Latest / The Search Signal / Which AI Search Platform Should I Prioritize?

Transcript

Which AI Search Platform Should I Prioritize? · The Search Signal

• michael: Hey, welcome back to the search signal. I am Michael Transon. I'm the founder of a search marketing agency called Victorious. The search signal is how we help marketers understand and take action on what is happening in the world of search marketing. So as a team that works on hundreds of websites annually, we take what we learn from running search campaigns for some of the biggest and also some of the fastest growing brands and bring it. To these conversations so you can apply it to your own work. So before we get into today's topic, I do want to spend a little bit of time bringing you up to speed on what is happening this week in the world of search marketing. So a couple of episodes back, I told you about a new search console feature called platform properties. And this was right when Google had first announced it. this week, they rolled it out worldwide. And we know a lot more about it right now. So let me bring you up to speed on what it is. And a quick catch-up. If you had missed the initial announcement, so platform properties is a property type that verifies your social accounts like your Instagram and your TikTok, your X and your YouTube accounts, and then reports how those posts perform in three areas of Google, Google search. Google Discovery and Google News. And in addition to taking the rollout worldwide, Google also shipped a companion guide on how to use this new data to help you and me market our brands better. And what it shows is how to spot search terms people are using that are related to your content on those platforms and also how to compare your formats against each other, such as your YouTube shorts against your. full length videos or your Instagram reels against your regular Instagram posts. it also walks through how to track the same piece of content across those platforms to see where your content is being seen the most and also reaching the most users. And then it also allows you to check whether a title or a caption rewrite changed how people ultimately found you. So the link to that full guide. Is going to be in the search notes. But for at most companies, social performance has historically been siloed into each of the platforms native analytics, which has meant if you want to see how your brand is performing in Instagram versus Google search, you needed to log into that platform to see that specific information. And as a result, the two teams that own those distinct platforms. Never really put them next to each other. So this is the first time your social contents search and discover performance shows up in the same tool as your website. So that is the biggest news from this week in search. With that being said, let's get into today's episode. Somewhere in the last two years, AI Search became a very significant new marketing channel that we needed to track and optimize for. And this channel did not show up one platform at a time. We had ChatGPT and Claude and Gemini and Perplexity and Google's own AI services, they all arrived practically on top of each other and in just a few months' time. And when a category, a whole category shows up that fast, the natural reaction is to treat it as one unified channel. And it makes sense when that happens to put one strategy together and hope that it all works out across all of them all at the same time. And that is mostly what has happened. Across our industry, we see marketers treating these platforms as if they're all the same thing. And the fact is They are not. And as this channel is going to continue to mature, we're understanding more and more just about how different they really are and how these differences should inform the way that you are optimizing for each of them. And that is what we're going to get into today. So we're going to start with what these AI search platforms have in common. And then we're going to talk about how they are. Different because they are so different. And one playbook honestly cannot cover all of them. I also want to talk about why it is just so hard to know for sure which platforms your buyers are using and also what tools are available right now to help us get an answer to that question. And then with all of that outlined, I'm going to then break down how to resource a more nuanced. approach to AI search that helps prioritize which platforms you should focus on based off of, you know, how you choose to put budget to AI search at your company. And by the end of today's show, I want you to have the information that you need to be able to set a platform specific strategy in motion that makes the most sense for your specific business. So I want to start by saying that AI itself is not a monolith. And not all of these platforms operate the same. I think a lot of us have approached it as if it's all this this one thing. And that's, in my opinion, partly because this is the first time so many search surfaces have launched all at once. We got used to having one major player in the search scene and maybe like a few others. scattered and options that really were not worth paying attention to. You think about it, for the last two or so decades, the usage gap between Google and whomever came next was wide enough that a dedicated Yahoo or Bing strategy just really never made a lot of sense. And that divide has just persisted, right? Today Google is still Responsible for over 90% of searches worldwide. And Bing, you know, the second most popular search engine sits at like 5%. And because of this huge gap, marketers have typically run one search marketing strategy pointed at one search surface. And that was a very appropriate read of that situation until AI search started to emerge. And AI search platforms have been so disruptive to buyer behavior that none of us could reasonably keep operating as if there's only one search experience for us to optimize for. SimilarWeb estimates that chatgpt.com has had over five and a half billion visits in just the last three months. And