Register for Napier’s on-demand webinar and discover how generative AI is affecting marketers as it impacts SEO and search advertising tactics. We will cover:
- How Generative AI is changing search
- The impact on SEO
- How is search advertising changing?
- How might your website traffic change?
- Five things to do now to manage the impact of AI
Register to view our webinar on demand by clicking here, and why not get in touch to let us know if our insights helped you.
Napier Webinar: ‘The Impact of AI on SEO and Search Advertising’ Transcript
Speakers:Â Mike Maynard
Good afternoon everyone. Welcome to our latest webinar.
Today we’re going to be talking about the impact of AI on SEO and search advertising.
So particularly, we’re going to look at generative AI in search results pages.
So what I’m going to do is I’ll get started, and we’ll start talking about this important topic.
So why are we doing this? Well, it’s pretty simple. I mean, other than the opportunity for me to use AI to generate some crazy robot characters, really, it’s about stopping the robots eating your SEO and SEM lunch. So one of the big concerns that a lot of people have is as generative AI starts dominating the results on search, the question is, what’s going to happen to my website traffic? Am I going to lose all my traffic? Are people going to stop going to the site? Am I even going to appear? So this is a very valid concern, and it’s something we’ll talk about today without wishing to spoil the whole webinar and cut to the end, I will say this is a rapidly changing situation. So whilst I can talk about what’s happening today and what to do, there isn’t any firm answer as to what’s going to be working in a year’s time, because, frankly, I don’t think even Google knows what its search results page is going to look like in a year’s time.
Anyway, let’s have a look at what we’re going to talk about today.
So I’m going to start off by talking about how generative AI is changing search I suspect pretty much all of you understand that, but I think it’s important to get onto the same page. We look at the impact on SEO and also advertising. And I think advertising is something that is often underrated as being impacted by these generative AI search results.
We’ll look at what’s happening with website traffic and how it’s likely to change, particularly how it’s likely to change in the near future.
And then lastly, we’ve got five suggestions of things to do to manage the impact of AI.
So lets look at a typical search. So choose a microcontroller. So this is one for our semiconductor audience here.
So it’s quite an informational query, and you’ll notice that I’m tending to use informational queries. This is not because informational queries are the only things that produce generative AI results, but at the moment, you’re much more likely to get a generative AI result when you’re asking a question that needs to be answered with lots of information, rather than a simple search that is much more transactional.
So if I run this search, and this is what I did when preparing the webinar, this is what I get as the results, and I think it’s really interesting to understand what happens as we go down the page. Now this is particularly chosen as a page that doesn’t have a lot of ads against it. In fact, there’s no paid results on this particular results page. Doesn’t mean people aren’t advertising against terms like microcontroller. It just means I’ve managed to find a search where we didn’t get ads to confuse the situation.
So the first thing we’ll see is, right at the top, we’ve got the generative AI result. This is the text that Google’s AI Gemini has generated to try and answer the question.
It’s been generated using certain sources, and you can see here there’s references. So if you want to be in the top part of the page, you’ve got to be one of the websites that’s referenced by this generative AI result, and you’ve got to be referenced early on it in that generative article, because what we’ve seen.
Now is that Reddit is next. That’s not that unusual. Reddit is certainly seems to be the current website of choice for getting really good ranking. At the moment, not sure that’s always going to be the case. You get the people also asked, and then finally, you get the first website results. So this is, you know, right down at the bottom of the page. Now, I mean, almost off the page before we get a result from a website, from somebody that’s trying to promote or sell microcontrollers. In this case, it’s a distributor.
And so, as I mentioned, no adverts are shown. If they were we’d actually see everything pushed down.
Now what we can do is we can click on the Show More section of the AI overview. I’m sure you’ve seen this, and what we’ll see is we’ll see a much longer answer from Google, an answer that, in fact, is covering the whole page. So we’ve now wiped out all search results, and all we’ve got is the references. And you can see there’s a lot of references that are on this page, so it is quite important to remember that references do matter. That’s what we’re trying to achieve when we’re trying to deal with the impact. However, if you caught the previous top search result. It’s interesting that although tme was the top search result, it doesn’t get quoted in the references. So this is the first thing to recognize, is that actually the generative AI results and the pure search engine result pages results, they are going to be different, and you may need to do different things to appear in the generative AI summary than you do to appear in anything else. So in the results below it, it’s also important to notice that there’s not just these generative AI summaries. So there’s AI shopping as local results, travel results, informational results.
So AI is going to produce a lot of results on Google, and increasingly, it’s going to produce more and more results on Google. Google has actually experimented with 100% generative AI results pages. So it’s clear that Google is looking at what delivers the answer that the searchers want.
