{"id":51678,"date":"2025-01-27T20:41:27","date_gmt":"2025-01-27T20:41:27","guid":{"rendered":"https:\/\/peraltafinancing.com\/business\/marketing\/how-youtubes-recommendation-system-works-in-2025\/"},"modified":"2025-01-27T20:41:27","modified_gmt":"2025-01-27T20:41:27","slug":"how-youtubes-recommendation-system-works-in-2025","status":"publish","type":"post","link":"https:\/\/fivemor.com\/?p=51678","title":{"rendered":"How YouTube\u2019s Recommendation System Works In 2025"},"content":{"rendered":"


\n<\/p>\n

\n

In a recent video interview<\/a>, YouTube Liaison Ren\u00e9 Ritchie spoke with Todd Beaupr\u00e9, YouTube\u2019s Senior Director of Growth & Discovery, to discuss the platform\u2019s recommendation system functions and what creators can expect this year.<\/p>\n

Their discussion revealed how time of day, device type, viewer satisfaction, and the advent of large language models (LLMs) are reshaping YouTube\u2019s algorithms.<\/p>\n

Here\u2019s what you need to know about YouTube\u2019s recommendation system and how it works.<\/p>\n

Personalized Recommendations<\/h2>\n

One of the central themes of the interview is YouTube\u2019s focus on matching content to individual viewer preferences.<\/p>\n

According to Beaupr\u00e9:<\/p>\n

\n

\u201cOften times creators will say hey, uh the recommendation system is pushing out my video to people or why isn\u2019t it pushing out my video yes they they may ask that and the way the work it works is it\u2026 isn\u2019t so much about pushing it out as much as it\u2019s pulling\u2026\u201d<\/p>\n<\/blockquote>\n

He goes on to explain that YouTube\u2019s home feed prioritizes content based on what each viewer is most likely to enjoy at any given moment:<\/p>\n

\n

\u201cWhen you open the homepage, YouTube is going to say hey Rene is here, we need to give Rene the best content that is going to make Rene happy today.\u201d<\/p>\n<\/blockquote>\n

Metrics & Satisfaction<\/h2>\n

While click-through rate (CTR) and watch time remain important, YouTube\u2019s system also accounts for user satisfaction gleaned through direct surveys and other feedback signals.<\/p>\n

Beaupr\u00e9 notes:<\/p>\n

\n

\u201cWe introduced this concept of satisfaction\u2026 we\u2019re trying to understand not just about the viewer\u2019s behavior and what they do, but how do they feel about the time they\u2019re spending.\u201d<\/p>\n<\/blockquote>\n

He explains YouTube\u2019s goal is to cultivate long-term viewer satisfaction:<\/p>\n

\n

\u201c\u2026we look at things like likes, dislikes, these survey responses\u2026 we have a variety of different signals to get at this satisfaction\u2026 we want to build a relationship with our audience just as creators want to do with their fans.\u201d<\/p>\n<\/blockquote>\n

Evergreen & Trending Content<\/h2>\n

YouTube\u2019s algorithms can identify older videos that become relevant again due to trending topics, viral moments, or nostalgic interests.<\/p>\n

Beaupr\u00e9 cites the system\u2019s ability to pivot:<\/p>\n

\n

\u201c\u2026maybe like right now there\u2019s a video that that reaches a certain audience but then like in six months\u2026 that makes this video relevant again\u2026 if it\u2019s relevant and maybe to a different audience than enjoyed it the first time.\u201d<\/p>\n<\/blockquote>\n

Context: Time, Device, & Viewer Habits<\/h2>\n

Beaupr\u00e9 revealed YouTube\u2019s system may show different kinds of content depending on whether someone is watching in the morning or at night, on a mobile phone or a TV:<\/p>\n

\n

\u201cThe recommendation system uses time of day and device\u2026 as some of the signals that we learn from to understand if there\u2019s different content that is appealing in those different contexts\u2026 if you tend to have a preference for watching news in the morning and comedy at night\u2026 we\u2019ll try to learn from other viewers like you if they have that pattern.\u201d<\/p>\n<\/blockquote>\n

Fluctuations In Views<\/h2>\n

Creators often worry if their views dip, but Beaupr\u00e9 suggests this can be a natural ebb and flow:<\/p>\n

\n

\u201c\u2026the first thing is that that is natural\u2026 it\u2019s not particularly reasonable to expect that you\u2019re going to always be at your highest level of views from all time\u2026 I would encourage you not to worry about it too much\u2026\u201d<\/p>\n<\/blockquote>\n

He also recommends comparing metrics over longer periods and leveraging tools like Google Trends:<\/p>\n

\n

\u201c\u2026we do see seasonality can play a role\u2026 encourage you to look beyond\u2026 90 days or more to kind of see the full context.\u201d<\/p>\n<\/blockquote>\n

Multi-Language Audio<\/h2>\n

Many creators are exploring multilingual audio to broaden their audiences.<\/p>\n

Beaupr\u00e9 highlights how YouTube has adapted to support dubbed tracks:<\/p>\n

\n

\u201c\u2026we needed to add some new capabilities\u2026 aware that this video actually is available in multiple languages\u2026 so if you\u2019re a Creator who\u2019s interested in extending your reach through dubs\u2026 make sure that your titles and descriptions\u2026 are also uploaded [in] translated titles and descriptions\u2026\u201d<\/p>\n<\/blockquote>\n

He also emphasizes consistency:<\/p>\n

\n

\u201cWe\u2019ve seen in particular creators who dub at least 80% of the\u2026 watch time\u2026 tend to have more success than those who dub less\u2026\u201d<\/p>\n<\/blockquote>\n

LLM Integration<\/h2>\n

Looking to the future, large language models (LLMs) enable YouTube to better understand video content and viewer preferences.<\/p>\n

Beaupr\u00e9 says:<\/p>\n

\n

\u201c\u2026we\u2019ve applied the large language model technology to recommendations at YouTube to\u2026 make them more relevant to viewers\u2026 rather than just memorizing that this video tends to be good with this type of viewer\u2026 it might actually be able to understand the ingredients of the dish better and maybe some more elements of the video style\u2026\u201d<\/p>\n<\/blockquote>\n

Beaupr\u00e9 likens it to an expert chef who can adapt recipes:<\/p>\n

\n

\u201c\u2026we want to be more like the expert chef and less like the\u2026 memorized recipe.\u201d<\/p>\n<\/blockquote>\n

Key Takeaways For Creators<\/h2>\n

Here are the top takeaways from their 21-minute conversation on the YouTube recommendation system.<\/p>\n

    \n
  1. The recommendation system is about \u201cpulling\u201d content for each viewer, not pushing videos universally.<\/li>\n
  2. Metrics like CTR and watch time matter, but satisfaction (likes, dislikes, surveyed feedback) is also essential.<\/li>\n
  3. YouTube can resurface older videos if renewed interest emerges.<\/li>\n
  4. Time of day and device usage influence recommendations.<\/li>\n
  5. View fluctuations are normal\u2014seasonality, trending events, and external factors can all be at play.<\/li>\n
  6. Dubbing and translated titles may help reach new markets, especially if a high percentage of your content is available in the same language.<\/li>\n
  7. Large language models empower more nuanced understanding\u2014creators should stay attuned to how this impacts discovery.<\/li>\n<\/ol>\n

    Watch the full interview below.<\/p>\n