that's just one platform, right? Up to now. Nothing has challenged Google's search dominance like that. And the fact that all of these platforms, ChatGPT or Cloud, or Gemini, or Perplexity, and even Google's own AI services, all emerged in rapid succession and with such fast user adoption, it's forced us all to acknowledge that we just can't continue to optimize for traditional SEO on Google. And expect to compete well in our industry. And because they did all emerge around the same time, marketers responded quickly and in the only reasonable way that they could. They kept the SEO strategy that they had and then folded AI search strategies alongside of it. And every new AI platform that launched. Just went into that AEO bucket with one platform behind it. But as the platforms have evolved and as we've expanded our understanding of how they work, there is now enough data to start pulling apart the idea that we can optimize for all of these AI search platforms the same exact way. So let's start with the Two things that all of these AI search tools do have in common. And they are very, very basic. And the first one has to do with how your own website serves information. And then the other has to do with what other websites say about you. And the first one is purely mechanical, and it comes down to how the pages on your website get built. So I'd like you to think about how your pages on your website get presented to a visitor. Like ordering a meal at a restaurant. Some websites hand over a finished and composed plate of food. And then there are others that hand over the ingredients and a recipe. And then your browser on your computer cooks it for you and does it in a fraction of a second. And as a human, you would never know the difference. Either way, you're just looking at a normal website page. Crawlers for AI search tools like ChatGPT, right? They don't cook. They take whatever they are handed. And if that is a pile of ingredients, they walk away thinking that there's really nothing there. The recipe version is something called client-side rendering. And most AI crawlers just can't read it. That includes the ones behind ChatGPT and Claude and perplexity. the big exception is Gemini, which actually handles it fine because it rides on Google's own crawler infrastructure, which is already built to do that. So you don't need to go any deeper than that. just know that if the content that matters on your website only appears after a browser assembles it, most AI crawlers will not be able to see it. The second thing that all AI search platforms have in common is a reliance on third party mentions. So that's when another page names your brand, whether it links to your website or doesn't link to your website. our own independent research at Victorious found that as a brand's third party mentions increased, so did its mentions in AI answers. And I walked through this and the numbers a couple of episodes. Back in the search signal, but you get these two pieces right. And you have what I would call baseline visibility on all of the platforms. It's more of like a rising tide lifts all boats. But let's now talk about how these platforms are different and why that is. And so to start, each one of these platforms was trained on a different data set. So their baseline knowledge is very unique. Right. They also supplement that baseline data with live web searches in very, very different ways. So when they go out and look something up, they are pulling from different sources. plus, on top of that, they're triggered to run a live search by different kinds of prompts. So some search for almost every question that gets asked, like perplexity does this, and then some search a lot more selectively, like Claude. so even if you run the same exact question or query on each of the platforms, they're not going to go looking for answers in the same places. And there ends up being very oftentimes little overlap in the sources that they reference. And even Google's own AI products are vastly different from each other. there's a study that we have in the show notes where five answer engines responded to prompts across Several different industries. And they found that Gemini's most cited sources overlap more with chat GPTs than with Google's AI mode or Google's AI overviews. So you've literally got a Google product whose citations look more like one of its competitors than like Google's own search products. And what that does is tells us something very practical. Showing up in AI overviews and AI mode does not mean you're going to show up in the Gemini app, even though it is all Google. And then on top of that, each platform has its own unique little tendencies, right? Claude, for example, favors a lot of established sources. and that means that the pages a brand would publish itself and coverage from like professional media. There's a study that looked at this, 27 million or so odd citations across six different AI platforms. And with Claude, roughly two-thirds went to pages that the brands own themselves. And then most of the rest went to media coverage. Social sites like Reddit, for example, were barely registered in the study. So on Claude, for example, your own content and coverage in Trusted publications are more likely to get you cited, and presence in Reddit threads, for example, probably won't. And their behavior also evolves in different ways and also on different timelines. So case in point, perplexities citation mix is constantly changing. earlier this year, our own research found it favoring. News outlets and specialist publishers. And then just a couple of months later, there was another separate study that found YouTube had become Perplexity's most cited source. And there was this old assumption that Perplexity favored Reddit, which is also out of date. And Reddit's share of citations have been dropping off of a cliff even just a few months back from now. So if you optimize For what perplexity references today, you should know that those sources might keep shifting as that platform continually learns and updates. And that's likely just as