Campaigns take flight with Viasat Ads. Unlock access to over 250 million passengers annually across leading global airlines, with high-engagement ad formats and real-time delivery. Viasat Ads provides access to a verified audience in a captive environment, so your message reaches passengers when they are ready to engage. Join their journey with Viasat Ads.

Marcus Johnson (00:00):

Take your brand to new heights with in-flight advertising, powered by Viasat Ads. High-engagement formats, targeted delivery, and self-service tracking make it simple. Reach millions of travelers across leading airlines in a premium captive environment. Join their journey with Viasat Ads.

(00:23):

Hey, gang, it's Friday, December 5th. Jacob, Nate, and listeners, welcome to Behind the Numbers, an eMarketer video podcast made possible by Viasat Ads. I'm Marcus, and joining me for today's conversation, we have two folks. We start with our analyst who covers AI and technology for us, living on the West Coast, is Jacob Bourne.

Jacob Bourne (00:40):

Happy to be here, Marcus. Thanks for having me.

Marcus Johnson (00:42):

Hey, fella. And our principal artificial intelligence analyst living on the east one is Nate Elliott.

Nate Elliott (00:49):

Hey, Marcus. Thanks for having me.

Marcus Johnson (00:51):

Hey, fella. Welcome to the show. We start, the fact of the day.

(01:00):

Okay, certain species of a land snail, like the desert snail, can hibernate or estivate, which is hibernate in the summer, for up to three years during extreme environmental conditions like prolonged drought or heat. That's too long, isn't it? Well, you know when you're asleep for like 12 hours straight, you know how groggy you feel afterwards?

Jacob Bourne (01:26):

It sounds long, but I bet you there's some species out there that can hibernate for longer. That's my guess.

Marcus Johnson (01:31):

This is the longest.

Jacob Bourne (01:33):

Is it the longest, really? Are you sure about that?

Marcus Johnson (01:36):

Apparently. Yeah. Really. Don't question my... Yeah, apparently this is the longest. Three years-

Jacob Bourne (01:42):

It's a long time.

Marcus Johnson (01:44):

It's 12 hours times 2,000. Too long. Can you imagine? Snail wakes up, his friend's like, "Phil, you're okay?" Phil's like, "Where the hell am I?" So long. Just a different year?

Nate Elliott (02:03):

I just want to know how long a snail lives. What portion of a snail's life is that?

Marcus Johnson (02:07):

True. Big part of his life-

Jacob Bourne (02:08):

I imagine the hibernation might extend their lifespan a bit, would be my guess as well there.

Marcus Johnson (02:13):

Hopefully Phil went to sleep in his 20s, he wakes up 75, like, "Jesus, where did..." Anyway-

Nate Elliott (02:21):

I like that you've named your snail as well.

Marcus Johnson (02:25):

Of course. Of course it's Phil. Is that a good snail name? I don't know.

Nate Elliott (02:30):

Sure.

Marcus Johnson (02:31):

Anyway, today's real topic, how AI has changed search already, and what happens next?

(02:40):

So ChatGPT comes out in November 2022, and every headline, ad, earnings call now includes the letters AI. Because ChatGPT helps folks find information, people have been trying to figure out what this means for search. And so we'll start with a very simply worded question, but maybe very complex to answer question, which is how has AI changed search already? Nate, I'll let you take this one first.

Nate Elliott (03:12):

Well, for some people, it's changed it fundamentally, but for most people it hasn't changed it at all.

Marcus Johnson (03:18):

Well put.

Nate Elliott (03:19):

There are a lot of really interesting dynamics happening around AI and search, and one of the points I make often is these tools can do amazing things. They can invent new content out of whole cloth. They can go through advanced reasoning with startling accuracy in some cases. And yet the most common way people use them is just to search for information the way they have traditionally used Google.

