, checked against YouTube's pages

Every Shorts algorithm guide repeats a few rules. This table sets the common ones against what YouTube's own help pages say, as read on November 7, 2026.

You will readWhat YouTube's pages say
Post every day or the algorithm drops youThe Shorts page says "there is no minimum posting cadence required" for videos to perform well. The general discovery FAQ says growth in views across uploads is not correlated with time between uploads.
Shorts will hurt your long-form videos"Shorts can help with audience discovery but can’t hurt your Video performance."
Tags and hashtags are the leverThe discovery FAQ says tags mostly help correct spelling mistakes. In Metricool's sample of 799,718 videos, hashtags drove under 1% of views.
Set your channel country to reach a marketLocation settings are not used to inform how videos are recommended.
Limited-ads or demonetized videos get less reachThe recommendation system does not know which videos are monetizing. Violent or graphic content can be recommended less, because of the content, not the ad status.
Post at the right hourPublish time is not known to affect long-term performance. Posting when your audience is active can help early viewership.
A swipe-away rate under some percentage kills a ShortYouTube publishes no such threshold. Socialync gives 65% retention for Shorts under 30 seconds and says on the same page its thresholds are "not platform confirmed"; Likes.io says above 50% is strong and below 30% kills distribution, from "anonymized telemetry" of its own campaigns.
Cross-post to other apps to win YouTube's favorNot a listed signal. Metricool, reading the same pages, says cross-posting can bring outside traffic but is not something the ranking system counts.

Treat the last two rows as a rule for the whole table: a percentage with no YouTube page behind it is a test to run on your own Shorts, not a rule to follow. vidIQ describes an explore and exploit pattern, a small seed audience first and a wider one if it responds. That fits YouTube's description of testing a Short on the people it is recommended to, though YouTube does not use those words.

What a large 2026 sample shows, and what it cannot

YouTube publishes no outcome data by account, so the best public numbers come from vendors. Metricool's 2026 YouTube study covers 799,718 videos from 71,177 accounts. It is a vendor sample, correlational, and its customers are not all of YouTube. Metricool's own caveat applies to every year-over-year view figure: any comparison across March 2025 straddles a change in what counts as a view.

Finding in Metricool's sampleNumberWhat to take from it
Share of organic views from the Shorts feed61.54%The feed, not search or subscribers, is where Shorts are found
Average views per Short, 2025 to 20262,343 to 5,321Reach rose, partly because the view definition changed
Average watch per ShortAbout 16 seconds, roughly half a minute less than a year earlierThe first seconds carry most of the watch
Engagement rate, 2025 to 20264.94% to 1.85%Engagement per view fell as counted views rose
Views per Short by posting rate5,872 at 2 to 4 a week; 5,087 at 4 to 7; 3,535 at half a post a week or lessPer-post results are best at a moderate pace; monthly totals still rise with volume
Accounts under 2,000 subscribers versus over 1 million482 to 1,290 views per Short, up 168%; 46,326 to 11,954, down 74%Small accounts can reach strangers because the feed ranks the Short, not the channel

The posting-rate row is a per-post average, so it says nothing about what posting more does to your total views. It also does not contradict YouTube's line that there is no minimum cadence. Use the numbers to size a routine you can keep, not to set a quota.

Make the first seconds earn the stop

Appeal is the first test, and your analytics show it in two places on the Shorts report: shown in feed counts how often a Short appeared, and How many chose to view is the share of those showings where the viewer watched and did not swipe away. YouTube's own advice puts the weight on the opening: the first seconds are when viewers decide to stay or move on, and the intro should deliver on the promise of the title.

Two habits follow. Put the topic and the payoff on screen in the first second, and cut any greeting, logo or setup. And frame a niche idea around a problem a wider group shares: YouTube suggests this so a viewer who is not an expert still has a reason to stop. Compare each Short's chose-to-view share with your own median. YouTube publishes no pass mark, so the only fair comparison is your own recent Shorts.

Cut what makes people leave

Average view duration and average percentage viewed are the two engagement signals YouTube names. Its FAQ adds that relative watch time matters more for short videos and absolute watch time more for longer ones, so a 20-second Short watched to the end sends a stronger signal than a 3-minute one abandoned early. Metricool's 2026 sample put the average Short at about 16 seconds of watch, so a setup that takes ten seconds has spent most of the viewer's time.

Open the audience retention graph for the Short and find the second where the line falls fastest. YouTube suggests measuring retention at key moments, and that second is yours. Cut it, or move the payoff in front of it. The retention rate and watch time pages define both terms if you share the numbers with a teammate.

