B (4:02)

But and it's the context piece that's really interesting to me because the signals they're talking about, the more you read into it, it's about the sequence of things that you maybe have searched for previously.

A (4:12)

Yes.

B (4:13)

And where you might end up.

A (4:14)

Absolutely. So, so a lot of their like performance Max, for example did started the ball rolling on this. So you know that was able to look at a lot of context across a lot of different Google properties at lots of different stages of the funnel. Right. To like get a picture of, well, you know, from what we know about the people using Google and what does the data say about who's interested in what and who might be enticed into something completely new that they haven't perhaps thought of before. So you know, it was a fantastic, it was the first time we'd ever had, you know, one campaign type that could do like full funnel stuff.

B (4:48)

Right.

A (4:49)

And anybody that's, you know, seriously pursued this has, you know, realized fantastic gains from it. It was, it's brilliant. It gets harder as everybody jumps on that bandwagon because suddenly you don't have that competitive edge. Your competitors weren't using it. I think probably pretty much everybody's probably using Performance Max now. Some better than others, has to be said. But the thing with this is there's a whole bunch of new signals I think Google have now got. So, you know, people, as we know for a lot of searches, if they're, particularly if there's like an AI answer, like additional summary boxes, you know, a lot of publishers have seen a massive drop in the number of organic clicks that they've been getting. And Daniel, you were saying, even at Target Internet we've seen some big shifts, like big improvements in, in like impression rates for sure.

B (5:35)

But the problem is interesting.

A (5:36)

No one's scrolling down to where the organic listings are because they are, they're getting the answer up top.

B (5:42)

Yeah. So just to give people some context, within search console you can see how many search impressions you got and you can see how many clicks that you've got. And our impressions are an all time high. We continue to publish content, we continue to optimize it, where our rankings are getting better all the time. But actually our number of clicks is not an all time low at all because it's, we've been going up gradually over the years, but it's, it has dipped in the last six to 12 months and that, that percentage ratio between them is different. And you're absolutely right because there are ads at the top of the page, there's AI overs top of the page, there's all those, you know, different results. Like we're seeing Reddit show up more, we're seeing more videos show up because they seem to be more human based. We're just not getting the clicks in the same way.

A (6:20)

You see, Google aren't getting the clicks in the same way either. Right, right. So that creates a problem. So previously they would have been able to see, you know, who's clicking on which organic links and that, you know, for their system that might put you in the running for certain ads. Right, that would make sense. Now they're not getting that data either. But what they are getting is a ton of data on who's asking AI what, you know, who's looking at which AI. Summaries, who's dwelling on a particular summary and then what's the follow on question?

B (6:46)

Right, that's AI.

A (6:47)

You can use AI to string together lots and lots of signals like that for every single individual and produce what I'm assuming is what they're referring to by the keywordless signals.

B (6:58)

Right. And I think that's the whole thing is that you might show up, you get really good results with four or five search terms. You're not getting a click, but you might get a six step because it's a sequence of kind of deduction that people are going through to some extent. So it's really interesting.

A (7:09)

Well, it's even more than that in a way, which is think about it because actually somebody might just be researching, I don't know, they might be researching baby names for example, right. And that combined with lots of other behavior, what videos are they watching on YouTube, you know, what comments are they making in the comments section? All these things are all, what emails are they getting from who on Gmail? You know, if you've got the right permit, given the right permissions, it's able to use that. If you've got Chrome, what other websites that are looking, it's all being fed into the mix. And actually when you start to join the dots for a whole bunch of different signals in this way and you start to train your AI to make use of this, of course there's a ton of value that you can extract for that in terms of, you know, who would be receptive to, to which kind of advertisers and what kind of message. I think that's, I mean it's mind blowingly complex. That's the first thing I would say. This is increasingly, I find it very difficult to get my head around. Okay, how do you unpick this, how's this actually working? But in a way you don't need to, Right?

B (8:06)

Well, that's one of the things I was going to say because increasingly you've got a couple of, there's three kind of main slider buttons and it's. You slide one to say, well yeah, expand my keywords to keyword list signals and things like that. And then you talk about creative assets. Yes, you can, you can create new creative assets and then it goes through and says, and actually can we use different landing pages? We're going to select the most appropriate landing page. Now my terror with that is historical, which is looking at expensive word campaigns where basically what would happen is that it would start, find, oh well, you're going to rank well for this keyword. And I'm going to send you through to this landing page. And the reality is I'm already ranking for that organically, and it's not a term that's going to convert at a high rate. Actually. These campaigns are only optimized for achieving your desired outcome.