Now the big problem we have with this is that although there’s lots of references compared to a Google search results page, there’s actually not that many, because each of these references apply to a particular section of the of the generative result, and we’ve basically only got this one page, so although we might get 10 references, we haven’t got page two, which is 11 to 20, page three, etc. So it’s really important to recognize that this is very much a winner takes all game. If you get into generative search results, you are going to be beating a lot of your competitors.
But before we talk too much about how to get into those results, I think one of the important things to say is that appearing in Gen AI results, it’s what’s called Zero click SEO. So Rand Fishkin, well known in the essay SEO community, he came up with the term of zero click SEO. And here what you’re trying to do is get an impression. You’re trying to show yourself ahead of anybody else. So you’re trying to show yourself an ingenerative ai results. You’re not necessarily trying to get a click. Now, of course, we’d love it if people look at the references click through to our website and then read our website. That’s fantastic. But the reality is that’s not the behavior of most people who are doing searches. So most people will read the Gen AI result, and your goal is to get your opinion, your point of view, and also reference to your product into those results. It’s not necessarily going to drive traffic to your website, and that makes measurement really, really hard. You’re not going to necessarily see traffic in the same way as you did from your standard SEO campaigns.
However, it’s not completely new, and RAM fish can actually introduce the zero click SEO term back in 2019, so this is best part of six years ago, when some research he shared showed that actually the majority of Google searches don’t produce a click. So even today, you’re being shown in Google search results, and it’s more than likely there’s not going to be a click to anybody, not just to you, but to anybody, from that Google search result. So more and more we are going to have to think about generating SEO and measuring it in a different way, purely from traffic.
So I mentioned informational queries. And you know, one of the things that certainly is the case is that disproportionately.
Gen AI results appear for informational queries. They do for more transactional ones. So it’s okay, because AI is not going to tell you what to buy, or maybe they will. Let’s ask which high performance ARM based microcontroller offers the best price performance.
And here you can see AI is certainly not at all shy at telling you what to buy. AI is recommending the STM 32 h5 and the Renaissance ra series as being best price performance. Now this is interesting, because actually, if you go through and analyze this, it’s really not telling you why these two products here are actually the best price performance, and in fact, it’s been created much more from an SEO type approach, in terms of linking words around price performance on the same page next to these products than it is necessarily from a technical analysis. So it’s super important to remember that we’ve had two product families that have ranked in the Gen AI for the best price performance. But that’s not necessarily driven in any way by their price performance. It’s driven by what’s said on the page about price performance, and therefore Gen AI has linked these two products to good price performance.
So this is a real problem. I mean, it’s no different in many ways to when we looked at Search Engine Optimization, in the good old days, we didn’t have to worry about Gen AI, where if you did do this query, you would get different products being listed in an order that doesn’t necessarily represent their price performance. And here you can see, for example, the ST product that is mentioned is actually the number one search result. So again, this is a search engine optimization challenge. It’s not AI providing analysis that really tells you which is the best price performance products.
So what’s the problem? I mean, basically, there’s a ranking of results, and that’s being done by Google, but it’s also being done by these generative AI tools like Gemini.
That’s always been the case, that’s not changed. But the issue is these generative AI answers, they’re quite detailed, they provide a lot of information, and there’s much more of a risk that people won’t read further than the generative AI and you’re putting fewer options into that generative AI summary. So the generative AI result being in it matters a lot, because people are less likely to look at your position in the search results page.
And, you know, they’re much more likely to pick something from Gen AI.
Of course, some people use chat bots. And you know, just to let you know, here, I’ve done Google for I’ve used Google for the search and therefore Gemini and I’m using chat GPT as an example of a chat bot. What I’m saying here applies across all sorts of AI engines. So whether that’s Bing, which actually is fundamentally, you know, driven from chat GPT, GPT, whether it’s perplexity, which a lot of people are using for search, or whether it’s something like Claude. So all of what this what this webinar covers, applies across the range to different AI models.
So I’m going to ask chatgpt, which high performance ARM based microcontrollers offer the best price performance. Let’s try the same question. I But before we get carried away, let’s not think that everybody’s doing this. So I talked to oops, I talked to a lot of people who say, I use chat GPT. Everyone uses that. Actually, it’s not the case. Google Search grew 20% in 2024 and actually, Google Search receives almost 400 times more searches than chat GPT.
So if we look at what’s happening, yes, some people are querying chat bots, but actually not everybody is. So whilst we’re looking at this, let’s not get too freaked out by chat bots, because it currently it’s nowhere near the majority of traffic or the majority of questions on the internet. Okay. Now, with that caveat in place, let’s have a look at what chat GPT tells us about the best price performance micro controllers.
