A Complete Guide to Cracking the YouTube Algorithm with Matt Gielen: TubeTalk 167

Do you want a heads-up on cracking the YouTube algorithm going into 2020? This week's TubeTalk guest Matt Gielen brings you all the latest unmissable insights.

We all seem to have a love-hate relationship with YouTube. We produce a video, and it does well, we love it. We produce a video, and it does poorly, then all of a sudden we blame YouTube and its algorithm. But is their algorithm really at fault here? Or perhaps it's our content? This week's episode of TubeTalk, the YouTubers podcast, will fundamentally change the way that you upload content to your YouTube channel.

Joining us today is Matt Gielen, who is a YouTube strategist, and a consultant to Netflix, Nickelodeon, MTV, and Comedy Central. You might have heard of some of these names. Matt is actually a code cracker, not in a hacker kind of way, but unpacking the YouTube algorithm seems to be his specialty.

In this podcast you will learn:

  • Why YouTube chooses to surface your video to non-subscribers
  • Why your Click Through Rate is so important
  • How user Watch Time history plays a huge role in the content they are served
  • How to pivot successfully so your new content works on YouTube
  • Why linking to your other videos in YouTube End Cards and descriptions is vital
  • Why viewer activity & engagement far outweighs a subscription to your channel

Understanding the YouTube Algorithm: Full Transcript

Liron Segev: So, Matt, one of your specialties is clearly understanding what this YouTube beast wants from us. Have you seen major changes in the algorithm and the way YouTube handles content?

Matt Gielen: Yes and no. So, basically, the algorithm, in my view, was very, very stable for about a solid three and a half, four years. Once they made the switch to watch time, there were a bunch of little things that caused big waves for certain areas of YouTube, but it kind of got to a very stable point and stayed very stable for two or three years. Then starting in late August, September of 2018, things just started going crazy. Things were a bit stable between, I want to say, October and November, about eight weeks. Then right about December 15th, everything went crazy again, and then, again, on January 15th and then February 15th or 18th, somewhere in that range.

Basically, every six to eight weeks we're seeing these massive changes in what is or what isn't being promoted. Some of that is algorithmic, and some of that is structural and policy. By structural, I mean making very specific decisions as to what will be put into suggested or browse, which seems to be a layer overriding the algorithm in some capacity. Or from a policy level, where some big news story will come out. YouTube will add something to the policy layer. Everything goes crazy. Then the pendulum swings back after a couple of weeks, once they start getting their feet underneath them again in response to whatever the issue is that week.

We had the outbreak that caused a huge issue in the kid space. Then we had, most recently, the Steven Crowder, right-wing personality and conspiracy theorist war. That's probably the most fresh. There's been a lot of pretty major hiccups over the last, I guess, nine months now that have caused fairly radical swings.

But I think, even more importantly, we saw the introduction of the DPP optimizer, which is Google's attempt to create, quote, unquote, diversity in the feed, which we just put out a video on our YouTube channel, ‘Reverse Engineering the YouTube Algorithm: Part Three’, because it was a massive structural change to the algorithm. We released ‘Reverse Engineering the YouTube Algorithm: Part Two’, I want to say, in 2017, 2018, which was a Google whitepaper that explained how they created the algorithm. It basically said there's two main big filters, candidate generation and ranking. Well, the DPP optimizer is a third massive filter. What the optimizer essentially does is, if you have a feed, like in browse for example or the Facebook feed as an example, typically, the way they sort the items in that feed is from best to worst. User A is likely going to be interested in these 100 videos. We've ranked them from 1 to 100.

Well, for that user and for any user, the optimal actual layout of that feed might not be to go from best to worst. It might be like put the best one at spot seven. Put the worst one at spot 24. That's basically optimization by group as opposed to optimization by individual video. It's really fascinating if you read the paper, or you can watch our video to see an overview of it. YouTube has come out and said we want to provide more diverse opinions, or we want to provide a wider set of topics, and it's really not that. Basically, it's we want to hack your brain to make you stay on the website longer, and what we found was, by pulling more videos from your history and leaning a little bit heavier into collaborative filtering, we found that we can make you stay on our website longer. It's basically like, hey, we made a better heroin. It has nothing to do with exposing people to different opinions or exposing more creators. It's 100% solely based on the fact that we can get you to stay on our website longer if we pull more videos from your history.

