Brand drift happens because AI powered search engines and AI models fill in gaps because they don't have they don't have enough information from you as a business to tell your story, because your story now is told by all this artificial intelligence that crawl, all of the contents that you have available, and sometimes content, you're not even aware they're crawling because maybe it's a very old file in a public folder on Google Drive, but it still reads it and it uses obsolete information, or it invents information. It hallucinates.
Yeah. If I was trying to play that back to you in in language I can understand, if it doesn't know something about your brand, it'll make shit up. Basically. Welcome back to Digital Marketing From The Coalface. Um, this week Julie's over in Miami, supposedly on a trade mission, but almost certainly just laying around on the beach before heading off to New York to go and spend the weekend with her son. So lucky. Julie. But it's been fortuitous because Lesley has been dying to speak about video. You've been experimenting and playing around with video and how AI is is looking at processing maybe a better word, isn't it, how AI is processing video and the importance of it and everything else. So I think you have a little introduction to sort of set the scene, but, um, Lesley's looking very serious as well. So we're going to be talking about video and you want to set the scene with a little bit of a bouquet.
Yeah, absolutely. So I was just generally.
Interested in like in the relationship between AI and video. Most of the time when people picture the use of AI in video, it's more like generative AI, creating clips and stuff like that.
So I need a video.
Of.
A horse jumping over the moon. Create the video for me, that sort of thing.
But AI in video has been used for quite some time in like within professional software being like transcript generation on premiere pro or other sort of options like that. But I've been increasingly exploring the link between video, SEO, video strategy, video creation, and it turns out that now AI in video SEO is just something very, very surprising and a bit scary.
You see, what you found when you did your research was both surprising and scary. Yeah. Yeah.
Okay. So basically I found these two articles from and I'm going to pronounce her name French style Myriam Jessier. So she's an international consultant and trainer, specialised in technical SEO. And she's I guess, um, evolving with her time. And she dives into the topic of how AI crawls video content. Okay. And what it means for businesses online presence as well. Because at some point in one of our articles, she talks about brand drift. So I found that very interesting. So basically the, like the name of the article is how to optimise video for AI powered search. It was published last January. Okay. And one that is slightly older from August last year. How generative AI is quietly distorting your brand message. Mhm.
Okay. So we're going to have a little dig into that and look at it. Okay.
Like, I know.
This might be one of those podcasts where I don't say very much, but I'm quite unusual for podcast for me not to say anything.
Uh, so since I'm in charge of video creation for us and for our portfolio of customers, it's just very interesting to see how AI powered search impacts the way we will produce video in the future. Mhm.
And so are you saying that we will review we are now reviewing how we're creating video based on our understanding of how AI consumes video.
Yes. So basically it's not I mean, I remember this podcast episode we recorded in twenty twenty two, I think with Elliot Hornell. So he's a videographer and video strategist, and he was talking about how filming without a proper strategy, about what your content will add to your brand or how it will interact with your constellation of existing content will fail to deliver the results you hoped for and you invested budget in. Yeah. But now I think- So, that's something that should already have been part of the video strategy video production strategy too.
Yeah. Why are we doing it? Who's it for? Where do they hang out? All that kind of stuff.
But now, more than ever since, AI is actually scanning, watching, listening to your videos and making connections between, uh, the, like the data stream it gets from the video, it's even more important now to do that. Otherwise, you're just going to, well, risk that brown drift that Mariam is talking about and that we're going to talk about. Okay. So yeah. And I think that's just very interesting to explore the aspect of videographer will need to take into account, like the impact their work has on strategy, like working with marketing strategists more closely, maybe. Mhm. And I guess also SEO strategies, like, like marketing strategies would also have to rely on video producer to make sure that what they produce is not only like for bots to read either. And I think that's a problem. We, I mean, you had in the past when writing blog posts.
Yeah. So when you're writing content and you hope that content's going to potentially attract customers, you have to write it so that it's engaging for people. Yes. Human beings to read and make sure that the way you put it together helps or improves, or rather gives it a chance of being found in search engines. And what we're saying is we've always done that with video as well by using transcripts and titles and descriptions, etc. but what we what we now have is the AI's ability to go inside the video and extract data that isn't explicitly produced or provided by the, the producer of the video, as in it's transcript. It can actually go further than that. Mhm.
And so I guess the risk would be to just create a video that is empty of like depth or relevance for like human viewers and try to optimise it only for bots. And that wouldn't work.