Well, st, Mike, electrons have done a very good job, because again, that lovely 32 h5 series is ranked as one of the best price performance, again, not through any analysis of price performance at all. This is based upon effectively linking words together, so a kind of SEO type approach. Interestingly, the Renaissance product didn’t make it onto the chat GPT summary.
We see a number of other products here that didn’t make it onto the Gemini summary. So there’s a definite lack of consistency. We’re actually seeing quite different results. The other thing as well, I’m sure you guys all know this, but if you don’t, it’s important to remember that actually there’s a level of randomness in terms of the results that any AI engine produces. So an amount of randomness is introduced. It stops the answers always being the same, and it means that sometimes an AI engine will recommend a different product, rather than always recommending the same one. So even though I’ve seen these through a couple of my searches, it’s very likely you’ll go and try and replicate this, and you may see different recommendations, either from Gemini or chat GPT, and almost certainly will see something slightly different from something like perplexity.
So it’s a similar sort of thing driven by, you know, the concept of content and people describing products as being certain things.
But the question is, now, you know, how important are these chat bots?
Well, there’s not much evidence. I mean, we can see here. So the Google trends analysis of my controller, it kind of backs up the research we looked at from Spark Toro that said that actually the number of searches is growing, so even in an engineering environment where people are probably leading edge much more willing to experiment with technology, we’re actually seeing Google, the legacy technology, growing and expanding market share.
So the good thing is, we don’t have to worry about this too much at the moment, but I do think we have to worry about the Gen AI results on the search engine results pages.
So let’s go back to search engine optimization.
You’ve got to appear in these Gen AI results. It’s really important, as I say, it’s very much a winner takes all situation. But you also need to appear in standard search results, partly because some people are going to scroll past the Gen AI. And recently, there’s been some research I saw which actually said that engineers are becoming less willing to trust AI generated results and much more keen on looking at results that are human generated. They trust them more. So you know, the standard search results still matter, and AI impacts ads too, but in a slightly different way. What AI is doing is pushing the ads down the page.
So here we go. We’ve got another query. We’re talking about variable speed motor drives, or variable frequency motor drives. We’ve done a search. The AI view sits at the top, and then the ads are below. Now this is going to have an impact, because previously, prior to Gen AI, those ads would sit right at the top of the results. So we are going to see lower click through rates for these ads than we would have seen previously without Gen AI, and that’s just highlighting it there.
And obviously, just as we do with organic search results, if we click the expand, the ads disappear off the bottom of the page. So again, you know, if people start engaging with these generative AI results, we’re going to see fewer and fewer clicks on our paid ads on Google as well.
So there’s less we can do about this, because we can’t get those paid ads at the moment into the Gen AI results, but it’s certainly something to consider going forward.
So what’s going to happen to your web traffic? Well, I mean, one of the concerns is all this generative AI that basically it’s going off the edge of a cliff pushed by the AI robots.
So that is a real concern, but I don’t think it’s really this bad. So let’s have a look at a different search. This is an interesting search. It’s about a topic called mental availability.
Napier actually ranks pretty consistently top for a search on mental availability.
As you can see here, though, we’ve been pushed out by the AI search result, although the a search result does reference Napier.
The good thing as well is, you know, we’re here number one in the organic search results, and we’ve actually beaten the erenberg bass Institute, which was really the people who originated the concept of mental availability. So it’s really good that we’re ranking high, and that has obviously helped us rank high in mental availability, but what’s this done to our traffic? So let’s have a look at how much traffic is driven to the Napier website based on the search for mental availability.
And here’s the traffic by month, looking back over almost a year.
And a half, and you can see it’s pretty random. It’s jumping up and down. You might be able to argue that there’s some seasonality there, but it’s really quite hard to see.
But what you see is that when Google started testing what the time they called the search generator experience, we had a low month, then we had a high month, then they released it from beta. We had a high month that month, but the following month was low, and then we had it in the UK, and it was low, but the next month was high. So it was all a mess. So what can we do? It’s very difficult to understand this. I mean, we want to beat the bots, but, you know, are there strategies we can adopt that are going to make sense. They’re not going to harm our organic search results, but they’re going to help us with generative. Ai, well, I think they, there are. I mean, the first thing I’d say is, really just keep calm. A lot of people have freaked out over Gen. AI, actually, the reality is, as we’ve seen with a lot of these results, if you’ve got good SEO today that’s going to impact your generative AI results. It’s going to help, because what you’re doing that appeals to Google in terms of ranking you on the organic search results pages is also going to be something that causes you to be ranked higher or more likely to be included rather in generative AI results.
