Why on Earth would you NOT optimize a video? The whole idea sounds preposterous if you’re trying to build a following or a brand on YouTube, but it turns out sometimes that’s the smart move.
In today’s episode of Tube Talk, Jeremy Vest speaks with Dane Golden of Hey.com, where they talk about whether Google focuses more on channels than individual videos, the times when it’s probably best to let a video die on the vine, and why when it comes to sweeping changes on your channel, a little at a time is better than a big massive change all at once.
Enjoy, and please share if you find this episode valuable!
It’s time for Vidpow Tube Talk, everyone’s favorite podcast for video marketing tips, brought to you by Vidpow, your video marketing team. Get your free YouTube channel audit by going to vidpow.com/vidchop.
Jeremy Vest: Hey everybody, this is Jeremy Vest from Vidpow.com. Today, the amazing Dane Golden, how is it going?
Dane Golden: Hey, what’s up Jeremy.
JV: Dane, you told me what topic we were going to be talking about today. What are we talking about today?
DG: Well, we’re talking about you Jeremy.
JV: That should be every day.
DG: Okay, so I emailed you with my suggestion of the topic today. Here’s the thing, most people, they’ve heard us on the podcast and they don’t necessarily go to all the conferences we go to, Vidpow, Vidsummit, etc., but you’ve seen me right? I know you, I’ve talked to you a lot. I am always in the first or second row anytime you speak, anytime any of our friends speak. I want to get every kernel. Am I wrong about this? Do you see me in the front asking questions?
JV: Absolutely, I appreciate it. It’s like my mom, then you are next to her.
DG: Except I’m using this for business. I really do take to heart everything you say and I’ve been particularly interested in your idea of saying there are times, even though we optimize videos, back catalogs and future videos for a living, there are times you should not optimize a video under any circumstances. I’ve never heard anyone else say this, but importantly, the first question I want to ask you about is titles. When should you not change the title of an old YouTube video that has been on a channel for awhile?
JV: To start this conversation and thank you for the kind words Dane. Likewise, I’m your biggest fan. To start this conversation off, Vidpow, we have over a billion organic views for our customers. Most of the time, because I’m kind of known for optimization techniques on YouTube, customers come to us with the idea that they want to re-optimize or optimize their older content. As they’re making new content, you know, we will optimize that as well. I have about four or five years of optimizing old content. A lot of the things I’m going to be explaining to you guys is from a lot of experience.
One thing we found out, real quickly, about three or four years ago, is if you have a video that has a lot of search and organic viability. People are finding it through suggested videos, they’re finding it through search and you change the title, you can change anything else, you can change the tags, the description, you can change what playlist it’s in, you can change the thumbnail, but if you change the title, it seems to reset completely the ranking of that video.
DG: Not in a good way necessarily?
JV: Not in a good way at all. For example, the biggest blunder I ever did was with a video with 32,000,000 organic views. We changed the title-
DG: Why not make it better!
JV: Yeah, we can always make it better. 32,000,000 views is for suckers, no. Immediately, it lost 90% of it’s views on a daily basis for six months. It did end up coming back up in the rankings but it never landed where it was before. If you think of this just from a common sense perspective, YouTube is serving videos based on recommendations, how long people watch stuff, how much they engage with it and hundreds of other things we have no idea about. They are generating people to see your videos based on ranking algorithms. If you all of a sudden change what your video is titled, they seem to reshuffle where your video ranks altogether.
DG: Based on only new data essentially.
JV: Exactly, or they’re just like, “Hey, what do we do with this thing now? It’s something else, it’s not the same.” Like I said, we’ve definitely done tons of multi variant tests on this and-
DG: I’m sorry to interrupt you. Now, you get paid for optimizing back catalogs of videos right?
DG: I get paid for doing the same thing. You’re saying that even if I change one character in the title of a video that’s doing super awesome over a period of time, it can really harm the whole video permanently?
JV: I’m saying that it’s 70% to 80%. From within 48 hours, you’re going to lose 70% to 80% of your daily activity and views on that video.
DG: That’s incredible.
JV: You could literally just put a dot, a period or a comma. It doesn’t matter what the change is, it can be very minute. It doesn’t matter.
