Garrett Sussman: Phil is a video strategist, and speaker. He’s kind of like coming out of a hole to really butterfly out to all of us what he’s been working on. He has six adamantium plates in his jaw-I’m sorry. Six titanium plates in his jaw, meaning all his talks are generated with an AI voice. One of Phil’s passions is wildlife photography. So at the algorithms after party, you might see him with his camera, taking pictures of all of us, getting jiggy with it. No. Wrong rapper. My bad. When boxing, Phil floats like a hummingbird, stings like a panda. Okay. Presenting YouTube Keyword Research 2.0. Please welcome Phil Nottingham. 

Voiceover: We were hoping he’d have something to say. But the agenda said he’d be right here to share his video SEO expertise.

Phil Nottingham: Everything’s video these days. All of this is video. Video ads for videos on a video platform. Videos. The offices of LinkedIn. Now basically a video platform. Even this, while I’m meant to be on stage over there with you, is a video recorded on a phone. Not even the best phone, just a phone, a device we all have in our pockets capable of producing content that will be seen by millions. So I have a question for all of you. The question I’m going to ask soon as I get to the stage. 

Why are you all still worried about text? The entire world has gone video, and everyone in this room is worried about meta tags, blog posts, product pages, completely ignoring the main thing you need to be doing which is video. And many of you will be thinking, well okay, but you know, we do do video, it’s just it’s held with the the social media team. Video is a massive search channel. YouTube is enormous and you’re almost certainly sleeping on it.

Let’s have a look at YouTube. Right? So YouTube has about a billion searches a day conservatively and YouTube has 14 billion. Okay. That’s quite a disparity. But what’s the number one search thing on Google? Yeah, YouTube. And then there’s a whole bunch of queries where the only thing that ranks is YouTube. If you’re not creating a YouTube video for this query, you’re not ranking. There’s no way a blog post is gonna rank for it. So if we add in kind of all of those queries, the YouTube search volume for things that are on YouTube and queries where basically the only thing that ranks is YouTube. Okay, it starts to look a bit more competitive. It starts to look a bit more like YouTube’s seriously significant. But of course, search is actually only a small part of the YouTube ecosystem. If we look at YouTube search and Google search driving views to YouTube, for most big channels, that’s maybe like 15-20% of the overall views, and the rest of them come from the browse features, suggested videos, links, shares, dark social, all that kind of stuff. So if we add all that together, YouTube might start to be driving as much attention, much influence, about half as much as Google.

How many of you are worried about chat GPT and optimising for that? Everybody. That’s how many searches ChatGPT gets at the moment. Okay. It’s going grow. We know this is important. Not denigrating that. But how many of you are worried about this and sleeping on this? A huge number of you I know. And I know why you’re doing it as well. The problem is that video is hard. It’s expensive and it’s time consuming. Maybe not expensive in terms of the gear, but it’s expensive in terms of creative expertise. People who can come in and do a really good job. And to work out what to do is challenging because what you don’t wanna do is spend a ton of money, a ton of time, and create something that you’ve put your heart and soul into that then fails.

So how do we fix that? Well, to invest, to solve this problem, we need confidence. I’m gonna try and give you that confidence through a new methodology of YouTube keyword research. I think there are kind of three core questions that we need to worry about with relation to YouTube that kind of encompass YouTube keyword research. The first, what are we gonna make? What are we gonna create videos about that are likely to generate lots of views for us? Second, okay, we’re gonna create videos. Well, what length and format, you know, what kind of videos are we making? There’s a lot of different types of videos out there. And then lastly, what’s working for our competitors so we can actually effectively benchmark what we’re doing and see whether or not we’re competing and determining whether or not we’re doing a good job. I’m going try and answer these for you today.