So you can’t do a lot about Google ads. Your Google ads are probably going to reduce performance. The most important thing is to become about that and understand that as generative AI pushes Google Ads down, maybe you’ll see a slight drop in click through rate. Make sure you’re not making decisions about ads, assuming the ads are poor simply because they’re being pushed down by Gen AI. So really understand what’s going on and don’t panic.
Secondly, I would say it’s really important to understand whether you show up in AI results, and you can obviously do that to a large extent by looking at Gen AI search results pages and chat bots, understanding where you don’t show up and understanding why you don’t show up, so why others are preferred to you.
But you can also use tools, and there are a couple of tools shown here, so one is profound. The other is otterly. And we can do something like, you know, ask otterly what happens if we put this prompt into different AI engines, and otterly will give you a result. So it’s a paid service that will actually tell you whether you rank in Gen AI results for particular queries. So that is a really useful tool to understand whether you’re appearing in Gen AI, or whether you’re being excluded, there are lots of other tools as well, and a lot of the SEO tools are more and more beginning to add these Gen AI visibility tools. So it’s becoming a much more commonplace thing to use.
So those are two things we do to really, you know, firstly, establish there’s not a need to dramatically change what we’re doing. Secondly, we need to understand the situation.
The first real thing we could do is start writing answers to questions people ask AI. So I’m sure you’ve all experienced AI. AI is a very different experience, particularly the chat bots, where people try to engage and ask questions, it becomes quite conversational. And so the best approach that you know, people seem to believe today is to add FAQs or questions and answers onto a range of key pages on your website. And what that’s doing is it’s putting content that’s conversational, that can be very easily ingested by a an AI model, and you’re putting it on an important page, so therefore they’d be fairly highly ranked. You get some authority there, and hopefully also people will take them and maybe replicate some of those questions as well.
There is also the ability to do some technical optimization called schema markup, which is basically a tool to tell systems that you have a Q and A on your page or an FAQ on your page. It’s important to note that there’s no real evidence that any of the AI models are really looking at that schema markup today, but it’s generally felt that would be something that would help AI models get better answers for people, and it’s felt it’s really likely that more and more schema will be used by AI models today, it’s much more the content on the page, so that’s probably not something that’s going to make a huge difference to you the next thing is to appear on authoritative websites, and authority is important. It’s important for Google. We all know about that. We’ve all heard about authority and links and how they rank websites. But actually it’s important for chatgpt as well, and very crudely, if you want to.
On authoritative websites, news sites and directories tend to be ranked pretty highly. Even those directories you hate and you think are really low quality tend to get a reasonably high ranking. So making sure you’ve got a very solid PR program and you’re also actively listing yourself in directories is really important.
The other thing that does seem to rank very well, and there’s been lots of discussion around, you know, Reddit and Quora and other Q and A sites, over whether the AI companies can take their content or whether they need to pay for them. But it’s clear the AI content, the AI companies, so the open AIS and
other companies. They all value these Q and A websites. That’s very good content. It gets ingested well into the models, and it produces good answers. So they are, in some cases, paying, but they’re definitely looking at things like Reddit and Quora, and it’s important to have some presence on there as well.
And then lastly, and this is probably most important, is to have a Wikipedia strategy.
So you know this, this is really interesting, but without doubt, Wikipedia is something that is used by the Gen AI companies, and it’s used differently to the internet as a whole. So if you look at, you know how open a AI reported that they trained their models, they’ll talk about using Wikipedia and the internet. Well, Wikipedia is obviously part of the internet, so they’re separating out Wikipedia, and they’re actually making Wikipedia more important than the typical page on the web. So you really, really, really need to have a great Wikipedia strategy.
This is interesting. So we’re back on variable frequency drives, and this is a Wikipedia article about variable frequency drives, and the Wikipedia article talks about Stromberg as being the company that invented the technology. And if we go to Stromberg here, we can see that there’s an article that ultimately says Stromberg is part of ABB. So ABB ultimately acquired Stromberg. That’s one of the reasons why ABB is such a strong supplier of this technology. But you can see, it’s not a super highly ranked page. You know, it’s got some notifications that the page needs more references. So really, if you want to rank high and you want to link ABB closely to variable speed drives, there’s some things you can do in terms of tweaking Wikipedia to make Wikipedia reflect that ABB is the owner of Stromberg, put in more references, improve the ranking, and possibly put a mention of ABB onto the variable frequency drive page. Now, having said all that, it’s incredibly important to rank on Wikipedia and to have content on Wikipedia. It’s also incredibly easy to fall foul of the very arcane and obtuse Wikipedia rules. So if you don’t understand Wikipedia, please, please, please don’t sign up, get an account and start editing and particularly, don’t start editing content that directly relates to you. There are real strong rules and restrictions around Wikipedia. So please make sure you talk to someone who knows about Wikipedia and understands what can and can’t be done.