A (8:50)

Yeah.

B (8:51)

So they're not going to do the things it used to do when it was a little bit more basic. From my insight so far, they've also.

A (8:56)

Put a lot more control in than we had initially when Performance Max was released. So, for example, you do have the ability of turning off the. Well, let Google decide what landing pages it uses. This is the challenge here. Right. For years, everybody's been optimizing all of their websites and content for the same two to three word phrases. And now suddenly, genie's out of the bottle and no one's using those two, three word searches in the same way anymore. For sure, they're still using them. But a lot of people are, you know, skipping a whole section of the discovery phase that they would previously have done within search and through clicking on your organic links and stuff. And now they're just jumping straight to the answer. And so we're moving from a situation where there was space for, you know, three or four ads and, you know, half a dozen organic links to a place where, frankly, that's not the case anymore. And they're still there, but no one's clicking them. Like, it reminded me. I was chatting to somebody about it today. It reminds me of, you remember back in, probably go back to 2011, 2012, when Google had a whole bunch of not so hot ads down the right hand side, like down the bottom like 15, and at the bottom of the page, like 15 column, thin, skinny column of 15 blue links. And it was funny, it was reserved for the ads that weren't really quite as relevant, but we'll show them anyway. Yeah, just give them a chance. And they were terribly low quality and nobody got very good results from them. And there was outrage when people removed them because people thought, well, now I'm squeezed into fighting for, you know, top three position. Well, yeah, but actually, when we looked at the data, everyone ignored the long, thin, skinny column because it was terrible. Like, it wasn't really very targeted. Like, this is similar in a way. Right. We're going, we're now moving to a place where, like, just one thing, it's going to get the edge. And maybe a few things might get sort of submentions if you look at the sources, but actually yeah, the game has changed massively and what I'm excited about is AI Max, I think gives us quite early on tools to start chewing into that. You know, that's the exciting thing. I think also though, with that we undoubtedly we've lost a huge amount of perceived control that we had perhaps once I had over our organic listings. Like I think people have been asking me what can we do to rescue our organic position. I'm like, well, winging a prayer, but nothing really because, well, in a moment.

B (11:23)

Yeah, we'll talk about that organic piece, but on a moment because it's going to come up a load in a few more of the tools we talk about as well. Before we get into some of that heavier stuff, let's go to a fun talk. This is when you pointed out to me this is something in Google Labs. So if you're not familiar, it's Labs Google and it's just called Gen type. Well, again, as ever, everything's in the show. Notes, target, Internet.com, forward slash, podcast. So but you explain it because you showed this to me and I thought this was. It's actually a lot of fun and it could. It's quite useful as well.

A (11:49)

So you, you can create a full on like graphical Alphabet.

B (11:54)

A font.

A (11:55)

A font, yeah, a font set just by giving it a prompt. Right. So I got hold of this and I did something fairly pedestrian and mundane. I thought that's quite fun. That handed it to my 15 year old daughter Izzy. I said is, look at this. And she was like, omg. The first thing she did was she instructed it to create an Alphabet made up of vomiting unicorns, but they had to vomit rainbows.

B (12:20)

I'm like, wow, it's a thing of beauty.

A (12:22)

I'm waiting for it to produce this thing and it's never going to cope with that.

B (12:24)

This is going in the show for sure as well.

A (12:26)

Yeah, it totally did it.

B (12:27)