(03:44):

And so for the people who use AI a lot, this is something that for many of them has effectively replaced Google. And you hear that a lot around the industry, but the part of the sentence that most people skip is the next part, which is to say most consumers still don't use AI very much. We're seeing 40-ish percent maybe of US online consumers are using AI on a monthly basis. If you zoom into weekly, that number goes down by perhaps half. If you zoom down to daily, that number goes down even more. And so you've got this small group of people, maybe high single digits, maybe low double digits, who are using AI maybe on a daily basis. And for those people, AI has fundamentally changed the way that they find information online. But for the vast majority of the population, AI is a thing that they use once or twice a month. Maybe they use it when Google doesn't give them the answer they want. But for the most part, the vast majority of the population is still just going to Google a bunch of times every day.

Jacob Bourne (04:50):

Yeah. AI is changing search. I think it's a long process, and really what we're seeing today is just the fulfillment of a long vision. If you think back to 1996 when Ask Jeeves, that search platform first launched, I mean, the vision there was conversational search, but the problem was that the AI, the legacy machine learning models that powered that, just couldn't fulfill that conversational vision that today's generative AI models can. So I think AI has always been under the hood of search, and really what we're seeing is that search has evolved alongside AI. So current generative AI models can kind of bring us closer to that full vision of conversational search that I think was always there.

Marcus Johnson (05:35):

Is it fair to say that large language models then, they're becoming a new place to look for things and not a replacement? Because Patrick Coffee of the Journal was writing, "LLMs, as they're called, aren't replacing traditional browsers anytime soon, but have become another responsibility for brands."

Nate Elliott (05:52):

It seems that way. For one of my first reports here at eMarketer, I dug up the data on the amount of time people in the US spent with large language models like ChatGPT and Gemini and Claude and how much time they spent with traditional search engines like Google, and to a lesser extent, Bing, and what we found was that in one year, from August of 2024 to August of 2025, people started spending about four times more time with AI engines in August 2025 as they had 12 months previously. But while the amount of time they spent with these LLMs quadrupled in that year, the amount of time they spent with Google and other traditional search engines didn't change at all. I mean, literally the Google time-spent data from August of 2025 is almost identical, up I think half of 1% versus August of 2024.

(06:44):

So at this point, it certainly looks like even the people who are using these tools are also still using search engines. And again, because the heavy users of these tools, it still represents a pretty small part of the population, it's just not enough change to really make any kind of dent in traditional search behavior at this point.

Marcus Johnson (07:05):

Yeah. And you've got that data in your new research looking at AI search and what's going to happen to it next year. And it's so stark because you've put it into a pie chart, right? This is that number, and it's 3% slice for the AI things and then 97% for the traditional search things. And that really does put it in perspective when you see all these headlines about how many people are using AI within search.

Nate Elliott (07:30):

Yeah. I mean, listen, it's a deeply imperfect view of how people seek information and look for things online. It doesn't include site search on places like Facebook and YouTube. It doesn't include commerce search on engines like Amazon. But what it does is look at tools people might use for general online search, and right now that pie, from my perspective, seems to consist primarily of traditional search engines like Google and Bing and these new LLM tools. And if you consider that group of products to be the pie of general online search, you're right, only 3.3% of US online general search time in August of this year went to LLMs, and almost 97% went to Google and Bing. To put it another way, Bing got more time spent in August of this year than the top five LLMs combined.

Jacob Bourne (08:20):

Yeah. I mean, we're talking about a massive behavioral change here that I think... People have been using traditional list search for decades now, and so it's not going to go anywhere overnight. I think it's actually also, there's going to be a role for it that's going to continue into the future even as AI search expands.

(08:39):

A recent study from the Wharton School of Business found that even though AI search kind of speeds up the search process for people, that actually users' depth of knowledge about a given topic is actually much more shallow if you use AI versus traditional search. So I think there's something to be said for just different ways to search beyond the conversational search that's powered by generative AI.

Marcus Johnson (09:07):

We've been talking about consumers. We're talking about this from the kind of consumer angle, how their behavior's changed. But I want to talk through the click-through behavior and how that's changed and how that's affecting advertisers. That behavior is consumer behavior, but it's affecting the advertiser and publisher community. People are less likely to click through to the site when AI is involved. There's a bunch of research out there to show this.

