Why Video Discovery Sucks
How do you find a video you don’t know about?
Imagine this very common scenario. You are at your computer and want to watch videos to kill some time. You do not have any specific video in mind, but you want something that’s fresh, popular, and in this case funny. Without the knowledge of which channels provide that, you go straight to YouTube search and enter the word ‘funny.’ The results? You mostly see compilation videos of pranks, bloopers, and fails. Sure these types of videos can be entertaining, but they hardly reflect the top videos online at that given moment that are funny. This way of discovering trending videos is broken.
YouTube is the World’s second largest search engine.
It is hard to believe how ineffective YouTube can be at helping users find what they are looking for when you consider just how massive it has become. YouTube is the second largest search engine in the world after Google, with 3 billion unique searches monthly. Just because it is popular does not mean users are receiving the best results possible. YouTube search consistently pulls up videos that do not meet the criteria of users. Consider that YouTube videos have a bounce rate of roughly 43%.That means 43% of the time a user opens a video, they close it almost immediately, realizing it was not what they were searching for. At the scale YouTube operates, that is an outstanding amount of disappointed users.
It is not to say YouTube always fails its users. It is no coincidence the site has become what it is. If you have some idea of what you are looking for; a specific event, a Youtuber series, or a scene, its search works flawlessly. Their ranking system, which relies mostly on watch-time (the total amount of time viewers have spent watching a video) insures that the results you get are the most engaging videos with your keywords. The issue is that on a site that gets 300 hours of new video content every minute, it is difficult to know what is out there to search for.
Why does video discovery have to suck?
YouTube’s discovery problem is not unique. All of the big video hosting sites fail to provide the ideal video discovery experience to its users. They fall short because of the way search currently works. By scanning a tittle and description for keywords, search falls victim to the wording used by the uploader of the videos. If a truly funny video is uploaded by a master digital marketer it may include the type of wording that will show up in common searches, but that is rare. The majority of the time the type of videos that go viral are uploaded by an average Joe, who assigns tittles that describe the video but are not optimized to benefit search or discovery.
Is there a solution?
There is a simple solution to ensure that search results reflect the content of the video, and in part, better fulfill what the user is searching for. That solution is commentary. Relying purely on tittles and descriptions makes for ineffective results. Rather, going into the comments people have made about a video and indexing them allows for a far more powerful search. The manner with which people talk about a video in Comments reflects the same language and connotations used in searches. Dissecting what people have said about the video, and identifying what part of the videos they are referencing, allows for a deeper understanding of exactly what the video is about and what level of engagement it is receiving. Comments are much more likely to match the way a person would search for a type of video.
Unreel has pioneered this approach and incorporated comments about specific moments within videos into its search algorithm. This means that when you enter a keyword into search on Unreel the results are comments at time-stamped moments in the videos that have elicited a response that uses your keyword. If you search ‘funny,’ the videos’ results are the moments in videos that people actually called funny, not videos with the word funny in the tittle. This ensures you are seeing what you wanted to see, even if you were not sure exactly what that was.