Using YouTube Retention Data to Edit for the Web
Today, I was checking on some video analytics for a client’s YouTube account. I manage the postings for the account, and use the data to develop stronger videos as time goes on.
YouTube recently redesigned their analytics dashboard (again), and one of the changes was the addition of “Audience Retention.” This feature kind of existed before as “Hot Spots.” Hot Spots compared your video to other videos across YouTube of similar length, and showed you a relative graph of how much of your audience you were retaining relative to the other videos. The graph would show you peaks and valleys so you could check where people were likely skipping forward (valley), rewinding a section (peaks), or just leaving altogether (downward slope). It was a really good indicator of what people liked or didn’t like about your video, and I used it a lot to see where I was losing people. It was a bit vague though — it didn’t really offer any true numbers, just “hot!” or “cold!”
The Audience Retention feature now calls this “Relative Audience Retention,” one of two options available to view. The other option? Absolute Audience Retention. A real time, moment by moment graph of your viewers’ attention to your video – and ONLY your video. I was checking some videos today on the Absolute Retention, including one of the most popular postings in the last year. This video has gotten a ton of views and is currently sitting at number 4 in the top 10 videos on the account.
The retention rate for the final two minutes of the video: 15%.
This video is the last part in a four part series following two guys as they go through a school. It’s shown to new students in the classroom on the first or second day of classes, and I’ve been told that it’s helped a lot to improve graduation rates since it puts people at ease. We’ve gotten a lot of really good feedback and students are engaged in the video the whole time (mostly, anyway). They remember the information provided, and their questions to instructors are better since they are more informed about everything.
The structure of the piece is pretty simple, basically three quick acts. The setup: it’s the final day of class, the students have to pass a test, and there’s a dramatic build up. Will they fail? Will they succeed? Then they pass and talk about how it felt and where they go next. Informative (if a little too talky at times but that’s another post), sucks the students in, plays well on the big screen.
It gets a LOT of views on YouTube, but it simply doesn’t hold the attention of the viewer the same way. They barely make it halfway through, and certainly don’t make it to the end where we tell them what action they should be taking next. 7000 views in the last few months, but only 15% of those viewers are seeing the entire video.
This has made me think a lot about how to approach an edit for the web. In this case, the YouTube views fall off dramatically as soon as viewers know the outcome of the tests – basically, after the climax, they run for the door (typical.) The topic of the video is obviously a really popular search term since so many people hit it. But then they bail. Why? I’m assuming they just wanted to see what the test was like, and were engaged just enough to care if the guys passed. They aren’t a captive audience, they’re distracted by a million other things on the Internet, and while my video is answering their questions, they would prefer to be entertained instead. My conclusion is that I need to do more videos with this subject, and be more clever about how the action unfolds, and how I deliver the calls to action. The information you can gain from these graphs is not always so abstract. It can be as simple as checking to see if a setup falls flat, if a hook is grabby enough, or I suppose you could even compare versions of videos to see if different edits or graphics make a difference.
The interesting thing is that if you look at the Relative Retention on this video, it’s not awful. Remember, this compares the audience retention to videos of similar length to yours.
What does that tell me? It tells me that YouTube users have no frickin’ attention span, and it’s all part of the challenge when editing for the Internet.And there’s an awful lot of horrible crap on YouTube, weighing down the site.
And a lot of web video experts will tell you different things about this. The big thing right now seems to be “if your content is engaging, people will watch the whole thing no matter the length!” Yea, maybe for some audiences. But not all of them. This video I’m talking about is tightly edited, packed with good information, well-recieved among students, and answers a lot of questions. But YouTube viewers in my neck of the woods aren’t interested. They might stick around if there was a nude woman, maybe. And that’s a big maybe. I have to play a game with them and try new things to engage them and make them realize that this is the exact information they were seeking.
Go check out that Audience Retention tab on YouTube and see what juicy bits you can glean from it about your media. Are people rewinding a part to watch repeatedly, maybe a comedic cut? Are people fast-forwarding in a section that lags? Are you losing viewers in droves? Why? Of course, it’s all one big psychological study. Who really knows what goes on in the brain of a typical YouTube user? And do any of us REALLY want to know?