true for the rest of these platforms. And then there's also on top of this variants within each of these platforms as well. So remember, all of them were built as prediction engines, right? So the same LLM. if you ask it the same question twice, you are likely to not get the same answer. Right. there is a research team that tested this by running over 70 or so odd prompts, and they ran them over and over and over across four different platforms. And on questions where someone is still weighing their options, the mix of brands that were named stayed about 60% consistent from one run to the The next run, but on questions where someone was actually ready to buy, that consistency dropped to about 40%. So even on a single platform, right, the closer the question gets to a purchase, the less consistent the answers ultimately get. And point being in this whole thing, there's foundational work that pays off across all of these platforms. And then there's also more nuanced work that has to be tuned for each one of them. And there's a lot more research behind all this than I can possibly share in just one episode. So we're gonna link to a bunch of the research and data in the show notes if you do want to investigate this any further on your own. Okay, so we've established that these platforms are too different right now to treat them as one thing. as a result, your instinct might be to just pick the platforms where your customers are and then optimize for those ones. And I'll tell you, you definitely should consider doing that, but it's also harder than you might think. and definitely hard to make a data backed decision about where your specific customers. For your business are searching because at the end of the day, your AI visibility metrics, the ones that you track, show where your brand is already visible. And that's either by design or by accident. And that's not the same as knowing where your buyers are asking those questions. The tools that track your share of voice in AI answers have the same limitation. Both of them will describe where you are already appearing. And they will not describe where your ICP is looking for businesses like yours. And on top of that, they're lagging indicators, right? They can only report on what has already happened. So they are always going to be a step behind from where your buyers are going to be headed to next. So the next logical place to look for as you're trying to grab data on which platform your customers use the most. You could be like published research, for example. And that research exists, but most of it is in the form of survey responses. And an inherent limitation of surveys is that the responses are going to be shaped by the questions that were asked, right? So for example, Adobe had asked shoppers whether they had used AI for online shopping, and about four in ten said yes. Then McKinsey Asked people who were already using AI tools what they use them for. And only about one in five picked brand or product discovery from a long list of other options. And then a third survey asked whether people use AI for local business recommendations, and about half had said yes. So that is three credible surveys and three very different numbers. And it's because each one of them asked a different question. And, you know, none of it is broken out in a way that you can extrapolate anything about the platforms that your customers use. So you can't measure your way to this decision. You're going to have to make a judgment call. And because the platforms are different, and because no data source is going to tell you which ones your buyers use, you have to make a resourcing decision. About how granular to make your AI visibility strategy. And the way I think about it, you've really got three options. Option one is a generic AI search strategy. And you don't distinguish between the platforms at all. You go broad instead of going deep. And in practice, that is going to mean executing the foundational activities very, very well. So this is making sure that your It's making sure that AI crawlers can effectively read your pages. it's going to mean focusing on third-party coverage and reviews that are going to earn web mentions. Now, this takes the least amount of effort because this is work that's already underway, probably for your own SEO strategy, right? You got one team, you've got one work stream, you don't need to optimize for any individual platforms, but what you give is the ability to focus on any single platform clearly. And you'll have one blended AI search strategy. And if you're invisible on the one platform that your buyers happen to use more than others, your strategy is just not built to catch it. But then option two is picking two or three platforms to really go deep on and accepting that the others are going to get like crossover benefit from some of the foundational work I outlined in option one. And going deep adds two layers of work to the platform. The first is platform-specific work itself, which is going to be different work for each of the platforms, right? Getting cited on Claude, for example, would focus on citations in brand websites and in more established publications versus getting cited on perplexity would focus on YouTube videos and coverage and news outlets. The second part of that that we need to think about is also our reporting. We have to be able to segment and report on each of the platforms uniquely. And then you have option three. And this is going deep everywhere with a dedicated Strategy for each of the LLMs and for Google search as a surface in its, you know, in its own right. And that multiplies the, I would say, the per platform effort across, you know, five or six platforms at once. And this is the most resource intensive, but it will provide, in my opinion, the best outcome. So when you look at these three options, the Design constraint is just how many platforms you can manage to resource well. And from my perspective, I think the most, I think most businesses should focus on option two, which, as we talked about, is just picking a couple of platforms to go