So thank you very much. If anyone has any questions you know, please feel free to put them into the chat.
I’ll just summarize. So the first thing I’d say is don’t panic, and hopefully there’s a few geeks on here that understand the link between Hitchhiker’s Guide to the Galaxy and don’t panic. I would say definitely keep an eye on the robots and monitor the impact of AI. Focus on quality. And I think this is really important. Increasingly, it’s clear that there are definite moves from the AI companies to make sure that they’re ingesting content that they view as high quality. Wikipedia is a great example of what they rank as very high quality. So focus on that. Remember, other websites are as important as your own when it comes to appearing on generative AI, this, again, is a zero click play for most of the time. So what you’re trying to do is you’re trying to get content about you into the generative AI answers. If that content comes from the Wall Street Journal rather than from your own website, that’s fine.
Definitely create content that helps the AI, you know, let’s try and work together produce content that’s going to rank very well. And finally, make sure you ask experts. And I think I will go back and talk about Wikipedia. It’s very easy to fall foul of the rules on Wikipedia, and you really don’t want to do that.
So thank you for listening.
Hope this has been interesting. I hope it’s been helpful. As I say, there’s no immediate easy answer as to, how do you change your strategy for optimizing for generative AI results, but there’s a lot you can do to give yourself a much better chance of appearing in those results, and certainly, if you’re monitoring what you do, you can really learn as to what works for your brand.
We obviously keep these webinars going as a fairly frequent sequence.
The next one is going to be at the end of April, Tuesday, the 29th of April, same time,
although probably one hour different for people in the US. And what we’re going to do is we’re actually going to demonstrate some integrations of AI into marketing campaigns. So rather than talk on a more theoretical level, this is going to be very practical, and we’re going to show how AI can be used to really help and improve your campaigns. So I strongly recommend attending this. I think it’ll be a fun webinar, and hopefully we can potentially do some live demos, and they won’t go horribly wrong.
Okay, I hope you found this useful.
What I’m going to do now is switch over to questions. So if you have any questions, feel free to post questions into the chat or into the Q and A, and I’ll just give Everyone a couple of minutes just to post questions you.
I’ve got a good question here from someone, and they’re asking, is there a way that you can understand the algorithm used to decide which content is put into generative AI results, and that’s a great question.
The answer is very simple. There’s actually not an algorithm as such to decide that. So if we look at Google, when it ranks pages, Google has a definitive algorithm and that’s related to all sorts of things. It’s related to the content, it’s related to the number of links, it’s related to the freshness of the content, all sorts of things will cause your page to be ranked higher when we look at generative AI what we’re doing is we’re training an AI model. And an AI model fundamentally works by effectively deciding what’s the most likely next word. And anyone who’s an engineer, please, don’t freak out, because that is a gross oversimplification. So it builds up and effectively compresses as much data as it can into its model, and then it tries to work out the next word, so what’s most likely. And so there’s not really an algorithm. And in fact, you know, understanding how AIS work and why they produce certain results. It’s a subject of massive research. People have not got to that point anyway. So there really is no magic answer to this. There really is nothing that says if you do this, then you’ll appear in generative AI results. So it is a case of doing all the right things and then, to some extent, hoping that generative AI will produce you as the answer, as the reference, and as I say, also, generative AI does introduce some randomness. So if you ask AI the same question, it will quite often produce a slightly different answer, sometimes choose a very different answer.
And that means that even from answer to answer, you can’t predict whether you’re going to be in there. So it is a case of the fact that there’s an element of randomness, and that is something that we can’t overcome. So today, it’s very much a case of trial and error, and that’s why monitoring is so so important. If you put some effort into trying to appear into generative AI search results, it’s then very important to measure to see whether that had any impact or not, so the monitoring tools are key.
Okay, I haven’t seen any other questions, so hopefully this means that the webinar was clear and you all enjoyed it. If you get a chance when you finish, please do complete the feedback form, and if anyone has any further questions or would like to talk about how you know their organization can appear in Gen AI search results, please feel free to contact me. My contact details are there. You can email me directly. Mike at napierbe to be.com thank you very much for attending. I hope you found it useful, and I hope to see you on the 29th of April for our next webinar. Bye.