DG: Basically what you’re saying is, “I want to go out of business.” It’s the I want to go out business change.
JV: A couple of other weird things we’ve seen-
DG: I’m sorry, I’m sorry.
JV: Go ahead.
DG: Even if you instantly changed it back?
JV: I don’t know about that because I haven’t been brave enough to try it on a big video.
DG: If you changed it back like with that other video, did you change it back to the original?
JV: Yeah, we didn’t do it instantly but it still didn’t matter. The damage had already been done. Six months later the video bounced back up but it never got to where it was.
DG: It took six months.
JV: Yeah. That time changes a lot. I’ve seen one month, I’ve seen four months. A couple of other interesting things, one of our customers unlisted like 100 videos last year. The whole channel, the views for the whole channel died by about 70%. We put them back live and the channel immediately came back up.
DG: I’m sorry to interrupt you again.
DG: What about if the videos are out of date, no one wants them anymore, they have wrong information, even so?
JV: What we’re finding is you can’t change a huge chunk of your library at one time. Let’s say you wanted to un-optimize and not have older videos appear, you would want to definitely change the titles and things, but only do a few a day. You don’t want to go change 100 videos. Why would you want to un-optimize something? The biggest reason is if people aren’t watching much of the video, they’re not engaging with the video or it’s old content, you probably don’t actually want people to find that video.
DG: You’re saying you don’t want to take it down, but in this case the act of actually changing the title to something less helpful is a good thing?
DG: That’s so ironic.
JV: The reason you don’t want to take it down is because when you unlist, make private or delete a video on YouTube, the social engagement, likes, comments, unlikes, shares and views disappear from your channel. Most of the time you want to keep that content up. The trick would be to un-optimize it so it’s not showing up organically, it doesn’t have a great thumbnail, it’s not in playlists, it doesn’t relate in suggested videos because it doesn’t have the same meta data as the other ones. You put cards and on-screen annotations to the other videos that are newer or have replaced that one.
DG: What if it’s a bad video and it’s doing bad. Still take it off? Don’t take it off?
JV: I think it’s relative to how many views, likes, comments and engagement the video has. If it doesn’t have any views and it doesn’t have that social engagement, moving it, deleting it, unlisting it, it won’t matter. Never remove a big chunk of your library at one time. For some reason, we’ve seen some very extreme cases of the channel dipping and doing weird stuff and doing bad. The biggest thing I want to tell everyone is whatever you’re doing on your channel, don’t make huge changes to a lot of the library in one day or one week. Spread it out.
DG: By one video a day over a period of time?
JV: If you have 300 videos, then three or five videos a day is probably okay. If you have 500 videos and you change 100 in a week, that’s what, 20% change in your library. It’s my experience that weird and negative stuff happens when you make a huge, huge jump in change in your library very quickly.
DG: That’s super important to know for people like us who are listening. I wanted to ask you another question. That is, you’ve often said if a video sucks, don’t optimize it. If it’s a bad video, don’t optimize it. Isn’t that true?
JV: Absolutely, absolutely.
DG: Why? Why wouldn’t you want more people to see it?
JV: Basically, the way the algorithm starts as far as we know, is all about watch time. Views, average view duration, session starts, session duration, session ends, upload frequency, there’s a lot of parts of the algorithm on how people watch and consume your videos and other videos. If that’s the baseline core of how the algorithm works, then if people aren’t watching and consuming much of your video or leaving YouTube altogether because your video is so bad, it’s literally like putting a wrench in the sprocket. It’s going to completely mess up your chances to have a great channel.
DG: That’s a good point. If you optimize, then it bubbles to the top and someone says, “Oh, here’s a new video or it’s new to me,” then they watch it, they watch it for about 10 seconds and they go, “This is a terrible video.” That doesn’t just reflect bad on that video, it reflects badly on your whole channel.
JV: Well, a lot of people argue with me on that. A lot of people say a video is separate than your other videos on your channel. I have a lot of multi variant tests that prove that different for me. It’s my humble opinion that if you have a YouTube channel and 45% of your content is bad and 55% of your content is good, there’s some type of correlation there.