But I think the key thing is it's a nice little use case of generative AI creating images but doing something that's kind of fun with it as well. And actually there's loads of useful cases for it as well. But it is a bit of fun. But there's some nice stuff in Google Labs that you can experiment with and go, actually I could see how it's going to be applied as well. So that's gen type. Take a look at that. Right, let's move on. LLM Refs is one of the many tools that are appearing to try and help you with search engine optimization. But from the generative Engine side of things. So the idea of how do you show up in the AI overviews, how do you show up in ChatGPT and so on. The nice thing about LLM refs is they have a very robust and solid free trial. And Kieran pointed this out to me. We've been playing around with it a fair bit. And rather than doing what HubSpot AI Grader does, which is HubSpot AI Grader, you put your brand in and it tells you how visible your brand is. It thinks in a number of different large language models. This, you put a search term in and it tells you the brands that are doing well and it shows you how they're kind of branded. It gives you some scores around those kind of things as well. So as with any of these tools, I think there's two main issues. One, it starts to make you realize how complicated, really unpicking this Generative Engine Optimization piece is, how you show up in a large language model, how many things there are to think about. And we'll come to that point in a moment because Kieran was talking to me about that. But also, where's the testing to prove this? Where's the data to back this up? Where is the definitive case? Oh, this is what it is now. It's going to be fixed like this for a period of time. I think there's a lot of people just testing, trying, experimenting, coming up with tools. This is a great example of a tool, but I don't think there is a definitive answer to these questions yet. So that, and that's part of the game. It always has been. SEO is trying to understand it and game it and get there before other people and so on as well. But it is complicated. And I think that actually search has been more complicated than people have realized for a while because a lot of this stuff was going on in the background with the Google algorithm when they started to bring AI into the Google algorithm anyway. So really this idea of understanding SEO, Generative Engine Optimization, if it's the same thing or something separate, has actually got wildly more complicated and there's just so much more to do that it becomes slightly overwhelming. And therefore you're going to need help with this.

A (14:50)

It's every time this iterates, it becomes exponentially more complicated. Right, so what do I mean by that? Well, let's just go on a journey through time, shall we? Back in the day, there was Google and there was one like search Index. And then a few years later we had, you know, mobile search results were sometimes slightly different so that complicated things and then there was this massive revolution that sort of happened silently where actually there was no such thing as being number one in Google. You had to ask number one for who and then number one for who and where and on what device, right. And in what context. Right? So each time this happens, you're massively adding to the number of variables and it's led to like a running joke. Actually, I was at this year's Brian SEO, the running joke was, you know, when you ask an SEO any question, they say, you know, the answer is always, ah, well, it's complicated, right?

B (15:48)

Because it is.

A (15:49)

And it's getting more complicated by the day. So this is the thing, I think the I'm really quite sick of hearing senior marketers, you know, marketing directors, you know who you are, you'll recognize the people that have wound me up. You'll know who I'm having to go out, right? Like they go off, they look at some webinar that's shoved at them for free on LinkedIn and they come back thinking they've got all the answers because they've found this expert who thinks that, you know, this is real. And they announced to their team that we need to get on top of AI search. And like just because they've identified that there's an issue there, they've solved it, they've staked the ground and they now own this. Right? And it's just really annoying. So if you're in this caught in these headlights right now, I'm sure a lot of you are. Actually, the llmrefs.com demo will help you to explain to these seniors. Serial summarizers, we'll call them. That's what they do.

B (16:48)

They're like, ah, blatantly, I like this serial summarizes.

A (16:51)

Yeah, they're blatantly simplifying ridiculously complicated things, right? Because you can show them with this. Actually let's just, with using the free version, let's just run the numbers on one key phrase. Like pick one really popular key phrase and what it'll do is it'll ping out how like the top like 30 like references in not just one, but like almost a dozen different large language learning models and clawed and touchy. There's a little top right hand dropdown you want to look at, right? Because I hadn't spotted it initially, you sort of miss it in the interface. But actually you can choose the results that you're getting. Like, well, what do you want to look at? Do you want to look at Claude or chatgpt or do you want to look at, you know, Gemini AI or Google Answers or you know, what element you want to include within this summary and it calculates it all out for you and it is just mind boggling. And that's just for one phrase. So something we've gone from like one source to like like almost a dozen just in one foul leap. And the dust really hasn't settled on which of these are going to be most influential. No one knows which is going to be the most influential large language learning model in any one space this time next year or even a Christmas like Christmas this year. No one really knows where that's going to go or how that's going to roll out. Remember there's a massive arms race going on with the functionality that's available to various different people at the moment say so for that reason I think LL Refs deserves a very cool mention. You get a ton of value out.

B (18:16)

Of it if you the free subscribe.

A (18:18)

Yeah but if you subscribe you can do a lot more. Like it is very much locked down. But I think just looking at that demo, it's so easy to access. You just connect it to a Google account and away you go. Yeah.

B (18:29)

They've upset me with one thing though.

A (18:30)

Oh, they did upset you, didn't they?