Could we liken it to, um, somebody like Mark Kermode actually analysing a film and watching a film and then giving us information about the film and the nuance of the film and the and rather than just read like, this is a film about, um, you know, this is a spy thriller where, you know, the protagonist tries to, um, find a document that's hidden and hidden in a vault in Moscow or something like that, and he can get into it and explain the characters and how this character's mind works and, and how that feeds into the plot and everything else. So like in, you know, AI is able to do that. It almost able to see the nuance of what's going on in the video.
Yes. If you provide it with the right information, but nuance is definitely a word that comes back in my notes. Uh, so yeah, basically because we haven't explained exactly what AI is doing. So, um, videos are now watchable by AI, basically. That's what she says in her article. So that means when you produce a video, know that the video now can provide AI and should provide AI with a high density stream of data I quote to to help with indexing the content more accurately. Traditionally, as you said earlier, what was crawled before this AI powered search search analysis. That was all the metadata sunshine. So all the metadata related to a video. So title, description, transcript, uh, tags, this sort of thing. These are still important. But now AI powered search is tokenising. That's the word tokenising videos. So it's taking snapshots at regular intervals of your videos. So basically it analyses faces expressions the objects in your frame.
To say what does it do with objects. Would it would it, you know, if this video, if we videoed which we are videoing, but if we videoed this room, would it would it recognises there's a pushbike there? Would it be able to tokenise that and then realise by tokenising it? Ah, that's a pushbike.
Well, you know, the maximal effect we had that I'm pretty sure comes partly from that. Like iterate your face, it read your glasses, it made a visual connection that was totally irrelevant. So I guess right now it doesn't really know what it does, but it makes association.
Here's a question then. So if we did a silent video, yes no soundtrack and we did the just scanned around this room and then AI consumed the video, would we be able to ask questions and say, is there a is there a push bike in that video? Is there a white wall in that video? Is there a door in that video? And it would it would be able to do that. It would able to be able to like feed back.
So it would depend what type of model you're using and questioning. Okay. Because for example, if I'm thinking about the AI that YouTube uses, uh, to maybe link videos between one another and just suggest new videos for users, which is where we had the spikes from. Yeah, from yeah. To our playlists. Um, you can't question it directly about that, but it might make connections and sometimes failed ones, which is the problem. Um.
But my page has just gone off.
I mean, I know where I stopped. Yeah.
Just pause it just now. Right. So as you may have heard, my pager went off and I've, uh, just been out being a fireman for, uh. Oh, no, sorry, I can't say that firefighter for a little while. Um, and we're gonna now continue. So apologies to Leslie, who's put a lot of prep in for this. So it's sod's law that my pager went off. But anyway, here we are back. And you can remember where we were. Yes. Good.
So you asked the question about if, uh, since AI models can take screenshots of videos and they identify an object, an item maybe with a brand on it. And you took the example of this bike. Yeah. If we query them, can I say that if we query them, uh, can it actually make maybe the video pop up or, or. Yeah, I had the time to think about your questions. So that's the good thing. So I think first we have to make the difference between AI powered search engines that we mentioned since the beginning of the episode. So ChatGPT Gemini, you ask questions, you type questions and they answer, yeah. So I guess videos could come up maybe based on what you asked them. Yeah. They don't treat they don't read videos the same ways, but all the information that we're going to mention, they take them into account. I think they're just like comb through them in different ways. And there are also AI models that curate your experience for you, but you don't query directly. For example, the YouTube algorithm. Uh, I don't really know how it works. It's been purchased by Google for quite some time now, so I guess it's maybe linked to how. Okay. Yeah. So, so this, uh, I don't know, because the bike is there. Maybe it might suggest a video for you to watch next. Okay. With the same bike featured or or maybe after this video is posted online, uh, someone will will see a video of a review of this bike and then have a podcast episode. I don't know. Um, so that's.
What about, I don't know if you know the answer to this, but you know, notebook LM where you give it the sources you give, the AI, the sources you want it to use. Um, you can suggest YouTube videos. So is it, is it only looking at transcripts? Is it able to understand in inverted commas the video beyond a transcript in order to use it as a source when it's in in notebook LM are you not sure?
I'm not sure because I mean it's very impressive what it can do. Um, but yeah.
Like you're a big fan of notebook and I know I am.
