Now, you may have seen this thumbnail on YouTube. I mean, I can actually basically guarantee that YouTube has been shoving this in your face like, “Click! Click! Click it again!” So you might be wondering why. Why did this video in particular go viral? Well, that is what I’m gonna set out to explain in this video but I’m gonna start with something that seems unrelated, which is YouTuber burnout. Now, there’s no secret, there have been a lot of YouTubers recently talking about burnout: people like Casey Neistat, Superwoman, Ryan Higa, Grace Helbig, and many, many, MANY more, and I think, when you have all of these people with different YouTube channels, different experiences, they’re all saying something similar; I think it’s worth understanding that there may be sort of a common factor at play amongst all of them. So what I’m actually gonna present is a kind of theory of everything when it comes to YouTube; everything from YouTuber burnout, through to videos going viral; why do videos go viral on the platform right now. Okay, so, you know, I think as a viewer your response to burnout might be to say, “Well these YouTubers are just entitled” or, “They’re soft” or, you know, “Boohoo, you have the best job in the world” or your other response might be that burnout happens in every industry, so why should we care about this YouTuber thing, but I think that YouTuber burnout is a more specific phenomenon; I think there’s something particular going on here. and it all starts with the YouTuber life cycle. I mean, if you think about a YouTuber who’s complaining about burnout, you know they’ve had some success on the platform where their views are rising – and of course that’s exhilarating, it feels great and you’re working pretty hard, but you’re seeing it pay off, and then at some point your views start to falter so maybe you work a little bit harder but things don’t really change and-and the views, you know, continue to fall. I think there’s a certain thing to do with the psychology of anchoring here. You know, once you have a certain number of subscribers, then, you know, a hundred thousand views which might be amazing in the beginning feels like a horrible disappointment. So there’s a real problem between sort of expectations and reality, and to me, this is a big part of why YouTuber burnout happens. Now, if you look at what the YouTubers are saying about this, you find that a lot of them are taking responsibility themselves, and they’re also making some clear cause-and-effect relationships, which you know this kind of seemed to make sense: okay, so this idea we assume that views reflect something about video quality, we kind of make that assumption, and so when there are, you know, fewer views taking place, that means the video is lower quality. This is just kind of, I think, instinctive, but I want to question that. I also want to question these YouTubers who essentially put the blame on themselves and say, you know, they’re just not making as good videos anymore to me I think there’s something else at play, and it is the system. So, from time to time, I like to go over to Google Trends, where you can basically search any search term and see how much traffic there is, how many people are searching that term at any given time, and if you search for “Veritasium” basically since 2004 you see, well, this kind of curve, which might make you ask the question: “Where on the YouTuber life cycle do you think I am?”,
“Where on the YouTuber life cycle do you think I am?”, and of course this little spike here is from my black hole videos, this over here must be some Harry Potter fans who are bad at spelling. So this is how much people are searching for me, which I think kind of correlates to how much my videos are being shown on YouTube, and how much people are watching them and enjoying them. Now let’s look at a similar channel for comparison. How about the channel Numberphile? It’s also an educational channel, but it’s, uh, you know, very diff- made by very different people with very different schedules, very different topics that we’ve tackled, and yet these graphs look pretty similar, Uhh, let me have a look at one more. So this is the AsapSCIENCE curve. So here you have Veritasium, Numberphile, and AsapSCIENCE: three educational YouTube channels, which all follow a fairly similar pattern on Google Trends. So my question is: You know, why is this? Why should we follow a similar trend? Well, the obvious answer is the algorithm. And the algorithm has a very important job, because in traditional media, there are not that many works produced. I mean, in American theaters last year, there were about 800 films. 800 films. That’s it. On TV there were 500 scripted TV shows. But when it comes to YouTube, it doesn’t even compare. I mean, I don’t even think that they advertise the number of hundreds of hours uploaded every minute. The obvious analogy is that you’re really just seeing the tip of the iceberg, but in the case of icebergs, you see, you know, almost 10% of the ice above the surface. In the case of YouTube you’re seeing just (the tip of)x5 of the iceberg basically, statistically speaking, you are ignorant of everything on YouTube. And yet, YouTube claims that they have this algorithm, which you can think of, like, the brain of the platform that can connect an audience (that knows almost nothing of what’s on the platform), to the particular types of content they like. And in the best case, they would argue, YouTube would argue, that the algorithm is basically the audience. You know, the audience decides what they want to watch and-and the algorithm is just reflecting that, but of course, we are not in this ideal world, and the audience itself is always kind of shifting and changing who they are and what they like to watch, and so, the algorithm, you can think of as following the audience the algorithm is-is chasing this audience around and trying to reach them. Now, what YouTube doesn’t much consider, is that well, the YouTube creators use their content to try to chase the algorithm. So these YouTubers who would love to be creating content about, you know, whatever they’re most passionate about in whatever way they also see what’s popular on the site and they try to make content that will please the algorithm. And so there’s a way in which the content is chasing the algorithm, the algorithm is chasing the audience, and if the content actually manages to catch up with the algorithm, you get a-a kind of perverse situation, in which the algorithm is the content. I mean, what do I mean by that? Just a crazy thought experiment: if YouTube wanted more videos of snails, they could make that happen; they don’t actually have to go out and make any snail videos, if they just, you know, promoted videos with snail in the title, tomorrow there’s gonna be a whole bunch on the front page and trending and then creators be like, “Oh snails are the new thing!”, and they’re going to go make a bunch of videos about snails. YouTube would love to believe that-that creators don’t care about the algorithm – but they do! It’s like the core aspect of how creators decide what content to make. Just as a more, uh, down-to-earth example, I can show you a graph of the length of my videos from 2011, through to today.