B (18:31)

Yeah, quite, quite drastically so on. On one of their kind of blog type pages and we're just having a little browse around as we do make sure we kind of think the tool's okay and the headline said and I'm not going to go into this because I'm going to get dragged down a rabbit hole and I don't want to. He said meta descriptions are the new H1. Oh, and I then lost the will to live.

A (18:49)

Did that upset you, Daniel?

B (18:51)

Just because this idea, you know how long I spent, I reckon I spent eight years telling people meta descriptions had no impact on Google. Right. Other than it would show up and it might get a higher click through rate. But it wasn't directly impacting the algorithm. And I remember repeating things and then they took is that the same as meta keywords? And I was like no, that's a different thing completely. But your H1's really important. And, and now we're saying things like this is the same as this to what your point was earlier on. Like someone will hear something at a conference.

A (19:15)

Yeah.

B (19:15)

And go that's the gospel. Right. And it is that a little knowledge has always been a dangerous thing. I, I work in lots of organizations, worked in academia a lot to Big universities. And invariably a professor would come back, go, why aren't we ranking for this search term? And it would just be because it's not relevant, because it's just not in the user journey. And it was just one of those kind of classic.

A (19:33)

Well, my favorite because according to Google, only 30 people a month search for that. Like, you know.

B (19:38)

Right, exactly. It's like no one's searching for it. We're getting 10,000 visits a month on that search term. So I'm not sure that's true. But yeah. Anyway, let's move swiftly on LLM refs. If you're listening, you can redeem yourself by leaving us a comment of what you were saying with that and we will get into it. Right. Actually on this, I thought this is quite interesting. Google have published some advice, AI experience advice. And it's like you get the lovely Google documentation that says this is how search works and this is what you should do. And it gives you the really kind of straightforward stuff. They've actually written some really nice stuff about succeeding in AI search and it talks about those fundamentals of, you know, the good quality, content, understanding, user intent, knowledge graph, connections between things. So we'll put the link into the show notes because it's developer.google.com understand their search section. But this is something you highlighted for me. And it's not to say this is a completely robust tool, but it did get me thinking on the lines of actually, search has been more complicated than we've realized for a while, which was the Knowledge Graph explorer. So let's take a step back. The Knowledge Graph is the idea that Google works on entities, that is the connections between different things. It will know the target Internet as a company. It will know the digital marketing podcast is a podcast. Kieran's a person, Daniel's a person, and it'll kind of know the connections between those things. And that's how it's working out to some extent. Connections, the kind of semantic connections, the context, but also the validity and the kind of authenticity and authority of those things as well. So it's a very interesting kind of piece. But the knowledge graphics blur this thing, you can put something in and it will tell you the things it's connected to. Now, it's reasonably limited, but it does show you that, that you can kind of play around with this knowledge graph a little bit. It was interesting that I put the company name in and all it came up with was my name and our managing director's name. And then when I had, it had lots and Lots of different Daniel Roses. So I clicked through to some of those. Some of me as an author, others were me as a movie star. I was quite interested.

A (21:33)

You apparently were a movie star in a black and white movie back in the 1930s.

B (21:37)

I've aged very well, I'd like to say. Yeah, so. So it's very imperfect, but what it does get you thinking about is that connection between things is really important. And that's what these large language models, and that's what Google are trying to work out, is authority and authenticity and all those things that you've got in E experience, expertise and trust and so on. So really all we're getting to is SEO is dead again. And I'm sorry to go down this path again, but it's just like the more complex it gets, the less I'm going to think about it because it's just a bit overwhelming.

A (22:09)

What. It is overwhelming. But I think what was really interesting about that example you just gave was that clearly it had got it wrong. And this is what you all need to be doing is looking at tools like this and going, well, what has this got wrong about us, our organization?

B (22:23)

Exactly it.

A (22:24)

And so I've been using this and working out, well, why doesn't it know that we're a member of, you know, this big important thing that we're a member of?

B (22:32)

Yeah.

A (22:32)

Oh, because actually we haven't made it obvious on our website. Like, we've looked, you know, maybe we've linked to, you know, just the organization's logo, but we haven't like put the hyperlink in there or, you know, little things like what's on our about page, do we mention it there? Like, you need to make this stuff really clear and then the AI will trip over it and go, ah, okay, I understand the context. Like, it's smart, but it's not clairvoyant. Right. You have to create links, you have to create.

B (22:58)

That's when you. It's smart, but it's not clairvoyant.

A (23:00)

I like that.

