Algorithm & Discovery

YouTube Algorithm

The YouTube algorithm is not one model but a collection of ranking systems that decide what shows up on Home, the Subscriptions feed, Suggested Videos, Search results, the Shorts shelf, and Browse on TVs. Each surface uses a slightly different ranking recipe, but the core inputs are Click-Through Rate, Average View Duration, Watch Time, session contribution, engagement, and topical relevance to the individual viewer.

Last updated: May 17, 2026
Quick definition

The YouTube algorithm is the recommendation system that decides which videos appear on Home, Suggested, Search, and Shorts — optimised primarily for viewer Watch Time.

Why YouTube Algorithm matters for YouTube creators

Almost every creator's growth is governed by the algorithm. Subscriber-only distribution accounts for roughly 10-15% of total impressions for most channels; the remaining 85-90% comes from algorithmic surfaces. That means the algorithm decides whether your video reaches new audiences or stays bottled up inside your existing subscriber base. Understanding which surface delivers your impressions — Browse vs Suggested vs Search — also tells you what to optimise. Browse rewards strong thumbnails and immediate hooks; Suggested rewards session contribution (does this video keep viewers on YouTube?); Search rewards topical match plus retention.

How YouTube Algorithm works

The recommendation pipeline is two-stage: a candidate generator pulls a few hundred personalised videos for each viewer, then a ranker scores them against expected viewer satisfaction. Satisfaction is approximated from CTR, AVD, Watch Time, "Like" rate, survey responses, "Not Interested" signals, and dwell on the recommended thumbnail. The system favours videos that keep viewers on YouTube longer in the current session — not necessarily the same channel.

YouTube Algorithm in practice

A creator uploads two essays in the same week. Essay A gets 9% CTR and 60% AVD on Day 1 and goes viral via Browse, hitting 2M views. Essay B gets 4% CTR and 55% AVD and never breaks 80K despite the strong retention — Browse needs both signals.

A video tagged for "iPhone 16 review" wins Search for that query and quietly earns 1M views over six months — different algorithm, different success pattern.

See YouTube Algorithm on real channels

FameLifter pulls public youtube algorithm data for 500K+ YouTube channels — refreshed hourly via the official YouTube Data API v3.

Frequently asked questions

Is there really only one YouTube algorithm?
No. Home, Suggested, Search, Shorts, and Subscriptions each have their own ranking. They share core signals (CTR, AVD, Watch Time) but weight them differently.
Does the algorithm penalise older videos?
Not by age alone. Videos with strong evergreen retention can earn impressions for years. What the algorithm penalises is decaying engagement — falling retention or rising "Not Interested" rates.
Can you "trick" the algorithm?
Not durably. Misleading thumbnails get clicks but the resulting low AVD pulls the video back down within 48 hours. The most reliable path is honest thumbnails plus strong retention.