YouTube Shorts algorithm

The YouTube Shorts algorithm is the recommendation system that decides which vertical videos under three minutes appear in the Shorts feed for each viewer. It ranks candidates based on watch behavior, such as swipe-away rate and completion, rather than subscriber count or upload time. It operates per video, not per channel.

Why it matters

If you misunderstand this system, you will optimize the wrong things. Channels with zero subscribers routinely get millions of views on a single Short because distribution starts with individual video performance, not audience size. Spending months building a subscriber base before publishing does nothing; the feed tests each Short against cold viewers regardless of who made it.

It also changes how you read failure. A Short that gets almost no reach was not suppressed by an enemy; it failed a specific measured behavior, most often early swipe-aways. Creators who know this diagnose the first two seconds instead of blaming shadowbans, and they iterate on hooks rather than on posting schedules.

An example

Niche choice interacts directly with what the algorithm rewards. In ViewMade's keyword measurement dated 23 August 2026, the "aviation disasters" niche scored 94 out of 100 for demand but only 85 for saturation, giving it an opportunity score of 62, the highest among the niches measured. The top-performing video there, "How the Crash of Flight 4590 Destroyed Concorde's Mystique" from Smithsonian Channel Aviation Nation, had 10,982,012 views against a channel of only 663,000 subscribers. A video reaching far beyond its channel's subscriber base is exactly the pattern the Shorts algorithm produces: it distributes to interested viewers one by one, independent of the channel's existing audience.

Terms people confuse this with

• Faceless YouTube channel: a channel format with no on-camera presenter. The algorithm treats these videos like any other; being faceless neither helps nor hurts ranking.
• Faceless niche: a topic area suited to channels without presenters. The algorithm decides reach within a niche; the niche itself decides how much demand exists.
• Niche saturation: how crowded a topic already is with competing content. High saturation makes algorithmic wins harder because more videos compete for the same feed slots.
• Niche validation: measuring demand and competition before committing to a topic. It uses observed data rather than assuming the algorithm will find an audience for any video.

Where this shows up in ViewMade

ViewMade's keyword engine scores each niche's demand and saturation before you commit production budget, so the topic you pick matches what the feed can actually distribute.