And this is the answer. Okay, I’ll break it down. In a 12 step process, using a fictional company that I have come up with this morning. Now, one of my great passions, as mentioned, is wildlife photography, as we are beautifully seeing with the the background here, and I’m never happier than being somewhere remote in the world looking at an elephant. So this is something that I I’d love to do, and through that process, I have discovered an opportunity for new business, and this is bags. So what you need to do is, when you’re on safari, make sure you have a bag for your camera that also contains your first aid kit, water, all that other kind of stuff. So I’ve gone to ChatGPT and brilliantly designed this new bag, which we’re going to market, and of course, we need a great brand name, so we’re going to kind of think of a wildlife related brand name that we can come up with here, and what better name than panda and penguin? 

So let’s go through these questions. First of all, what should we create videos about? Okay. So we’re going to have to think about this. Well, we’ve got topics around maybe camera lenses that we should take, maybe about what kind of bags, what should you pack on Safari, what kind of camera should you use. There’s a lot of things that we could talk about in this. So what we might do is say, well we’ll go through some traditional keyword research. So how do we do that? Well, we come up with our big old list, as we would do normally in traditional keyword research. We’re then going to sort of write some things that we might be able to create videos about. We’re then going to do that classic thing that people do with old school keyword research. We’re to go to Google, get the organic click through data, get the difficulty data, the volume from Google Ads, and there we go. And then think, okay, well we need to now apply this to YouTube. So there’s a bunch of tools out there that will offer YouTube search volume for you, and you can go and get this data. So it’s vidIQ, Keyword Tool, TubeBuddy, and TubeRanker all purport to offer YouTube search volume. So this is vidIQ, and as you can see here, we have keyword volume of was it a hundred and thirty one thousand for New York City? Okay. That many searches a month, maybe. Apparently, is Phil Nottingham the best video SEO of all time is getting 2,000 searches a month. Sounds a lot. Phil Nottingham in New York City is 2,700. That seems unusual. Hold on. Is Phil Nottingham in the naked cowboy 2,000 searches a month? Really? And then YouTube keyword tool apparently, New York City bagel has no search volume. I searched for it this morning. But then Phil Nottingham’s favourite New York City bagel apparently has a total search volume according to Keyword IO of nearly 300,000. Now, it’s Russ and Daughters, if anyone’s asking, but I think that’s an unlikely situation. And then TubeBuddy just gives me this score, film loss from New York City. Again, there’s going be no search volume for this. I’m pretty confident. 40 out of 100, what does that mean?

Okay. This data is junk. The problem with it is that all of these tools are trying to replicate the stuff that we think we want. What we want is search volume. But nobody has access to YouTube search volume, except Google, and they’re not giving it away. So in trying to replicate this old school process, we’re telling the tools that what we want is something that they can’t provide, and they’re coming out with this kind of generic crappy data that doesn’t tell us very much, and it isn’t very useful. So we can do better than this. This is what we come up with. This data tells us nothing. All similar numbers, apparently, that’s the keyword volume. It bears no resemblance to reality. Alright. We can do better than this. Imagine a world where every single search result in the Google SERPs had the data of the traffic that that particular page was getting on the search result. That would be pretty cool. That would change the way we thought about YouTube, about YouTube, about Google, about search. It would change our method of analyzing competition, demand, all that kind of stuff. And yet, we have this with YouTube. Every YouTube video tells us how many views it got and how old it is. And actually, this very simple piece of data, we can extrapolate some extremely important insights. 