Deliver what the opening promised

YouTube says it reads enjoyment from likes and post-watch survey results. You cannot open the survey results, but you can read likes and comments against your median. YouTube is clear that likes are one of hundreds of signals, so a Short with fewer likes is a prompt to read the comments, not a verdict.

The common cause of high watch time with low likes is a payoff that does not match the promise. A hook that says "3 mistakes" followed by one mistake and a plug does that. Rewrite the ending to deliver what the opening promised, then compare the next Short's like share with your median.

Judge each Short against your own median

YouTube advises looking at both Shorts that did well and ones that did not, and not measuring every video against a viral hit. Its audience retention report offers a comparison with YouTube videos of similar length, which is the same idea: your recent Shorts are the baseline. Divide each figure by the median of your last ten Shorts, and change one thing per Short so you can tell what moved the number. YouTube publishes no thresholds, so the comparisons in this table are ours.

What you see in StudioLikely weak signal groupOne change to test on the next Short
Shown in feed is low next to your medianTopic interest or competitionReframe the idea around a problem a wider group has, then compare shown in feed
Shown in feed is high, but few chose to viewAppealRewrite the first line so the topic and payoff are clear in the first second; cut any intro
Many chose to view, but average percentage viewed is lowEngagementCut the slowest middle seconds and move the payoff earlier
Watch time is fine, likes and comments are lowSatisfactionMake the ending deliver what the opening promised; check that the title and the hook say the same thing
Views up, engaged views flatDefinition change or loopsJudge the Short by engaged views and average view duration, not the headline count

Log each change and its result in the Shorts change log and mark a week later whether the Short beat your median. After five or six rows you will know whether your weak spot is the opening, the middle or the ending. The short-form strategy guide covers hooks and pacing, and the YouTube Shorts strategy guide covers the plan around them.

Where Superdirector fits

The algorithm above belongs to YouTube; no tool changes it. Superdirector, which publishes this page, handles the finding and remaking part of the loop. It tracks YouTube, Instagram, TikTok and X accounts, competitors, keywords and hashtags, flags the posts and formats taking off in your niche, and rebuilds the same hook and structure as carousels, X posts and video scripts in your brand. On Growth it posts Instagram carousels and single X text posts. AI-made video and full autopilot are coming soon, so a Short is still filmed, edited and uploaded by you, and you remain its publisher. See what runs today. The competitor content guide shows how to keep the log of what is working in your niche.

Frequently asked questions

How does the YouTube Shorts algorithm work?

It matches each Short to individual viewers and ranks it on how they respond when it is shown. YouTube names three groups of signals: appeal (do they choose to watch or tap not interested), engagement (average view duration and average percentage viewed) and satisfaction (likes and post-watch surveys). Viewer history, topic interest, competition and seasonality also apply. YouTube publishes no weights or pass marks.

Does posting more Shorts help you get more views?

YouTube says there is no minimum posting cadence required for videos to perform well. In Metricool's 2026 sample, per-post views were highest at 2 to 4 Shorts a week (5,872 on average), while monthly totals still rose with volume. Treat that as a vendor observation, not a YouTube rule.

What is viewed versus swiped away on YouTube Shorts?

It is the share of times a Short was shown in the feed and the viewer chose to watch and did not swipe away, listed in Shorts analytics as How many chose to view. YouTube says it feeds ranking as the percentage of viewers who chose to view, but publishes no percentage that guarantees more reach.

Did YouTube change how Shorts views are counted?

Yes, twice. Since March 31, 2025 a Short's view counts when it starts or replays with no minimum watch time. Since August 24, 2026 views on all formats are counted from the first frame. YouTube keeps engaged views, past the initial seconds and without loops, and says recommendations did not change.

Does YouTube penalize reposted or re-uploaded Shorts?

On October 1, 2026 YouTube said it is updating recommendations to reduce the reach of content re-uploaded from other creators without adding anything of your own. The notice concerns other creators' content; it does not mention your own videos posted on more than one platform.

Do hashtags or tags change how Shorts are ranked?

YouTube's discovery FAQ says tags mostly correct spelling mistakes, and its Shorts page lists sounds and hashtag pages as one way viewers find Shorts. In Metricool's sample, hashtags drove under 1% of views, so treat them as context, not as the lever.

Related practical guides

Score your last ten Shorts today

Open the Shorts report, copy ten Shorts into the diagnosis sheet and mark which signal group is weakest for each.