I can show you a graph of the length of my videos from 2011, through to today. And what you see is, at the beginning, uh, all of my videos averaged about 2 or 3 minutes in length, but these days, I’m fast approaching an average length of video of 10 minutes. Now, YouTube never came to me and said, “Derek, you have to make longer videos”; all they said was, well videos that are longer and get longer watch time will be promoted more on the site, so, people like myself and, you know, all the other creators who wanted to be seen, we made longer videos. It’s just, what we did. Now, we know that the algorithm is always changing; they’re always trying to optimize it make it work better, but they also change what it’s trying to optimize for. In the beginning it was simple: just views, but they quickly figured out that, of course, like, a 20 minute view is worth a lot more than a 5 second view, so the actual metric should be watch time. And this, uh, was advantageous to people like gamers, because people like to watch people play video games for long periods of time. They also reduced the latency of the algorithm, by which I mean it updates every 10 minutes instead of, say, every day, and that’s beneficial for news content and things that are really time-sensitive, and to me as a YouTuber what this means is it’s impossible to be an expert on YouTube. If you look at what it takes to become an expert at anything, according to the Nobel prize-winning economist Daniel Kahneman, you need, of course, many, many hours of practice. Effortful, deliberate practice. And on YouTube I would argue we get that. You also need timely feedback, and, again, on YouTube we’re given plenty of tools to be able to see how our videos are performing; but, when it comes to a reliable environment – this is an environment in which the rules of the game aren’t changing: so you can think of something like chess it’s possible to be a chess expert, because in the middle of a game of chess, the rules don’t suddenly change on you. Now, on YouTube, that is not true. We are not playing by the same rules the whole time, because the algorithm is constantly changing. And so, that puts us in this position, where it’s kind of impossible to be experts, and instead we’re a little bit more like [page flips], caged pigeons. Okay, go with me on this. Uh, there was this famous experiment where, you know, pigeons were put in a cage with a lever and if they push the lever then they would get some food, and so they very quickly learned that they needed to push the lever and the food would come out, so that’s fine. But in a another part of the experiment they disconnected, the lever from the food output and they just supplied food into the cage, kind of, at random and what did the pigeons do? Well, they started to do the types of behaviors that they were doing when the food appeared, so, if one was preening itself then it would keep preening itself, with the expectation that food was coming, and this one, maybe he was pecking the ground; so, the point being they started to engage in these superstitious behaviors, um, because they had related some sort of cause and effect, but there actually wasn’t a cause and effect there, and to me, this partly explains the whole YouTube burnout phenomenon, because you have these creators who are stretching themselves to try to chase that algorithm meanwhile the algorithm is constantly shifting and so you never feel like an expert and never quite know what’s going on even though you’re constantly trying to relate cause and effect now I can understand why the algorithm has to change. I mean in the early days of YouTube, it was this situation where if you had subscribers you were golden because the next time you uploaded some content then YouTube would show it to all your subscribers it’ll get a ton of views
downloading game may take longer then YouTube would show it to all your subscribers it’ll get a ton of views
then YouTube would show it to all your subscribers it’ll get a ton of views YouTube would say “oh this is a popular video”, show it to even more people, getting you even more subscribers so the next time you launch another video then it goes to more subscribers more views and it was just this positive feedback loop where you would grow and grow and grow a channel so the rich got richer and small channels stayed small so obviously we needed something to break the cycle and so YouTube started doing some experiments where they would essentially change what was recommended to your audience so they would stop showing some of the content from certain subscribers and they saw what happened to views so they started disconnecting these subscribers from the channel No longer was a subscription like “I want to see every video by this person” it was more of a suggestion that maybe something this person creates is something I might like to see and what YouTube found when they did that was that the number of views on the platform the amount of watch time all rose dramatically so for YouTube to get growth of the site they found that reducing the importance of subscriptions was essential and this kind of makes sense if you think about it I mean if you make one great video, that doesn’t necessarily mean that every video you’re gonna make is gonna be great so it makes sense for YouTube to step in and play some role in terms of determining what gets shown to which audience members Now, as has been observed by MatPat previously, moving away from subscriptions moves YouTube back to what it