When I realise this, I mean it’s so simple and yet it completely changed my thinking about YouTube keyword research. Over the past couple of years, I’ve been developing a methodology that has been very very beneficial for lots of clients just by starting with this core insight. We have the data and it’s more important than keyword volume. Okay. So how do we think about this? Well, let’s start with the view. Okay. So we have the view count. We’ve also got the age. So when the video was published. So there we can actually look at the monthly views and say, well, how many views has something got over time? And then if we get that data at scale, if we get that data for every single video ranking for a given topic or query, we can then look at how many views things are getting over time overall, the monthly view velocity. But then we might want to say, well, there’s a lot of videos in there, some of them might be getting few views, but some of them getting a huge number. Something went viral, a lot of ad spend was put on it, whatever. So actually, the view velocity is not really telling us anything that important. Okay. So then we need to find a method of of coming up with an average number. Well, then we can take the median. So we have this metric, median monthly view velocity. This is so much more valuable than keyword volume or what is pretending to be keyword volume because this is true, and it tells you a reasonable proxy for demand for a given topic. So what we can do is get median monthly view velocity, and I’m gonna show you how to do it. So you start by just getting your big list of keyword ideas the way you would with keyword research anywhere else. VidIQ is actually really good for this. I know I was having a go at the keyword volume data, but it scrapes YouTube suggests, and that is very useful for that. SMRash, YouTube, of course, just look at what other competitors are doing, Google Trends, all that lovely stuff that we’re used to. Nice and simple. 

Step two, a bit more complicated, but don’t let it put you off. Here, we need to actually go and build and work with a scraper. Now, I’m not that technical, It took me about a day to work out to do this, so everyone in this room can do this. You can use Playwright Selenium, which will use Python and JavaScript just to kind of build a very simple scraper. If you don’t want do that, there’s also like Browse AI and No Kite option, or you can even just pay for existing ones on Appify that will give you this data. So don’t let a bit of scraping put you off. Get the top 500 search results for each keyword, and then upload that to your favorite LLM and categorize the output into topics. I’ll explain why this is important in a bit. But get every single keyword that you’ve got and say, well, come up with a list of topics that I can then categorize these by. Then do our most favorite thing in the world as SEOs: Put this into a massive big old spreadsheet. Crash Chrome. Do whatever you need to do, and have all this database in one place. And from here, you can then do something else we love to do as SEOs. The good old fashioned pivot table. And through that pivot table, divide the number of views by the number of by the age of the video, and you can then get the median monthly view velocity for all your different keywords, and organise them by topics. Now this is the start of something quite special because here you can actually see a proxy for demand. You can get an idea of, okay, well actually best camera for Safari, lots of demand there relative to other videos. These are like the median number of views that every single video is getting per month on that particular topic. So if I’m able to rank, if I’m able to get into that top 500, that’s how many views I’m likely to get.

Okay. Nice. Then we might say, okay, well that’s good. We can see there’s a lot of competition. What’s packed for Safari? Not so much there, but the camera comparisons is really hot. Something about the lens for bird photography seems like that’s got a lot of demand as well. So we can start to prioritize some ideas based on demand. And then we can go a step further and say, well, that’s demand, but what about competition? Because this best camera for Safari might be a really popular term, but maybe there’s just so many videos on that that I’m never going get in the list. Okay. Well, what we can then do is find the median number of subscribers. And this is actually a really good and effective proxy for competitiveness. Because usually channels that have done very, very well over a long period of time, end up getting a lot of subscribers. So if a channel has a huge number of subscribers, that tells you it’s quite competitive. So here, they’re comparing two different Canon cameras. That seems to be quite competitive, but we might see an opportunity here around that best lens for bird photography, where there’s a significant amount of volume, but a little bit less competition. So here, we actually have something tailor made for YouTube that is far more effective, and actually is a reasonable proxy to say, demand and competitiveness. The kind of thing that we want in keyword research, but actually tailored to YouTube properly. Alright. So then we’ve got basically question one solved. This is going to tell us roughly what we should make. 