LS: The longer you stay, the more ads they can serve. The better the business grows, all of that good stuff. YouTube's still a business at the end of the day. The interest isn't to shove you off the platform

MG: Exactly.

LS: That makes sense. But now does that mean that YouTube looks at individual videos compared to a channel as a whole?

MG: Oh, yeah, I don't think they look at channels as a whole at all, actually, other than to make recommendations to audiences based on which videos to serve them next. So in part two and part three, they go into a lot depth, but basically saying, to program our website, we program to the individual. So each individual has a distinct watch history, and we look at their watch history in order to make recommendations to them for what they are going to want to watch in their next session.

So, for example, if you're a big fan of Rosanna Pansino and you have all of her videos in your watch history. What YouTube will do is, next time you come to the platform, they'll say, wow, this person always engages with Ro's videos every time we put one in front of them. She is just releasing a video, so let's put this video in front of that person, assuming that that video has done well with other viewers on the platform that also have similar watch histories or similar demographic information. So if you upload a video to your channel and your audience doesn't choose to click on it or, if they do, they don't stay on it for very long, you don't get good average view duration, YouTube will say, well, this video isn't very good when we serve it to people that are really engaged with this channel. Let's not serve it to any more people like that. Let's use that space for something else, whether it's another video from your channel, or a video from someone else's channel.

LS: Got it. What about your next video? So in other words, you served a video. It did really well. Is the likelihood of your next video being looked at more favorably, or is it each video on its own merit, good or bad?

MG: It's each video on its own merit, good or bad with one exception. Let's say you average 1000 views per week on your uploads. If you have a video that comes out and does 2,000 views in that first week, well, now you have doubled the number of people with one of your videos in their recent watch history. That is a strong signal to YouTube that your next video should be served to a wider set of people, because that previous video is in more people's watch history. So that next video might have a better chance of doing well because there are more people with one of your videos in their recent watch history.

So, yes, your videos are going to impact each other. It's a symbiotic relationship, where if you have a video that's doing really well consistently and that's bringing more viewers to your channel, it doesn't matter if that video was posted a week ago or 10 weeks ago or 5 years ago, those people that are watching that video within those last seven days, one months, two months, all have one of your videos in their recent watch history, even if that video is from long ago. So your new videos are going to be put in front of those people, which is one of the main reasons why, if someone shifts their content type and does it too abruptly, they can really struggle to grow that new content type because people aren't there for that new content type. They're there for that old content type, and that leads to eventual death spirals on YouTube, often.

LS: Could we be forcing relationships on YouTube between our videos? Obviously, we put end screens. We put in cards. We put in links in the description, put in a comment with links to our other videos. Does YouTube take all of that into account as well when deciding to suggest videos?

MG: Yeah. They absolutely do. I would say that that should not relieve the content creation or the channel manager of the onus of creating those connections. But, essentially, what you're describing is collaborative filtering or one of the things that goes into collaborative filtering. So you can look at collaborative filtering in a few ways. One of the ways you already kind of mentioned, where if you upload a new video and your viewers really love it and click on it a lot, they will serve it to more and more people that have one of your videos in their watch history, but maybe aren't as active on the platform or it's a little bit older. Then they'll serve it to people beyond that that just have expressed interest in the topic or have similar watch histories.

MG: The other aspect of it is, if a whole bunch of viewers watch video A and then they go on to watch video B, well, people who have watched video A but have not watched video B are more likely to be recommended video B because so many people who watched video A went on to watch video B. That's in suggested or even in browse. If you can create those kind of viewership paths within your content by end screens, playlists, call to action, links in descriptions, pinned comments, links on your channel, etc, then you are more likely to have YouTube promote more of your own content against your own content, but also put your content in front of similar viewers throughout the platform.

LS: To me, this makes complete sense. Shoppers that have bought A would also like to buy B. It's the same kind of Amazon mentality, but on videos, where they like this content, they're more likely to like that content - let me help them along with that. Now you've mentioned a lot about views, but you didn't mention subscribers. Are we making too much of a thing about subscribers because subscribers have changed over the time? Is it all about the views? Or are subscribers still a thing?

MG: I'm not sure, but I don't think it matters. So, generally speaking, I don't know if you watched my response video to Derrick from Veritasium's video, My Video Went Viral. Here's Why. Essentially, back when Google first released the One Channel, it was wrapped up with Google+, and it was a whole initiative that fell flat on its face and was absolutely awful. I mean, look, I love YouTube. YouTube has allowed me to be unbelievably successful and start my own business, and I will jump to their defense whenever I can, but sometimes we've got to call them out because sometimes they just make terrible decisions. It's not always their fault.