was like to sell newspapers on the street you really had to inspire people to buy on the spot you have to serve up really sensational news items. This is also called yellow journalism. it was only once the newspapers built subscription models that newspapers could make sober journalism a reality, something that they focused on because they already had your eyeballs, they didn’t have to fight for that sort of attention they didn’t have to be as sensational but YouTube is going in the opposite direction reducing the importance of subscribers means increasing the importance of sensationalism and click-baity thumbnails and so a few weeks ago I was out at the creator summit in New York and I met MrBeast. I showed him some of the footage from this video I was working on about the shade balls in LA Reservoir and MrBeast took one look at it, and he said to me that’s a banger which I think it’s pretty funny but I think clearly he knew that this was going to be a video that would take off and we talked about titles and thumbnails you might have noticed that the thumbnail borrows a bit from say his work with cereal and Orbeez in backyards and I told him that I was thinking about calling it something like, Throwing Shade Balls and he was like, “No”. He suggested, why are there 96 million black balls on this lake Now I changed it to reservoir because I mean let’s be real that’s what it is but it just goes to show how important the title and thumbnail are what I learned talking to MrBeast was that going viral you can boil down to two metrics: you need to do two things with your video in order for it to go viral and I mean tens of millions of views as his channel shows Okay the first thing, maybe not surprising, is watch time. so when people click on your video they actually have to watch a significant portion of it It’s useful if that’s sort of seven or eight minutes now that is the actual time that they watch, so you need a longer video than that, say 15 minutes, if you want the average watch time to be around that 7 or 8 minute mark Now for me, watch time is not that exciting of a metric to know that I have to hit because I know that most people already watch most of my videos so for me the big insight came with number two if you want a viral video you need to have a high click-through rate so that is the total number of clicks on your titles and thumbnails
so that is the total number of clicks on your titles and thumbnails divided by the number of times that title and thumbnail have been shown that is your click-through rate and MrBeast showed me this graph showing that as you approach 10% 20% 30% click-through rate then the number of views and the number of impressions that video will get Just skyrockets. It jumps dramatically now on the one hand that kind of makes sense because you know people are clicking a video obviously you’re going to get more views but it is such a dramatic increase that it really turns the site into a place where the title and thumbnail are everything you can have a great video but unless you have a great hook to get people in it’s not going to go viral and so it was amazing to hear all these different YouTubers talk about how they optimized a thumbnail many people told me that when they were working on a thumbnail for a new video they would actually Photoshop it onto a screenshot of their current YouTube home screen in various positions and just see how eye catching it was they’re really working hard to make these thumbnails as clickable as possible it’s like weaponizing the thumbnail and that arms race is only set to increase with the introduction of real-time click-through rate which is coming out in a month or two so what you can bet will happen is that creators will launch a video and then they’ll be sitting there with all these different variants of thumbnails and they’ll be swapping them out and looking at what that does to click-through rate and then going with the one that leads to the greatest click-through rate and you can imagine, if you don’t do this, you’re gonna be left behind because all the other creators are sitting there swapping their thumbnails to get the highest engagement that they possibly can on their videos and if you’re not doing that, then your videos will end up getting buried So, acknowledging that this is the reality of YouTube I want to outline to you my plan for what I will do going forward First and foremost, I want to keep making high-quality videos because if I don’t do that I’m not going to be happy I won’t be satisfied and neither will you but second of all I’m gonna have to choose topics that are more clickable because frankly if I’m going to work for days and weeks and months on a video I would like that video to get say 10 million views instead of half a million so a lot of that comes down to “is this topic clickable” and so or number three I’m gonna have to use some clickbaity titles and thumbnails and I apologize for that but it just seems like it’s an existential threat that if you don’t do that your channel will not be around for very long and you’ll be on that sort of downward slide I’d love to know what you think, If you think it’s a good idea a bad idea I’m totally open to hearing you out CaseyNeistat talked about how he’s got to trust his audience, you know to come back to him the truth is, the audience only knows you’re there if they see your titles and thumbnails I went to buy something at B&H a little while ago, and the guy was like “oh hey I love your videos!” and then he said, “are