Question two. In what length and format? Okay. Well, now we can take our data a little bit further. So there’s a few different length formats that I tend to think about for YouTube, and this is just sort of born off a bit of old school TV formats and more broad formats with online video. So you have Shorts, now 0-3 minutes, important format for YouTube, 16×9, usually ads, clips, that kind of thing. And you have snacks, 3-15 minutes, something that’s gonna take as long as time as a cup of coffee to consume. And this is often going be shorter videos, covering a single concept, maybe instructional videos, that kind of thing. Episodes, longer things, things like podcasts, things like that sort of TV format, and then features, which are those really long, high investment videos. And what you can then do is say, thinking of these four different buckets, how does that compare to what’s out there on currently for a given keyword? So we can then find the average length of videos through our scrape, and then also find the variance of that length. So here we can see for these three middle one in green, these videos are kind of 12-40 minutes in length, and there’s actually low level of variance. So that tells us if we want to rank for this topic around best lens for bird photography, we are going to need a video that’s about 10-15 minutes in length. Anything longer than that is probably not gonna rank because that is the kind of video that people seem to be watching and wanting for that keyword. However, for what’s in my bag for wildlife photography, okay, that’s got a high average length, but actually a high variance as well, which means that there’s lots of different kinds of lengths of video ranking for that. Maybe some short, maybe some long, so you can be a bit more creative in the format that you’re going after. And this again just gives you that bit of data to give you the confidence to know, well, what I’m going to make and recommend is highly likely to meet the demand that’s out there on YouTube.

There’s also kind of different aesthetic formats, I tend to think about them in these two different things. There’s your static videos, so this is your simple shot things. Things in studios, things out on location maybe, but just usually talking to heads, lectures, podcast panels, that’s stuff that has very very simple camera work. You’ve then got dynamic formats, so this is stuff that usually has movement in the camera, maybe close-up product shots, archival things like documentary filmmaking, all that kind of thing that requires a bit more complexity with camera. And then lastly, you have artistic, which is probably the most aesthetically complicated format. So this is things like big scenic shots, this is abstractions, animations, cinematic video. So all that kind of stuff that maybe takes something with a camera and abstracts it further, or would work without a camera at all, and is done through animation. And what we can then do is work out, well, which of these types of shots do we need in a particular video? Now unfortunately, is the one part of the process that is slightly manual, but it’s not complicated. So what you can do is go and find those top videos by view velocity back in our massive spreadsheet, and then we can go and pick some of them, and say, well let’s just quickly watch them through. So here’s one of those videos, and Okay, so we can say this has got monologue content. It’s on location. There’s some product close ups in there, and it seems like there’s some b roll as well from other videos. Okay. Cool. Those are the shots that we need for a video that’s ranking right now for this particular topic.

What we can then do is go and create a creative brief, and say, okay, creative team, go and make a 15 minute video in a monologue style, shot on location in the Maso Mara, about the best lenses to use for safari photography. We want B-roll shots of lions and close ups of lenses. It’s a pretty good creative brief. Somebody can go away from that and say, okay, we know what we’re doing, we’re gonna go build this video.But I know it’s been about 20 minutes and I’ve not mentioned Generative AI yet, so I’m gonna sort of do that now. If you go and take that brief and give that to an AI, you’re gonna get something that looks like this. Okay. It kinda makes the brief. It’s not very good. It’s not gonna work. It doesn’t matter how good AI gets, it’s never gonna be the solution for this kind of video. For most of the videos that you’re making on YouTube, it’s going to be camera driven. Where AI really helps is in this kind of format, It’s in the artistic stuff. It’s gonna be really useful for things like ads, music videos, fun creative stuff. It’s important, it’s useful, but here is where it helps. Because it necessarily thinks abstractly. Most of the stuff that does well on YouTube is a lot simpler. It has static camera work, it has some dynamic camera work, and all that stuff, what matters is that it’s a human speaking. So AI is never going to be the answer to that. I don’t think AI avatars are going anywhere except for maybe product support at the margins. It’s not going to be good for marketing.