So during that time period, what they did was anyone who created an account was basically auto-subscribed to dozens and dozens of channels across many different verticals that had nothing to do with any interest they'd ever expressed. So you had these channels that went from a million subscribers, to several million subscribers, and most of their subscribers had never seen a single video from them ever before, had no idea how they were, didn't click on their stuff. A lot of them were like I can't get any viewership because you're putting my videos in front of people that have no interest in them because you auto-subscribed them.

YouTube was like, oh, well let's just make subscriptions unimportant. So it was like, wait, what? They're like, yeah, we screwed this thing up, and so we'll just keep screwing it up. Unfortunately, they can't undo that. So they were like, okay, now we're going to have people subscribe, and then they're also going to ring a bell. Then they're going to have options for that bell. Then even if they select some options, we still might not send stuff to them.

I've worked with channels across the board, big, small channels, everything in-between. Almost none of them get more than a handful of percentage points of their viewership from notifications. I can tell you on the Little Monster channel, we're brand new, so all of our subscribers are pretty fresh and pretty active. For us, I think it's like 3% is from notifications, which is, I think, the highest I've ever seen it. Over the last 28 days, it's our seventh biggest traffic source. It's responsible for 3.7% of our views, and that's 300 total views, and we've released 5 videos. It's just like, okay, I can't focus on notifications. I don't really have any control over that. Telling people to ring a bell is way less valuable than telling them to go on and watch more videos, so I ignore that.

Then if you look at subscriber views versus non-subscriber views, that can vary wildly. I've seen channels doing hundreds of millions of monthly views that have 3% from their subscribers and some that have 20% from their subscribers. But if you look at the traffic sources on that, it's all browse and suggested. So whether someone has subscribed to your channel or not is such a low impact signal from what I've seen, in terms of what gets suggested to them, that, really, exerting any sort of effort there is not nearly as good as exerting effort towards understanding your audience and your viewership, what they're interested in, and thinking about your programming towards that general audience, because YouTube cares far, far, far more about how someone has interacted with your content the last time it was put in front of them or the last several times it was put in front of them than they care at all about whether or not someone subscribed.

LS: So at the end of the day, we're once again circling all the way around back to the viewer. It's all about them, their experience, and what they would want to most probably see next, and then engage with it. Speaking of engagement, do you find, in your experience, that things like give the video a thumbs up and leave a comment ... Does that also add to good signals as well as the watch time and when somebody rewinds your video and watches it again?

MG: They are all taken into account, and YouTube said that they were looking at total time spent on site, including time commenting. It's like, yeah, they've been doing that for five years. It's called watch time.

The more engaged your audience is, the better. If someone's just passively viewing and it's kind of like they can watch or they could pass on it, well, you're kind of in a bad situation. You want people hyper engaged, if you can. Some of the ways to do that is to get them to like the video and to comment on it.

One thing they did say is we're taking things like what people like and what they dislike into account. It's like, well, are they saying like and dislike from the engagement or like and dislike based on the implicit signals that they're sending? Because if you look at Reverse Engineering: Part Three, I'm not sure if I cited it in the video, but in the actual paper they say that explicit signals are actually not very good from viewers, with the exception of surveys. But they're not surveying you after every single video you watch. They're surveying you on a session basis to say how was your last session. Are you satisfied with your time on YouTube? Because YouTube doesn't care about the individual video. They don't care about the individual creator. They care about the individual viewer. They want to know if the viewer's experience was good.

You can't be confident in the accuracy of their answer, and so they look for the signals that you don't consciously think you're sending by clicking on that one more video or clicking on that one thumbnail with the scantily clad girl. When you see that it's just some dude playing a video game, you're like, oh, I'm out of here. It's a guilty pleasure thing that you're not going to admit to, but YouTube sees those signals. They know exactly what you're doing. So they make those suggestions based, I think, far more on the implicit feedback than on the explicit of like, dislike, comment, survey kind of thing. But the one thing I haven't said yet, which is by far the most important thing, is click-through rate.

Click-through rate is the number of impressions divided by the number of people that click on it, It’s king, by far, 100%, and will be king until YouTube fundamentally changes the way they promote videos to people. By fundamentally, I mean, right now the way they promote videos to people, essentially, is by showing them a thumbnail and title. Now with the mobile feed going to a preview and now basically just playing videos in that mobile feed if you sit on it long enough, that's going to change things up a little bit. But right now it's click-through rate, basically.