you still making them?” and I was like, “yeah yeah I’m making them. How do you not know?” and I also see comments on my videos where people are like, “oh great here’s a veritasium video see you again in six months” but I’m like, “no I’m making these every week, every other week, like I’m making them you just got to know that they’re here but you can only know that they’re there if they surface to the tip of the tip of the tip of the tip of the iceberg which now is a click-through rate game On the one hand, you could be a little bit disappointed by the way the game is working right now but on the other hand I challenge you to think of a better way because they have this limited real estate in which to show people potential videos they might be interested in and of course they want to put the videos there that people are most likely to click and then most likely to watch once they’ve clicked it and like part of me is like, yeah I guess that’s kind of fair would you make the argument that you should put some videos there that people are less likely to click? No. I mean, that doesn’t seem to make any sense. so I kind of feel like this is the inevitable, rational endpoint for any algorithm that is trying to optimize what people are watching we want to give you the things that you want to watch and you indicate your desire by clicking on them but by doing that it also means that things that are just less catchy, less sensational don’t rise into this level here and I wonder more broadly, how does that affect us? How does that affect the YouTube platform? if the YouTube platform becomes a real clickbait site Who wants to go there? What types of audience will come there? Will there even be an audience for educational content? I think it’s all these decisions about the system and the algorithm that determine what the actual media looks like what content is created what people come to the site what audiences are served there And I think this is a broader issue than just YouTube. I mean, everywhere now we can be served the things that we you really want. but what does that world look like? I mean you might have thought it would be this utopia but I would argue that it’s not. because it silences all these voices that you would otherwise get you know in a more curated marketplace if you’re just going for the things that drive the clicks That is a very skewed view on the world and I’m not sure it’s for the best But there is one way on YouTube at least to short-circuit this effect and it is to ring that bell and you’re like, “oh the whole point of this video is to get to this point and tell you to ring the bell!” I mean, no, if you want to ring the bell, I encourage you to do that now because I feel like I never encouraged people to subscribe or ring the bell on this channel because it seems needy and I think you can do whatever you want, whenever you want it but in this ecosystem where click-through rate is king, where millions of black balls on a reservoir is the thing that is going to rise to the surface, the only way to get around that is to have people who are notified every time I make something so I want you to think about the videos that I’ve made you know in the past and so far this year and think about do you want to know about every single one of those? and if you do then I encourage you to click the bell and the more people who do that, the less I will be driven to make clickbaity titles and thumbnails and videos because I’ll know that I can reach my audience without having to appeal to, you know, the basest instincts of people So, if you want to see more of what I’m doing, click to subscribe, ring the bell and if you don’t then please you know unsubscribe because it’s good to give YouTube as clear of signals as possible but I think there is a bit of hope for the future because what I’ve described is the way YouTube appears to be working right now but I think that, well we know that the algorithm is always shifting and what I’ve heard is that the direction YouTube wants to go is optimizing for long term, satisfied, watch time That’s got three pieces. You have to look at, are people coming back time after time If they watch someone’s videos do they come back time after time for months That would be one signal. Are they watching for a long period of time? That’s also a sign that people are enjoying it And finally, there’s the satisfied piece. Are people satisfied by the things that they’ve seen? This is a new metric to add in And how do you really gauge satisfaction? Well, YouTube is experimenting with surveys. Surveys just like this one. and they will hopefully allow us to come to a different place where the algorithm isn’t just about click-through rate and watch time but it’s also about satisfaction. How much you’re driven to come back How much you really love the content. So my great hope is if we move in this direction there may be a time in the future where click-through rate is no longer king I’d love to know what you think so let me know down in the comments and I’m gonna jump straight into end screen here I’m going to post a few thumbnails for videos that are not particularly a click-baity so feel free to click these. I mean you’re gonna look at them and be like, “I don’t want to click that” but that’s exactly the point. these are apparently good videos that maybe just aren’t as catchy in terms of titles and thumbnails If you want to subscribe you can click up here and thanks for listening to the rant that I’ve given many times to whoever will listen