Yeah, I think it's very useful to summarise or rephrase existing content, but I don't, I haven't explored that. But, um, so yeah, they work differently when they analyse video. But yeah, basically what Miriam says is that this idea of the video taking snapshots, uh, and then after the video, listening to your audio tracks and making connections between the two, they basically do all of that just in slightly different ways. So yes, to answer your question as well, it also sees watch like the titles like text elements, uh, and labels on products. And when recording a video about a product, even doing a demo, it's apparently recommended that you show it under different angles so that you turn it around so that the AI understands like, okay, it sees it under different angles.
Yeah, yeah.
So that's, that's an important thing. And maybe even more so in a niche like ourselves, if customers want to advertise about very niche products, technical products, that might be something that you want to do. It's always nice to do. Yeah. But yeah, so there's this. Then they listen to audio tracks for not only transcript transcribe. Sorry your speech, but also to read emotional cues. So if you sound authoritative or whatever and background noises as well. So maybe it tries to understand where your set.
Yeah, it's trying to create context, isn't it? Trying to understand.
That's the thing. And after that, so it's connecting the dots between the information it got. And that's why um the quality of production is important as well because the clearer an image can be or a sound is, the easier it will be maybe to make, not to um.
Interpret, um.
Have a clear understanding of what it sees and what it hears and make a connection and not guess.
What I find with video is it's rarely the case that the video quality is poor with videos, you know, with, you know, with the kind of videos that, you know, you just user generated content companies shoot their own videos or whatever. Um, but the sound quality often is poor. It's the same with podcasts when we know when we started podcasting, we bought this lovely kit and it's the sound quality is great. And you know, the challenge, I think, I wonder how challenging it is. I think, you know, getting what I was interested when I was reading some of the notes that you sent over before we recorded this, that the modern age, there was often a mantra, don't worry about the quality. Don't you know, you don't need to be, you know, like creating what used to be corporate videos, which were very polished and very boring and very like, you know, video today it can be rough and ready. But what you're saying with the AI is that production quality does matter in order for the AI to be able to interrogate it.
Yeah, yeah. The clearer it is, the just, the better it will be for, for it to read your video. And at the same time, Miriam JCM mentions the fact that uploading a 4K video that is super long and super like big in terms of file size.
Yeah.
It's not, it's not necessarily the smartest thing to do because that might be too much. There is like a balance to find.
Yeah.
So I don't know to which extent, but that. Yeah. And uh, yeah, so basically you can't really opt out of your content being crawled by AI in any way. Like, I mean, like any sort of content uploaded online now.
It's just like with pages, with content, you can, you can actually, uh, tell the crawlers to ignore.
Yeah.
That's you saying you can't do that with videos.
No, not to my knowledge. No. So you have to pay attention to, to well, you just have to keep all this information we're discussing in mind when you're producing video now. Uh, that's the thing to come back on how like how to optimise a video, I guess to be like read by AI crawler. I think I can talk.
To be to be understood in inverted commas.
Yeah, absolutely. So Miriam says that the content needs to be clear, specific and high quality. Like to to relay a very clear, specific, high quality information. Right. In the sense that, I mean, I think she said that it's more impactful that long video, longer long form videos. But, uh, she says that long form videos are a smart thing to go for because the longer the video is, the more snapshot.
Okay.
Snapshot, sorry can be taken. It can be tokenised more, um, like with more density.
Are we saying like the AI is snapshot in a frame and then matching that with what's with the sound that's happening at the same time and at the same time it's trying to create.
Uh, but the thing is, I think, uh, I don't have the number right now, but it's mentioned in the article.
And you did mention it when we were talking earlier.
Yeah. Uh, I don't think the ratio of tokens that is taken for images and for sound is the same. I think less token are being taken for for sound than for images that might not occur at the same time. Okay.
Um, so we're saying that the quality is important if we want AI to be able to do something with it so that maybe it shows up in AI search and maybe it puts your brand in front of people, etc.. But the good news is that that's good for people anyway. That's good for humans, isn't it? You know, we're not saying that we have to or I'm asking the question, are we saying that we have to dumb it down so that the AI can understand it better? Or as long as it's good, clear content that's that's well produced, then that is like making it great for people, just like we did with SEO. Make the content good for people as well as good for search engines. So like we're ticking both those boxes. Is that what we're saying?
