YouTube’s Video Search Algorithm
We discussed in class how search engines like Google utilize page rank algorithms to prioritize certain websites when ranking the pages relevant to a search. Google famously bought the popular video-sharing site YouTube in 2006. As an avid YouTube watcher, I was curious how YouTube determined which videos to rank first. How did they determine which videos were relevant? Was the title of the video the core determiner? The video view count? The video duration? What made YouTube’s ranking system so accurate? Because Google owns YouTube, I was surprised to learn that YouTube does not actually use PageRank. Google Search and YouTube Search determine rankings through different criteria, but all with the intent of finding the most relevant page. Google’s PageRank works by comparing the number of in-links of each page. The most relevant page will tend to be the one most prominently linked to by other pages.
YouTube’s algorithm cannot work the same way as PageRank because YouTube videos do not often link to each other. Most videos even contain basically identical content (i.e. a review of the same movie, gameplay of the same video game, etc.). Before 2012, YouTube prioritized video views. When the user entered a search term, the suggested videos would be strictly those with the most views. but matched the search term. This became an issue, however, because the top ranked video did not always match the search term better than lower-viewed videos. For example, a search for an event (a recent fight perhaps) would rank first a reaction to the event if the reaction video contained more views. When a user clicks on a video but finds it irrelevant, he or she will click immediately away. YouTube weigh such a view as much as one who saw the entire video. Because even these fake videos were garnering views and thus page rank, irrelevant videos would show up even more often.
YouTube addressed the issue by implementing a new video search algorithm in 2012. Now, the website would track how long a user stayed on a video after entering a search term. This way, the videos that showed up first would be the ones viewers stayed to watch after searching for something. This improved algorithm increased YouTube’s daily watch time tenfold (in addition to YouTube’s overall growth). YouTube content creators began investing in longer videos and thus stronger content to grow viewership. The increased view time also expanded the opportunities for ad revenue. YouTube has followed Google’s lead in placing priority on user-tailored web searches. By analyzing a person’s behavior and trends, websites improve the user’s experience and keep people using their sites.
Ultimately, YouTube search differs from Google search in that it does not track endorsements from other sites but instead tracks endorsements from invested viewers. The Youtube model caters to its video streaming capability, and the success has shown in its popularity.
Main:
Also Read: