How does the YouTube algorithm work?

17 Tem 16:05

YouTube's recommendation algorithm is like TikTok's "For You" filter, but more technical and based on historical data. It doesn't just look at what you watch, but how long you watch, what percentage of the videos you complete, and whether you interact (likes, comments, channel subscriptions). If you abandon a video after 10 seconds, the algorithm understands that you're not interested and reduces its appearance in your recommendations. Another key point: competition between channels. If you always watch cooking videos but one day search for something about DIY, the algorithm will prioritize channels that mix both topics (like *Tasty* teaching budget-friendly college recipes). It also penalizes obvious "clickbait": if a video misleads with its title but doesn't deliver, the system shows this with lower retention and files it away in the fridge of unpopular recommendations.

17 Tem 16:29

To be honest, when I first started uploading videos on YouTube, something happened that made me think a lot about how the algorithm works. I uploaded a super detailed tutorial on how to create a responsive menu with CSS Grid, and at first, the views came in really slow—like nobody was watching it. The funny thing is that after two days, I changed the title to something like "How to Center Divs in 5 Minutes" (it was more technical before) and shared the link in a Facebook group that was exactly looking for that. Within two weeks, I went from 20 views to over 5,000, and almost 40% came from "Recommended." What I noticed was that the algorithm prioritized videos where people stayed longer (mine was 7 minutes long, and most watched it until the end) and where there were a lot of "Likes" in the first few days. I also saw that quick comments like "Thanks, this helped!" helped the video gain more traction. The craziest part was that the second video I uploaded on the same topic, even though it was less detailed, was already on the first page of search results for "responsive menu CSS" in just three days. It was like the algorithm already knew my channel had authority on that topic because of the engagement from the first video. Since then, I understood that not only does the content matter, but also how people interact with it in the first 48 hours.

17 Tem 18:23

YouTube's algorithm actually takes into account much more than just your watch history—it kind of "pampering" the viewer to keep them more engaged with the content. I think it's built on three main pillars: **personal preferences (watch history)**, **entertainment value (watch time and replays)**, and **interaction patterns (likes, comments, shares)**. For example, if you watch the first 15 seconds of a video and then close it, the algorithm marks that video as "weak" and doesn’t recommend similar content as often. But if you stay for 10 minutes, it starts adding related videos to your recommendations. Based on my practical experience, the algorithm is also sensitive to the viewer’s **emotional reactions** (like frustration or excitement). For instance, if you linger on a sad video and click the "Comment" button right away, the system will prioritize more controversial or emotional content on that topic. When I teach Arduino projects to students, I also ask open-ended questions like, "How would you improve this?" in the middle of the videos. This encourages viewers to comment, which triggers the algorithm to recommend more content on that topic—the system is basically running a kind of "engagement hunter."