Well, I think, uh, there's very interesting. And I think that, I mean, that just leads to the conclusion I wanted to talk about in this episode is just that it's about, I think, creating a constellations of different form of video content around your brand. Okay. Because longer form videos will be more dense sort of contents for AI to crawl and people. Yeah. And sometimes it's just the nicest formats to go for. It's just that with Instagram and TikTok, we just heavily edited down videos with barely time to breathe for people when they talk, because it has to be dynamic and it has to, you know, you have to keep people's attention because they have a low attention span on social media, but.
Particularly prevalent on, on TikTok where, you know, if you, if you swipe onto a TikTok that's trying to flog you something, they start talking immediately. You know, they try to, you know, to me, that's a cue just to keep swiping. I mean, just to move on. But yeah. But yeah, those, those kind of, you know, there's a style of video, which I've watched where it's clearly been edited in a way where like I'm talking now and I'll have to pause there. And I've left another pause there, just on, just because that's how I'm conversing. But they'll actually chop all that out. Yeah. So it's like boom, boom, boom, boom boom. I've heard podcasts doing exactly the same. You know, it's not just videos.
Okay. Okay.
Yeah. It's like a technique.
Yeah, absolutely. I mean, it's people decide to watch your video or not on social media in a matter of seconds. So you have to just hook them the right way. But if a video lasts less than a minute, it doesn't give me more like much time for AI to crawl to take snapshots. Like it's going to take snapshots. I don't know every 10s probably not.
Yeah.
But if you have a ten minute long video that is engaging and quite nice, um, it will be able to take more snapshots.
So long form content, long form written content has long been believed to be effective. Yeah. You know, long form where you do get into the weeds and explain it all. Yeah. It's funny because it's effective from an SEO point of view, but probably there aren't many people will read it from start to finish unless it's covering something that they really need to know about. It's one of the reasons we put the audio overviews, not not overviews. We put the audio version on our on our blogs. So people like me who are lazy, you can just go, you know what? It sounds interesting, but I can't be bothered reading it. I'll just listen to it.
Yeah, yeah, I guess we questioned and we as like, like marketers worldwide. I guess we wonder this past couple of years if long form video content was dead or yeah, you know, at some point, because everybody was talking about TikTok and it really shaped you like a specific way of editing videos. And it really frustrated me as an editor because it's just so annoying to do that because sometimes you just, you can't talk about a topic. No under one minute.
No.
That's right. And also it's just a very dry thing thing to do as well. Um, this sort of editing, but we've naturally come back to like longer form content. And I think that's something that is encouraged by the current CEO of YouTube as well.
What does long form content constitute with with video? I mean, if I'm thinking the written word for me, long form content, two and a half thousand, three thousand words, if you want to listen to it, it'll take you twelve, thirteen, fourteen minutes or whatever to listen to it. What is classed as long form video content?
Well, I should have said longer forms of content because long form content, I guess when you say a long feature film in cinema, it's above one hour and like, like a moyen age in French. So like middle, like medium feature film, I don't know how to say that it would be around.
thirteen and a half.
thirty minutes or even thirty minutes. But on YouTube, when you say longer form of content of contents on YouTube, when I click on a video that is between forty minutes or one hour, it's very like very long for me, like on topic. I know for me it's like political stuff and these sort of things. I know I want to watch from start to finish. It's about like constructed argument around a topic that I'm interested in.
Is that maybe because, I mean, I, I often judge podcasts, if I see a podcast and it's like twenty minutes long, fifteen, twenty minutes long, I'm not that excited. Even if it's a subject I'm interested in. I almost want to invest forty five minutes to an hour in a podcast. You know, I kind of like that, but that's partly because I can do something else at the same time. I can drive, I can cycle, I can walk, I can do something. Whereas with video, obviously, you know, it commands people's attention. So you have to not do anything else and sit and watch a screen. Yeah. So maybe ten minutes is kind of considered kind of more sort of long form with video.
Well, yeah, honestly, when, when we talk about that, my mind keeps going back to YouTube. When we talk about like watching video and the thought of things, of course, it's not the only place where to watch video, but you have to question when you produce a video and you realise it's going to be long where I'm going to, where am I going to just upload it and, uh, and YouTube is more and more, it has more and more of its viewers actually watching content from YouTube on their TV, like really increasingly so in the end, if relevant for your business or like talking about a topic that is not dry and commercial, maybe, maybe people will end up watching that on their TV and maybe one hour is fine because they won't skip it. They will watch it as a program.
So, so my a rule of thumb, be it make the video as long as it needs to be.
Yeah, absolutely.
Don't try and say, well, it must be ten minutes because the AI algorithm won't won't like my video if it's more than ten minutes long. It just really make it as long as it needs to be, which works both ways, obviously. Don't make it an hour long because you think it's long form content and it's got more gravitas.
I guess it's just keeping like sticking to a script that is clear and well organised. And then if it's like, if you have a long content, you can always chop it down to smaller, like shorter form of content and make shorts out of it. On to post on YouTube or on Instagram or TikTok, and then it links to your long form video content. It's just, yeah.
The equivalent of posting a snippet about a blog post on LinkedIn to try and interest people. Certainly click through and read the whole thing. That kind of thing.
Yes. And so I mentioned YouTube Seo's CEO, sorry, earlier. Uh, and I actually noted down something that I found interesting. Um, so I guess I'm just going to read. Okay. It's not very long, but I'm going to read. Neil Mohan's two twenty twenty six roadmap reveals that YouTube is evolving into a unified ecosystem where creators, commerce, AI, and entertainment converge. For digital marketers, the opportunity is not to chase every new feature, it is to design integrated, integrated systems that use shorts for discovery, use long form for depth, use creators for trust, use paid media for scale, and use commerce integrations for conversion. The marketers who succeed in twenty twenty six will not be the ones who produce the most videos. They will be the ones who build the smartest video ecosystem. Yeah. And so I really, I mean, the key word ecosystem for me here was very important. So the idea of ecosystem and a constellation of content, like creating clusters of content across content form not only around like video, because I said you can have, um, like a long form video and then you can create short shorts from that, but you can also create a video from your blog post and create a podcast of your blog post or do a, like an audio version of your blog post. These sort of things. So yeah, uh, it makes sense, particularly with because we haven't talked about brand drift right now. But the thing is, brand drift happened because of, uh, AI search engines.
What, what did brand drift? Yeah. Sorry.
Uh, it happens because AI powered search engines and AI models fill in gaps.
Yeah.
Because they don't have the right. They don't have enough information from you as a business to tell your story, because your story now is told by all this artificial intelligence that just crawl all of the contents that you have available, and sometimes content, you're not even aware they're crawling because maybe it's a very old file in a public folder on Google Drive, but it still reads it and it uses obsolete information, or it invents information, it hallucinates, or it takes information that a competitor of you. Mhm. Maybe they have a similar product with a specific feature. And because you don't advertise about the fact that you don't have this feature or, uh, that the feature is different, it will just tell people when they look up your product online that, um, yeah, it exists. This feature exists on your product, for example, when it doesn't exist. So she mentions the fact that video content can be a way to correct this brand drift that can happen. So basically, she says that the idea is to give the AI algorithm as much updated information and facts about your brand to help it not make assumptions about your business. That what we just talked about, that it doesn't start hallucinating facts about your brands, your services, your products, or your team. And because, because if you don't provide it with any sort of information, it will start comparing what you offer with your competitor's available content and make assumptions about your business based on what content is readily available in your niche market. Yeah. And that's how brand drift happened. But basically, I'm just summarising what we've been talking about.
Yeah. But if I was, if I was trying to play that back to you in a, in a, in a language I can understand, if it doesn't know something about your brand, it'll make shit up. Yeah, basically. And we know, we know that AI does that. I mean, one of the things I find myself doing increasingly, the more I use AI is I have to be really careful if I'm asking if it's helping me write a report. I mean, I was working on something yesterday and it made a statement of fact and I said, well, how how do you know that? And he said, well, actually, yeah, you're right. I don't really know that. What I, what I've done is I've gone and seen how many of a certain type of car are registered as existing in the UK. And I've based that data, I've used that data to figure out that there's no search volume for the thing that you were trying to research because. But, you know, it's a fair enough assumption to make, but it was really just, it almost feels like AI is so keen to please it doesn't ever like saying, sorry, I don't know. I mean, the only time it falls over is when it hits technical barriers. If you're trying to get it to review a web page. And I was trying to get I was trying to get AI to review a car to tell me about a car on AutoTrader. AutoTrader uses JavaScript to generate the pages. AI couldn't understand it. Couldn't read it. And in the end he said, look, I'm really sorry, but, you know, give me a screen grab of it, which I did, and it worked. But it said, I cannot look at this web page. I can't read this web page, but very rarely does it do that. Mhm. So but I wonder why like we can, we can help to avoid the brand drift and the AI hallucinations using video. I wonder what it is about video. Is it just the volume of data in inverted commas, information that go into a visual communication method like video?
It's one means among others. Yeah. Because, uh, when I was talking about creating clusters of content, these sort of things, uh, I guess AI powered search engines are just going to find any sort of information that are available. I mentioned all files in public folders forgotten. So it's going to look for like official materials. Information about your logo is going to sift through user generated content about your product. It's going to make a compilation of all of that. And well, just video is just part of it. It's just that traditionally, I mean, so far it hasn't really been able to tokenise video the way it's doing it right now. And I don't know if video is the form of content that is most produced. I would be interested to know about the ratio between written content and podcasts. But the fact is, it's quite prevalent because because social media and these sort of things. And, um, and I forgot a question. Uh, how. Yes. Well, I think it's just part of how to avoid brand drift because it's a way to show how your products work, because you can do demo videos. So you can just, it can help you if I can find in my notes. Exactly. Ah, yes. So it's about providing AI with nuanced correction and trust. And I guess video is a media that allows you to do that in many different ways because you have walkthrough, video. Demo video, explainers, video. Um, and just, uh, make a video that introduce your business, your team. It's just about finding creative ways to share information because they're going to be cruel anyway. So just film. Yeah. Even like putting things on social media that are like relevant, uh, you being at an event, these sort of things.
Um, so would you say like, for example, if we took our website where we've got, uh, educational content, blog content, um, podcast content that we like, like the content we're creating at the moment. And then if we come down to the service pages where we talk about specific services that we provide, would they all benefit from, you know, a nice, clear, concise video which explains that service like piece to camera, like me talking to camera or you talking to camera, that would be something that might help us if we want to make sure or at least encourage the answer engines to to use our content and include us in their citations.
I think it would be definitely worth experimenting with, um, to see to which extent it could have an impact. I think it would always be nice for people to have a talking head video on a page to watch because as you said, sometimes people would rather watch a video or listen to a podcast than read. Um, but that could also definitely have an impact for AI powered psychology. Maybe when people would look on like ask a question on ChatGPT about, uh, what is HubSpot? Maybe our video would be suggested on ChatGPT as an answer.
Well, I mean, if you think the, you know, it's often mentioned that the second biggest search engine in the world, and I don't know if this has changed is YouTube. So people are go on to YouTube. I mean, I, I go onto YouTube, I've mentioned this before. If I'm thinking of buying a certain car or a certain motorbike or a certain fishing reel, or it could be anything. And, you know, including when I'm researching some technical subjects that I need to try and understand a bit more in a bit more detail, I'm looking for video content. So yeah, I mean, it makes sense that AI is on a mission to index and understand in inverted commas video content, because there's going to be money there. Yeah. Because AI is now moving towards, you know, ChatGPT is going to start experimenting with, um, showing ads. Yeah. Supposedly ads relevant to the query that somebody typed into ChatGPT. Here's what I think. And by the way, here's some links to businesses that do what you're looking for sort of thing. I think that'd be interesting to see how that goes for them. Um, yeah, I don't know whether I mean, my tool of choice is Claude sometimes Gemini um, seems obvious that Gemini would do it because Google will be wanting to backfill the money it's losing from from PPC, from Google ads and things like that. Or it's going to try and integrate Google ads into the into the answers that Gemini is putting together.
It's very interesting because we don't really know where it's going, and it's going to be shaped by the way people interact with AI, and the way people interact with AI will be shaped by the way AI evolves. Yeah. And so it's almost impossible to predict. Mhm. So yeah, I guess we could conclude. Okay.
Conclude away.
Yes.
Um, what's the, what's the main takeaways? What should people be thinking about?
So basically, I think it's still important to guide AI with what we used to do before, uh, with, uh, structured text like titles, metadata or any sort of metadata like hashtags, keywords, transcripts. This is still very important. So creating not a lot of content like the CEO of YouTube said, uh, it's not really necessary to produce the video a week, but if you produce like regular content and maybe longer form of contents, that might be. I mean, that will in it will effectively bring more data to AI to grow. So that's, that's a good sort of thing to go for. Mhm. Shattering the video also is something you should have, you should do already if you post a video on YouTube, because it's been quite some times now that when you search on Google or I'm sure if you do that on ChatGPT or Gemini as well, that's the same sometimes chapters in a video happen because it has been chaptered. It links to you to an answer that is addressed in a video. So that's important.
It tells you to the exact part of the video, and you just click the button and it'll play that part, the video for sure.
So that can help search engine to best deliver the right information to potential viewers by just leading them to the right parts of your video, a specific time code. Uh, so resolutions and readability, important and font choice and contrast important in your video. Mhm.
Um, but are there important anyway?
Yes. For people. yeah, even for people with disabilities or things like that, you should have transcripts, you should have clear texts. It's a bit like, um, a graphic design one hundred one. If we would talk to Jenna about it, it's just very important. So drop the cursive font and yeah, low contrast sort of text. Think about the way your speech is delivered as well. I know it's you feel already self-conscious when recording a video most of the time for a lot of people. But the thing is, um, make sure you articulate and make sure you sound authoritative because everything matters. Yeah, yeah. But at the same time, you must feel comfortable. So that's just. Yeah. Mhm. Um, and I wrote something. I'm not sure about it. Visual anchors will help AI better understand video topics and themes. I guess these are like illustrations in your videos, these sort of things. It will just help put context around what you say and I think I can relate that to the market effect we had that maybe I didn't put enough marketing, engineering, uh, B2B related sort of, uh, information, visually speaking, for it to bypass the fact that you slightly look like him. You know what I mean? So that's something I will do in the future to see how it, how it goes. But yeah. Uh, so yeah, so basically not all AI models can watch videos yet. Uh, so yeah, transcript and all of that are still essential. And, uh, you still need to find out where your audience is and what they like to watch while keeping an eye on optimising video for AI powered search. So that's something we mentioned at the beginning as well. Yeah. Um, so that's why the idea to build a constellation of content and work from there, I think it's a win win strategy because it allows you to, I don't know, hook your audience with short form content and then link them to your longer form videos, which are something very nice to have for AI powered search engine to crawl these sort of things. So it's just about enriching what you're telling around your brand. Mhm. Because, uh, yeah, AI is just going to crawl through information that are available online and whether you want it or not, it's going to tell a story about your business. And if you're not here to help add nuance, as we say, and, uh, correct it and bring like trust, yeah, it will just hallucinating or just, you know, so that's just important.
To just, it makes it makes sense because if you, if you don't structure content properly, whether it's video content or whether it's written content or anything else, then the tools that are looking at your content, they're going to make their own mind up. A simple example, if you don't put a meta description on a web page that so that you control the little snippet of text which appears in the search result, if you don't put it in there, then Google will read the page and it'll decide what goes in there, which might be completely wrong. Yeah. So, you know, this idea of, of if you know, by following some simple rules, which hopefully we've highlighted and given people a steer on in this podcast, then the video content that you produce can be, you know, even more effective. It's already effective if it's in YouTube because it's a search engine, it's already effective if you're putting snippets of it on TikTok because there's an audience there who likes short form content. Um, but there's a, there's a, there's a whole new emerging audience, if you like, which are the AI bots which are actually looking at the video. And yeah, the amount of information that's in videos is, you know, it's huge. You know, picture paints a thousand words, as they used to say, you know, so, um, it's an opportunity that if you go about it in the right way and it doesn't seem to me to be an onerous task, we don't have to do anything particularly technically challenging or difficult. We just need to produce really good, well put together, well thought out, well, you know, considered content videos in this case.
And yeah, the production doesn't have to be expensive. No, it's going to be expensive if you don't do anything about it. I think because.
You could could be an opportunity missed.
Yeah. Yeah.
Okay.
That is.
It. Good. You've been listening to Digital Marketing From The Coalface. I think I said that, right? Uh, with me and Leslie. And, um, hopefully you've got some insights into video and AI and what you could be thinking about doing. Um, if you're not doing video content at all, then if you want to speak to us, speak to us. But I would certainly recommend that you do create video content. I think we should be doing more with video, you know, evolution, we, we know, we do these videos of these podcasts and we capture video of other things. But as the weather starts to improve and it is now twenty six minutes past five and it's still a little bit light, um, as the weather starts to improve, I think we should have a programme of, um, yeah, lots of, lots of video this year, lots of useful video this year.
Absolutely.
Okay.
Bye bye.