deep on and then get the crossover benefits of the others, right? Two or three that are done really well are going to beat. In my opinion, a more scatter shot approach that's going to be a little bit more difficult to execute on and also to measure. Okay, so my recommendation for most businesses is going to be option two, which is going deep on a few specific or chosen platforms. So the next question after that becomes which ones? And I'll explain why I think this, but my recommendation for most companies is to prioritize a mix of Google surfaces. And this would be traditional search, Gemini. AI mode and AI overviews. Now, I think about Google's AI products according to where people encounter them rather than the technology that is behind them. So the Gemini app, for example, that is its own destination. And it is somewhere where a user will go on purpose. But AI overviews and AI mode are different. They live inside of the same Google search experience that people have always used. And an AI overview converts into an AI mode the moment that you ask it a follow-up question. So from where your buyer sits, the way they look at it is it's all just Google search, right? Most people are already in the habit of going to Google when they want answers. And Google is capitalizing on that habit by rolling out its own AI into that experience at a very, very rapid pace. Right. data providers have been estimating that AI overviews are now showing up in more than 40% of searches, which is up from only 15 or so percent last year, the same time a year ago. and then AI modes, visits itself have more than doubled over the last year. And then also on top of this, at Alphabet's Q2 earnings call, they said AI mode had passed a billion monthly active users since they had gone global with that feature in October of last year. So the point being, optimizing for that one surface covers AI overviews. AI mode and traditional results all at once. And my read on this is that I believe AI mode is going to become the dominant experience in organic search. And I'm basing that off of three things. first, the AI overview handoff makes AI mode the just obvious next step for anyone who is asking a follow-up question. And then second, Google announced a redesign of search's entry point at their I.O. conference in May, which puts AI mode a step closer, in my opinion, to being the default search experience. And then lastly, on that same earning call that I just talked about, they also said that they had cut the cost of an AI mode response to its lowest level since they had launched it. So when it gets cheaper for Google to serve an AI mode answer, you can expect Google to serve a lot more of them. So investing, in my opinion, into the Google search experience is much more likely to pay dividends if AI mode takes that dominant position, because it's all native to the same place that users are already defaulting to when they are looking for answers. And That default is pretty sticky, right? Like a recent study that we had we had found saw that almost six in 10 consumers run a Google search immediately after getting a recommendation from a separate AI tool. So, point being is the AI answer for most people is not the end of their research. People are still coming back to the search platform. That they trust most to double check it, which means when you take this approach, Google search actually earns your attention twice, right? Once to get its own AI answers, and then a second time as the place that people go to verify all the other AI tools answers. So if you're going to resource one surface properly, in my opinion, that's gonna be the one. And that means maintaining your focus. On traditional search optimization and putting your resources towards AI overview and AI mode optimization. So that's the case for making Google your very first priority. But if you have got the resources at your company to go deep on a second platform, let me talk to you about how you might pick it. So one method you might consider is focusing on user numbers. So ChatGPT, for example, has published the biggest user numbers of any of these platforms. But you know, let's take all these numbers with a grain of salt because the companies that are publishing them benefit a lot from inflating them, right? A big usage number is how a company or a platform convinces the rest of the market that it's got a lot of staying power, right? also comparing user numbers across platforms can sometimes be an apples to oranges types of exercise. You know, OpenAI's 900 million weekly active users. they announced that in February and they did it alongside a funding round. and it hasn't been updated since, right? Anthropic and Perplexity, they don't publish their user accounts at all, right? Microsoft, which has Copilot, that reports paid seats and doesn't report people. And then Google discloses theirs on an earnings cycle, right? So that's four. Platforms with four different definitions of what a user is. And one of them isn't even counting users at all, right? So the platform with the biggest public numbers might look like an obvious choice, but that could be a trap, right? A usage number that you can't compare to anything else, in my opinion, is just marketing. So, you know, setting those. Unreliable usage numbers to the side, there's really no good, honest way to find out directly which platforms your buyers are going to use. So I would say start with the one thing you already know. Are you selling to businesses or are you selling to consumers? Are you B2B or are you B2C? If you're B2B, Claude might be the way to go because it has been very widely adopted by a lot of different businesses. So there is something called the RAMP AI Index, which is run by RAMP and it tracks corporate card spend across something like 70,000 different businesses. And what it's doing is watching what companies pay for rather than taking the survey route and asking anyone to self-report their usage. And it had recently showed. more businesses are paying for anthropic than for open AI. And right now it's like only about a point and a