When we think about click-through rate at Little Monster, what we think about is click-through rate actually starts on the whiteboard. Your programming choice has such a huge impact on your click-through rate. When MrBeast says, okay, we're going to make a video. What are we going to do? They don't go away and give $1,000. They give away $100,000. That might seem like a small programming choice, but it's a huge, huge deal. You're talking about millions of more people that are excited to see a $100,000 giveaway than a $1,000 giveaway.

What you choose to make a video about, in large part, is going to determine your title thumbnail. If you talk to a lot of YouTubers, a lot of YouTubers go, oh, yeah, I think what's going to make a great title and thumbnail, and then I make a video about that that would deliver on the value proposition of the programming choice as reflected in my title and thumbnail so that people stay and watch it. Because click-through rate doesn't really mean all that much if people click away from your video and you don't get the average view duration.

So, again, there's another one of these symbiotic relationships where it's like you can't just focus on one thing. You've got to be an amazing programmer, making a great video and choosing to make a great video. You've got to be great at production so that your content is high quality and ad friendly and all these other things. Then you also have to be an amazing marketer in that title and thumbnail.

LS: At the end of the day, it's the people that really understand not to go ready, fire, aim. You can't take out your camera, shoot for nine hours, and then go, oh, I wonder what I'm going to do with this footage. If you're very deliberate and very purposeful, you'll win so much more, especially when you're using the various tools, when you're using YouTube's own autocomplete. You're going to get all those hints of stuff that people actually care about, make your content around that. That is immediately a leg up to the previous scenario, where you ready, fire, aim.

MG: I would caution on the autocomplete. I think autocomplete is a great starting point, but you want to run all of those autocompletes through Google Trends, and make sure you're shifted to YouTube search, just to confirm, because some of those autocompletes are off topic or from too long ago.

LS: Relevancy for today is critical. The typical example just to prove it is when you go and if you do a search for, I don't know, Mother's Day now. That's come and gone. The interest of this is flat. But if you just did a search for Mother's Day, you're going to be seeing a million videos. So the two are very, very different. It's all about relevancy and time, as opposed to, oh, it's got six million views. It must be a good video. I'll make one just like it. The iPhone 4, when it came out, and the Galaxy Note 1, when it came out, I'm sure had millions of views. But how relevant is it today? Matt, do you think we should be focusing on writing titles and descriptions for humans? Or should we be doing writing titles and description more keyword stuffing for machines?

MG: Oh, I mean, obviously, you want to do it for humans. When we think about optimization, what we think is, essentially, what is more likely to get a viewer to click on this, rather than what is going to make a machine understand this, because click-through rate is so much more important than anything else. YouTube will figure out what your video is about. It will do that by listening to and watching, now, your video. I don't know if you had a chance to play around with the Cloud Vision.

YouTube are just going to get better and better and better at understanding your content, and they'll be far better than humans, because humans get tunnel vision. We don't pay attention to stuff going on, on the edge of the frame if the central focus is the middle of the frame, but a computer can see it all. Then they look at things like comments and how does your metadata match up to what they're seeing and is that accurate and have you been historically accurate.

But all of that is basically they look at comments, because in determining the language of a video, it's actually far more accurate to look at the language used in the comments than it is to rely on what the creator put in as their language, especially for older videos.

LS: Wow. Hold on, whoa, whoa, whoa. The obvious question is, what happens if you're new and you've got no comments?

MG: Well, one it's very important that you actually put the language of your video into the video metadata, in part because, for eight or nine years, that wasn't even an option. So there's eight or nine years of videos that probably just have no language set to them, other than what YouTube has gone through and determined, if it's made it to that video yet. It's relatively new, and a big part of how they make recommendations is on contextualization. So contextualization has everything to do with where is the viewer watching. What device are they watching it on? Are they watching it on Wi-Fi? Are they watching it on their cell phone or via a telephone etc. The age of the viewer, the race, other demographic information. They're adding these contextualization layers at multiple points in their recommendation. So what that's doing is, as it pertains to the previous comment about language, if the person generally watches content in the English language, they're going to be 1,000 times more likely to just serve them content in the English language. If someone always watches with closed captions on, they are 1,000 times more likely only to serve them content with closed caption. That's the level of granularity they're getting to, which it's mind-boggling

To actually answer your question, if you have no comments, then it's even more important that you're uploading the transcript. But if you have no comments, you don't have much of an audience, most likely, and so I would be far less concerned about YouTube understanding what your video is about and far more concerned with how do I get in front of people that I think will like this. An old colleague of mind at my previous job had a saying. If your mom and dad or your friends and family don't want to watch your content, it's probably the content. So it's like that's where you start.