DG: As I understand it, from your research and Matt Geelan’s research and other folks research, what YouTube says is that, YouTube really promotes channels and not videos. The video is only evidence of a channel being promoted. I would make for the audience, I would make this sort of example. If you go to a restaurant and you have a bad meal, you will say that restaurant is not a good restaurant. Every other meal they make may be a good one, but you had one meal, so you’re not going back. It’s sort of the same with YouTube algorithm wouldn’t you say?
JV: I would, definitely. I think that if you’re not getting a lot of subscribers and a lot of engagement, you’re not connecting to either the right people, like the right people aren’t finding your content, or you’re not making the right content. If I go to my favorite restaurant and I get food poisoning twice, that’s not my favorite restaurant anymore. I’m out forever, we’ll never go back.
DG: They may have some non-poisonous food but you don’t want to try it.
JV: There’s a really really big gaming You Tuber that’s experiencing a lot of those types of things right now. We work with a lot of brands. Brands aren’t human, it’s not an actual person. When you’re a 17 year old beauty blogger, you talk about make-up and your person experiences with the make-up, who you are as a person and people connect with you, almost at a subconscious level where they feel like they know you, they go through the journeys and trials of your life together. It’s a little creepy but the difference between a You Tuber and a brand is extreme. People can connect to YouTubers, people can’t connect to a brand.
That’s why amazing companies like GoPro, Red Bull and a lot of other companies have found other ways to connect with their audiences through extreme sports and all sorts of other types of content. The challenge with a brand, for example, there is a lot of content that brands put out that isn’t necessarily isn’t a good thing to put on YouTube, but they have to have it because their technical support people have to send it to someone. YouTube becomes a repository in these big organizations. Optimizing that content could be absolutely devastating for people’s experiences, normal people’s experiences.
DG: Now there was another thing you were talking about with deep learning AI and I don’t understand what any of that means, but why is that important in this discussion?
JV: I am going to pretend to know a little bit about this. I’m going to break it down real simple. It’s a little above my head. Essentially, as Matt Geelan has pointed out recently in this really amazing article, you should check it out, is Google put out this study, it’s called the deep, neutral networks for YouTube recommendation study. It’s a real fun time, you’re not going to be bored at all reading this. What’s interesting about this study and what’s really amazing is for the first time in many years, we actually get a dive into how the YouTube recommendation engine works.
Basically there are two neutral networks. There’s a candidate generation system and a ranking system. These two networks choose who’s going to watch your videos and how many views they’re going to serve to your video before they even serve it, then how your video is going to rank in conjunction with the candidates they’re going to send that video.
DG: How is this important to optimization?
JV: The reason it’s important is the way it works. Basically, watch time, session duration, session ends, how people are watching videos and their search history. The candidates are basically saying, “Hey, people are watching this video for a long time, they’re engaging with it, and these people over here, these viewers have watched similar videos in the past. This means they’re a good candidate, candidate generation to potentially watch your video. Your video is not going to rank or be shown to candidates if you don’t have proper meta, if you’re videos not watched very long or if you don’t have a lot of engagement or velocity to it.
The reason it’s important is it just gives us a little deeper dive to how all this AI business works, but basically think about it from a computer perspective. If you think like a robot, you’re probably going to do better on YouTube. Essentially, do people care about this video, are they talking about it, are they watching it? Is it related to the candidates history of what they like to consume? When you really think about it like this, it’s just a lot of common sense things.
DG: Right, and you’re speaking about this coming up soon?
JV: I am speaking at the social media marketing world conference in a few weeks in San Diego. We will be diving deep into this. More than anything, we’re talking about discoverability optimization, but we have to talk about how the deep learning system works to get in there.
DG: All right. If you guys are in San Diego for the conference, this is a must see, the Jeremy Vest presentation. You’ll hear more about this. I think we covered some good ground Jeremy. Jeremy, how can people find you?
JV: They can find me at Jeremy@vidpal.com or Vidpow Bam everywhere.
DG: Vidpow Bam!
JV: Hey, how can people find you?
DG: I’m Dane Golden from Hey.com, until next week, keep talking tube!