half. So you could say it's basically a tie. But what I would take from it is that Claude is becoming a default platform for a lot of businesses, which means there's a decent chance that it's going to be the assistant that is open when B2B buyers are starting to look for vendors or for software. Although if your buyers you know live inside Microsoft's ecosystem, which you know could be very true at more of the enterprise B2B level, Copilot could be something worth looking into. but if you're in, for example, B2C instead of B2B, Chat GPT's reach, in my opinion, still makes it a pretty good bet as your second priority behind Google. Pew research had found that. Chat GPT is just at the top of chatbot usage in pretty much every age bracket that they had measured. But lastly, I would also say you can lean on your own internal ICP data because you know, at the end of the day, you know where your buyers spend their time better than any survey or usage report could ever tell you. And we've established that. You know, in this in this conversation today, that each platform also has its own citation pool, right? Claude, like we mentioned before, runs on brand websites and a lot of established websites and publications. ChatGPT SKUs to Reddit and Wikipedia and some industry pubs. But then we also have Perplexity right now is running heavily through YouTube and news outlets. And then also, you know, our own research even found this, and it's well known that Copilot. indexes very heavily on LinkedIn and other Microsoft adjacent properties, and also does a lot of focus on B2B publications. So, what I would say is take what you know about your own customers, and I mean things like the publications that they read, right? what are the communities that they hang out in? what are the review sites that they like to check, and then back into the platform. Whose citation pool looks most like that list. So if your buyers are on, you know, LinkedIn more than that they are on Reddit, that would point you to a platform that would prioritize that source. And nobody can hand you a third-party measurement of where your buyers are asking these questions. So this pick really is yours to make based off of what you understand about your market. But I would also just treat it as a hypothesis, right? Pick your second platform, establish your baseline, put in the work, and then measure it for like a quarter or two. And then you need to be willing to adjust the strategy accordingly. And so, bottom line, there's really no universal playbook here, right? You're your vertical and your buyers, and then also your resourcing. Is going to decide which and how many of these platforms belong in your visibility strategy. And once you've established your initial strategy, remember that this is not a one-time setup, right? These platforms are constantly, constantly evolving, right? We need to test the assumption. We need to check the result, revisit the strategy, and then rinse and repeat. And that's the job. And if you a marketer, you probably already knew that before I said it. So I want to give you four things that I want you to take away from this conversation today. First thing is go and check whether AI crawlers can read your pages and go do that today. So just pull up one of your important pages on your site, right-click it, view the page source, and search for a sentence that you can see on your website on that screen. If it Is not in the code in the page source, that content is likely being addressed by the visitor's browser. And if that happens, most AI crawlers cannot see it. This takes about 30 seconds, and it is definitely the one thing that you can do right now. and it can legitimately cancel out everything else you're doing if it is not set up correctly. And then the second thing that I would tell you to focus on is Write down where your brand is getting named off of your own website. So third-party mentions, like we talked about, are the only lever that moves every platform at once. So make a list of what's currently happening, right? Your review websites, any industry publications, podcasts, or partner content, and be honest about how much of that came from being delivered and having effort. Versus just being lucky. And that list, and more importantly, the gaps in it is the plan that we need to go out and execute on. And then the third thing I would say is we have to stop staffing AI search separately from Google search. AI overviews and AI mode and traditional results all are pulling from similar work. So if you've got a team that is already doing organic search and you've got a separate effort that is spun up just for AI search, you are likely paying twice for one work stream and you're probably splitting the measurement two. And then the last thing I would say of the four main takeaways from today's episode is if you're going to pick a second platform, I would pick your second platform off of I would say your customer's media diet, not off of published user accounts, right? Write out the publications that your buyers read and the communities that they are in and match that list to the platforms whose citations look most like it, right? We talked about this earlier, like LinkedIn heavy buyers are going to somewhere different than Reddit heavy buyers. Right. Give that pick, whichever one you choose the platform, give it a quarter. baseline it and then also be willing to find out if you were wrong and you need to go somewhere else. But if there is a bigger point under all of this in today's episode, it's that AI search rewards the same instinct good marketers already have, right? You don't know exactly where your buyers are. So you make your best read, you commit resources to it, and then you stay close enough to the data To change your mind. And this is a very new channel, and sometimes it can feel very scary, but the job and everything that we can do to control it is not. So if you got something out of today's episode, please follow the show so the next one finds you. That is it for me. I am Michael, and this is the search signal. I will see you next week.