When I started the Little Monster channel, I had a small following on LinkedIn. 5,000 followers on LinkedIn is like 5,000 followers on Twitter. It's nothing, but I started posting our YouTube videos and would post teasers for the YouTube video to my LinkedIn audience, and got a little bit of traction of those people coming over to YouTube and watching and watching with their YouTube accounts. So from there, YouTube would serve it to more and more people. Then with a few of the connections I'd made on YouTube over the years, other creators on the platform, they would like the videos or comment on the videos. That would get a few more people watching and a few more people watching and a few more people watching.

That began to snowball where we started this channel in January of this year, and we haven't been great about it - trying to run an agency and a YouTube channel at the same time is not easy. We've got a couple thousand subs now. When we put up a video, it's going to get 1,000 views or so within the first week. That's not bad for a channel that was started in January for a company that basically had no audience.

You mentioned early on these big names that we consult for, and we're extremely well-known in the industry, which is great. But we're not very well-known in the creative community and in the broader YouTube fan community at all. Now we've given presentations at VidCon and that kind of thing, but we haven't been on a platform for years. It's like just no one knows who we are there. We just don't have any reach there. So it's part of why we're doing this, because we put all this love and work into these massive papers, and we put them out on Tubefilter, and they'd get tens of thousands, a couple hundred thousand reads, but I have no idea if that's industry or creators. Who's that helping? So it's like, well, if we're YouTube people, we should probably have a YouTube channel.

To be perfectly honest with you, part of the reason why we just never started a YouTube channel for the first two years is just me being self-conscious about being on camera. Finally, I was like, get over it. Just put yourself on the goddamn camera. No one cares.

LS: The philosophy of doubling down on content that works. That still holds water. So if people are here for a certain type of content, then you give them more of that content, and they're likely to watch it. Give them more of that content, they're likely to watch it. Rinse and repeat. That still holds true. Is that correct?

MG: 100%. I think it's more important than ever. You absolutely 100% have to be focused on a niche, if your goal is to grow. If you just want to make videos that are fun that you feel like doing, great, fine. But if your goal is to grow on YouTube, then, yeah, you have to find what that throughline is, and then you can kind of make whatever you want. So a lot of YouTube channels, the throughline is the personality. Sometimes it's a mutual interest in video games. Sometimes it's a mutual interest in woodworking or skateboarding, whatever it is. The fans are there because they like your personality and you have that shared common interest. But if you look at a lot of creators, oftentimes, it's just that they want to hang out with that creator. Those creators can be free to do any format they want.

MG: But if we take Epic Meal Time as an example, any video Epic Meal Time put up that was not in Epic Meal Time format just wouldn't do as well. Yeah, viewers liked the personality, but they wanted that format. They wanted to see them cook that crazy thing. They wanted that show. So if your audience has really evolved around one particular format, you have to be careful about going too far away from that format from a viewership perspective.

LS: 5-Minute Crafts does 5-minute crafts. If they did something weird, you'll maybe give them one or two opportunities, but most likely if they don't come back to their call, you're going to move on so they have to deliver. Do you see a difference between how-to channels and entertainment channels?

MG: Yeah, great question. So we refer to them as utility channels. So you have your entertainment channels where people are there to be entertained, and that can take on millions of different forms. A utility channel is a channel that people come to for information and then leave. Now the then leave part is not the desired intent of a utility channel. It's just what happens. The reason why is that utility is an exchange. I'm exchanging my time in watching this video and potentially a tiny fraction of a penny or whatever it is in exchange for the information you are giving me. What ends up happening is there's no reason for someone to watch your next video, often, because they've gotten that piece of information.

Now things that you can do to mitigate that is make your utility channel an entertainment channel. There's three main ways to do that. So first is the MKBHD way, where you make your videos so incredibly stylish that they become works of art, so that I don't care if he's talking about an iPhone or a TV or home lighting, I'm going to watch it because they're beautiful. The next way is from a personality perspective. This is more of a HouseholdHacker approach. I think it's HouseholdHacker, where people just love his personality, that they don't care if he's sitting in a gray room. They're just like, yeah, this dude's awesome. I'm going to watch anything he puts up.