B (23:01)

I think that it just ends up basically being so complicated because saying what shows up in organic search, what shows up in these different LLMs, what's being referenced in each of those LLM results, what are the AI? And you kind of get overwhelming and actually these fundamental building blocks of understanding user intent, doing your structured schema markup, and actually we will do an episode entirely on this because I think it's such an important topic, so we'll do something on that. So actually within the code saying this is an article this is an faq, those kind of things as well. Just making connections between things. Because the knowledge graph and so on, all of this is building up that overarching picture as well. Advocacy, really important, people saying nice things, leaving reviews, all those kind of things as well. So I do think, yes, there are new tools. It's changing very rapidly, but still coming back some of those fundamentals. Right. Another one, somebody did get in contact, because we're going to give the opportunity to get in contact somebody. But somebody that did get in contact was Ad Zoomer. And what Ad Zoomer allows you to do is to connect up all of your ad platforms into one place and manage them from one place, wherever you're an individual organization or an agency as well.

A (24:11)

I've used Adzuma. It's good.

B (24:13)

It's a great tool.

A (24:14)

It's a great tool. It does a lot of stuff. Like it looks around a lot of corners for you. And I like the fact that it, you know, like all of your ad platforms, it covers a lot of them. So, yeah, check it out. It's good.

B (24:24)

Yeah. It brings it into one place. And it will also Google Ads, LinkedIn Ads, but also it'll bring in your Google Analytics and those kind of things as well. And they've given us a discount code as well. So if you want to take a look at it, there is a free trial. But if you do sign up at a discount. Targetinternet.com forward slash, podcast.

A (24:39)

I've got an extra tool I want to squeeze in because I got very excited. This. I've only just started playing around with it, but it's a tool by Google and it's called NotebookLM.

B (24:48)

I mentioned this a few times, but I was gonna say, I don't think you'd play with it.

A (24:51)

No, I've been playing around with it.

B (24:53)

It's fantastic.

A (24:53)

What I love about it is you can upload your own data into it and then it becomes a language model, just purely focused on what you've given it. And I think that's massively exciting.

B (25:03)

One example of this, I was playing this the other day and I went through and uploaded five podcasts. It will transcribe them for you and then you can query them as a whole. I was trying to understand how much we'd mentioned SEO, geo, LLMs and all those kind of things, but what was lovely is it will create a mind map.

A (25:19)

Yeah.

B (25:19)

And actually in the last couple of podcasts, I've included the mind maps that it's generated back of it as well. Just so just give people get an idea of what we actually play with it. It's brilliant.

A (25:27)

It's just got so much potential. There's so many ways for learning.

B (25:30)

It's amazing.

A (25:31)

Well, well, it's got me thinking is maybe that's where we need to go with the complexity that we're facing.

B (25:37)

Well, we were talking about this, an organization I was in the other day about trusted sources and actually you can do this already with a decent prompt which you say, get me the latest digital marketing news. But only from well, how about you could create a version of Google or a large language model that just looked at the websites that you trusted and just said, I don't want to get it from everybody, I just want to look at these places. Now it might limit your viewpoint. You'll be really careful of the echo chamber type thing. That makes that even worse. Actually. This is kind of where it's going. If you look at Chat, GPT and Gemini, the fact they have connectors that allows you to connect your internal tools and things like that is definitely the direction of travel as well. And we will get into that whole agents piece again in future kind of episodes as well. I want to give a shout to one more tool which we'll put into the show notes before we finish up. And it's one that we've spoken about before, but our friends over at Nutshell. So if you're not familiar, Nutshell is a CRM system, but it was basically a bit of an outlier originally. It had a very specific use case and it's grown. And what they really pride themselves is the robustness of the tool and it is an absolutely excellent tool. But actually from a pricing point of view, it's much lower cost than a lot of the other CRMs that are out there. And it's very transparent on its pricing in terms of you did this package and then you've got 20,000 contacts, right? Here's your price. So I really do recommend it, but they have given us a discount code so 15% off when you sign up. If you are looking at CRMs, one take a look at it. And we are doing a big comparison of CRMs coming up as well. And Nutshell is coming out very well in our comparisons at the moment. If you have got a tool we haven't spoken about, please do submit it. We do take a look at them and thank you as ever for listening to the Digital Marketing podcast. For more episodes resources to leave a review or to get in contact, go to targetinternet.com podcast.