So that’s kind of where you need to think. More broadly, if anyone’s interested, what AI art is in the video world? Well, here’s what art is. A selective recreation of reality according to an artist’s metaphysical value judgments. It means it’s the vision of the artist made concrete. It’s their worldview. It’s the way they see the world. Their perspective on things. Now, what AI art can do is the selective recreation of reality. That’s what it does every time you put in a prompt. Can make something new. Okay. We’re going to choose these things. Here’s the thing. But what it can never do is metaphysical value judgments. That comes from the human. And what we find is when we’re creating art, if anyone goes and makes AI video, you find that essentially your job is to be really decisive, and you start playing with lots and lots of different elements, and you go, no, don’t like that, do like this. It’s kind of like being a director or a conductor. And your job as a sort of creative in that is to be very, very specific with stuff. So I love AI video and generation. It’s and it’s awesome, but remember what it is about and what it can do for you as a creative.

Okay. So then we’ve got an idea of format, and now we’re going come to our final question. What’s working for my competitors? Alright. So how do we work this out? Well, again, we go back to our lovely pivot table, and we can expand it further. And we can then go, okay, well let’s go and build a new one based on the channels that we got from our scrape, and the topics that we pulled from our LLM, and then we can map out how many different videos are ranking for each different topic. And from here we can then see, well actually we’ve got this content competitor here, Pangolin Wildlife Photography, who are really killing it in a whole bunch of directions, and they’ve got some good content on bags, cameras, lenses. So, we should go and have a look at what they’re doing, and work out what’s working for them, and then start to replicate that as well. So, we might then go and scrape that channel data. So go to their channel, run our scraper again, get all that information, and then go and watch those videos, discover what’s working for them, and then give that to our creative team as a reference point. Because we know that’s working for all the things we want to rank for, and therefore we do a good job or a better job, we’re likely to rank as well. And then you can also go, well let’s find all those competitors, get all this great data, and then say, how many different videos have they made for each of the different keywords and topics that we want to go for here? So we can then do a count of competitor videos that gives you an idea of firstly, like where they’ve invested and where it’s worked. So maybe you might wanna do the same, but also where those opportunities are. So maybe they haven’t made this video yet, and that’s something that you could go and make.

Okay. So we now have everything. We have the number of subscribers. We have the media monthly view velocity. So we have a view there of demand. We have a view of competitiveness. We have a view of age, which is actually going to tell us how long something’s likely to rank for if we get it into that top 500 for that query. We have the average length, which tells us how long a video should be. We have the variance, which tells us whether or not the length is important. And then we have the counter competitive ears to see what’s working for them and what we can replicate. And now, you have a YouTube strategy. Just through a good old fashioned spreadsheet. 

However, this is Mike King’s conference. And I know that you aren’t gonna do any of that. You’re not. So I’ve done it for you and built it all. So if you thank you. And this tool is going to give you all this data. It’s going to give you the information on the average view velocity. It’s going to give you the average duration, subscribers, all the stuff I’ve just talked through. You get the detail on the different channels, and everything you’re going to need to go make this spreadsheet. So it’s free. It’s all available to you at this URL here. Now, I have to run the scrapers to do it. We’re not super fast at the moment, so it might take me a couple of days to get this data. But if you go here and plug everything in, I promise you’ll get it and you can build this strategy. Alright. Thank you very much. This is the most important metric when you’re making your videos to optimize for. Impression click through rate, which you get in YouTube analytics, tells you how often people are actually choosing to click that video, and it’s defined essentially by how good your thumbnail is and how good your title is. 

So I’ve got another giveaway for you, which is a load of templates that I guarantee are gonna work. Now this is basically based off every single sort of loads of client videos that I’ve worked with over a long period of time. It’s about 80 million I think in total. So it’s a huge number that were based into this, and then we kind of mapped out the formats that worked really well. Got a designer to make some templates. Go and just take these. Use them however you want to. Upload them to chat GPT. Get it to remake the thumbnails. Do a Canva. Whatever you want. So here’s something that’s going get you started if you’re struggling to get that impression click through rate going. Make better thumbnails. Come up with more clickable titles, and you’re soon going to see your videos ranking more, more views, everything else. Alright. That’s me. Thank you.