Updated October 28, 2026. Based on YouTube’s official explanations of its recommendation system, including interviews with Todd Beaupré, YouTube’s Senior Director of Growth and Discovery.
YouTube algorithm, in short: YouTube’s algorithm does not push your video to an audience. It pulls videos for each viewer when they open the app, ranking what that person is most likely to watch and be glad they watched. Clicks and watch time still matter, but YouTube also weighs satisfaction: likes, dislikes, “not interested” clicks and post-watch surveys. To get more views, make videos a specific audience would choose and not regret.
Pull, not push: how recommendations actually work
Creators often say the algorithm “stopped pushing” their video. YouTube describes it the other way around. When someone opens the homepage, the recommendation system builds a ranking for that viewer, in that session, from the videos most likely to interest them. Apart from notifications to subscribers who opted in, nothing is sent to an audience when you publish.
This matters for strategy. Your video competes for each viewer’s attention against everything else they might watch at that moment, not against a fixed reach quota.
The signals YouTube uses
| Signal | What it tells YouTube |
|---|---|
| Click-through rate | Whether the title and thumbnail make people choose the video |
| Watch time and retention | Whether people keep watching after they click |
| Satisfaction surveys | Whether viewers say the video was worth their time |
| Likes, dislikes, shares, “not interested” | Direct feedback on the viewing experience |
| Viewer context | Time of day, device and what the viewer usually watches then; long videos tend to do better on TV in the evening |
| Viewing history | Channels and topics this viewer returns to |
YouTube has said that adding direct feedback such as surveys to ranking led people to come back to YouTube more over time, which is the outcome it optimizes for. That is why a video with fewer clicks but happier viewers can outperform one with a higher click-through rate.
Where recommendations happen
- Home: personalized for each viewer, mixing subscriptions, new channels and older videos that match their interests.
- Up next and suggested videos: based on what the viewer is watching now and what similar viewers watched next.
- Search: relevance to the query matters most, followed by engagement and quality signals.
- Shorts feed: a separate, swipe-based system. Since March 31, 2025, a Shorts view counts as soon as a Short starts to play or replay.
- Subscriptions: chronological, but YouTube has noted that many subscribers do not see every upload, because they follow more channels than they watch.
How to get more views in 2026
1. Define the viewer before the video
Because recommendations are matched to individuals, a video made “for everyone” matches no one strongly. Write down who the video is for and what they want after watching.
2. Make the packaging honest and specific
Title and thumbnail should promise what the video delivers in its first minute. YouTube’s Test & Compare tool picks thumbnails and titles by watch time share, which rewards packaging that brings the right viewers. See our thumbnail best practices.
3. Earn the first 30 seconds
Skip long intros. Confirm the viewer is in the right place, then deliver.
4. Build series and playlists
Videos that lead naturally to your next video keep viewers on your channel, which strengthens the signal that your channel satisfies that audience.
5. Treat a drop in views as information
Views fall for ordinary reasons: seasonality, a topic that has run its course, or viewers who have already seen your best videos. Compare against the same period last year and check which traffic source changed before rewriting your strategy.
6. Consider multi-language audio
YouTube’s multi-language audio tracks let one video serve viewers in several languages, which expands the pool of people it can be recommended to.
Myths worth dropping
- “Posting at the perfect time beats the algorithm.” Timing affects the first hours, but recommendations continue for months.
- “Deleting and re-uploading resets the algorithm.” It mostly loses your existing data.
- “Long videos always win.” Satisfaction, not length, is the target. A tight eight-minute video that viewers finish can beat a padded twenty-minute one.
Bottom line: The YouTube algorithm follows viewers, not creators: make videos people click and keep watching. Every change to the YouTube algorithm still rewards that, which is why chasing YouTube algorithm rumours rarely pays off.
Frequently asked questions about YouTube algorithm
Does YouTube favor long videos?
No. It favors videos viewers are satisfied with. On TVs in the evening, longer videos often perform well because that is how people watch at that time.
Do likes and comments help?
They are part of the satisfaction picture, alongside surveys and watch behavior, but asking for them does not replace making a video worth liking.
Why did my views suddenly drop?
Usually because audience interest moved, not because of a penalty. Check traffic sources in YouTube Studio to see which surface changed.
Sources
- Search Engine Journal: How YouTube’s recommendation system works (interview with Todd Beaupré)
- YouTube Official Blog: On YouTube’s recommendation system
Related: YouTube monetization requirements · YouTube keyword research