Then the third way is to have a format that is so interesting that people don't care what product you're talking about. They're watching for the format or a combination of all three of those. But if you're just topic ... You're like, oh, I've just got to get this iPhone the moment it comes out because it matters if I put my video up before someone else, well, you're never going to win that race. You're always going to be chasing YouTube search. You're never really going to grow. Most of your videos are going to get ... If you have 100,000 subs, most of your videos are going to get a couple thousand views if you're lucky. You're just going to be hoping that you get that one video that breaks or that you're just big enough to get a brand deal here or there or whatever it is, if you don't have one of those things that's driving everything. Generally speaking, we very, very much discourage a utility channel that's not pursuing being an entertainment channel.

LS: So a lot of us just had a mini heart attack as we listen to this. How do we convert them to become more entertainment?

MG: Well, to give myself a plug, I wrote an entire guide as to how to do that.

LS: Funny you should mention that, because this is the exact time of the show where I normally say to you, so, Matt, what have you got going on? Where should we find you?

MG: Well, you can certainly check out our YouTube channel, and on Twitter. You can also check out all our work on Tubefilter.. What I was going to mention about utility channels, the resource for them is The Taxonomy of YouTube Videos. It's a free paper. I put it out on Tubefilter, or you can watch our series on it on our YouTube channel. But if you read that, that's basically a guide to developing fresh and unique content on YouTube by understanding what's popular on YouTube now and breaking it down into easily usable items.

Cracking the YouTube Algorithm 2020

Then the big news is Cracking YouTube 2020. It is going to be looking at 200,000 YouTube videos, 20,000 YouTube channels, and seeing what we can draw from all of those channels of data. So I'm very excited about that. One thing I can spoil is that Cracking YouTube in 2017, what we saw was that videos that were between 7 and 16 minutes, videos that got 6 to 8 minutes of average view duration, had much greater performance than videos that were shorter or longer.

Now in Cracking YouTube 2020, which is the paper that's coming out, we've done a fair amount of the research so far. Now we've got a long ways to go, but what we're seeing, one, is that there's less dependency on generating high average view durations on average, but there's something that's very interesting that happens, where we're seeing videos that are a couple of minutes in length, they tend to perform not quite as well as other videos. But at four minutes, we're starting to see pretty decent first-day viewership, at least on par with five and six minutes. That starts to go down a little bit at seven, eight, nine minutes. But a really interesting thing happens when a video goes from 9 minutes and 59 seconds to 10 minutes, which is that viewership jumps by somewhere between 20 and 40% on average.

I was very, very skeptical when I saw the data. I went through, quadruple checked my work, but I went and I looked at 2017, and the same exact thing happened in 2017. There's a jump. I thought it was just an issue with the data when I presented it, and I didn't even bring it up, because I was like, this has got to be something funky here. I never had a chance to go and circle back. But when I saw the same phenomenon happen in 2020, I said, okay, what is going on? I went and I looked at it. This is, at the very least, true for first-day viewership.

Now this could be something algorithmic. Something with that kind of jump feels algorithmic to me, but, hear it me out, it could also just be viewership, just viewers, where a 9 minute and 57 second video just feels long. You're like, oh, God, 9 minutes and 57 seconds, God. But a 10 minute and 2 second video, that doesn't seem so long. I'm wondering if it's like, if you want to watch a short video, 9 minutes is long. But if you want to watch a long video, 10 minutes is short. I don't know. We have a lot more data research to do, but that was the nugget I was talking about earlier where it's like, okay, this is not a coincidence anymore. This is a pattern, and we need to look further into this. We're going to go into that. We're going to go into a ton of depth on frequency of posts, a lot on average view duration. We've got a lot of data around both suggested and browse.

Huge thanks to Matt for his insights. You can follow him on YouTube and on Twitter.

Subscribe to TubeTalk: The Video Marketing Podcast

TubeTalk has been informing video creators amd marketers since 2014, and is one of the most respected and listened to podcasts in the online video industry. Subscribe today so you never miss an episode! You can find vidIQ's TubeTalk at:

Don't forget to leave us a review if you love the show!

Want To Get More Views on YouTube?

If you want to take your YouTube channel to the next level and get more views on YouTube then make sure to download vidIQ. It will help you research YouTube, analyze videos, audit your own channel, and take actionable steps click here to install now!