(09:31):

One of the most prominent studies at the moment is Pew Research, written up by Athena Chapekis and Anna Lieb. And they found that Google users are less likely to click on links when an AI summary appears in the results. As you can see in this chart, users who encountered an AI summary clicked on a traditional result link in 8% of all visits. Those who did not encounter the AI summary clicked on a search result nearly twice as often, 15% of visits. And people also are 63% more likely to end their browsing session if an AI summary is presented in their search. So, Jacob, what are your thoughts on how AI search has already changed click-through behavior?

Jacob Bourne (10:12):

I mean, it seems like it certainly is having a big impact given that search is such a big portion of the internet. And if you have a link result in a traditional search, of course, then that implies that you're going to click on something. Whereas a conversational result that kind of gives you the answer that you were looking for, at least most of it, maybe then you don't need to click at all.

(10:35):

So I think this, I mean, it's the beginning of a bigger shift for the internet itself and how people use it and how brands need to stay visible on it. The pressure to appear in generative AI search results is going to keep on rising for brands. As Nate mentioned, most people are still using list-based search, so we're not at a dire moment at this point. But it certainly is changing traffic patterns, and especially for media publishers, certainly something that is a critical change that's going on right now.

Nate Elliott (11:20):

I will say, though, people like to think that AI invented zero-click search, in the same way people like to think AI invented lots of things that maybe existed more than 3 or 5 or 10 years ago. Even before AI overviews came to Google, only about half of Google Search Engine results pages got a click anywhere on that page, organic results, paid results, any of it. And yeah, we're seeing that AI overviews are driving that click-through rate down even further.

(11:53):

Likewise, I've seen data that says that as few as 6% of AI engine responses generate clicks, which is far lower than what we're seeing on search engine results pages. But there were a lot of people who were searching for things getting what they needed, either from the knowledge panel or from the two-line summary under one of the first couple of search results, and not clicking through. There are a lot of people doing that already. This is just accelerating and making it a bigger deal. But it's not like this is a new behavior. This is something that brands and retailers and publishers would've had to deal with anyway.

Marcus Johnson (12:32):

Yeah. And we should say Google is currently refuting these claims, stating in a blog post that despite these third-party reports, organic click volume from search to websites has remained consistent year over year. They were saying that a lot of these studies have very small sample sizes, and Google also saying the number of links showing up has gone up. BrightEdge report saying that search impressions, number of links that show up in searches, even if they aren't clicked, grew by nearly 50% in the year since AI overviews was launched.

Nate Elliott (13:03):

Yeah, but let's look at that for a second, because-

Marcus Johnson (13:06):

Please.

Nate Elliott (13:06):

... a lot of what AI overviews is doing is taking the 10 blue links that would've showed up anyway and sort of repackaging them into a conversational response that in most cases heavily feature the 10 blue links or some subset of the 10 blue links. So when they say that there are 50% more links, what they mean is the 10 blue links are still there and then AI overview is taking half of them, making them into a paragraph, and surfacing them again at the top of the page.

(13:33):

And likewise, when Google says overall organic search volume hasn't declined, it's because, again, AI search is only 3.3% of the time people spend searching for things online. It's not big enough yet to create a sort of macro trend in the data. I think Google may be hiding behind that a little bit. I actually trust the studies that say that we get a lot less clicks when AI overviews exist. I think that makes sense. The fact that, again, a full-fledged LLM response gets only about a 6% click-through rate tells me that this is real, it's going to continue, it's going to accelerate.

Marcus Johnson (14:14):

And this is a lot of data to support, so much data to support, that ARF's showing Google's search click-through rates of top pages fallen since AI overviews were launched, going from 6% to 3% within a year. Studies from Seer Interactive, from Authoritas, from sites in the British Paper, The Guardian, there's countless data suggesting that traffic click-through rates are going down.

(14:39):

We've been speaking a lot about how this is affecting Google because new players are coming into the market, like the ones we've been talking about, platforms from OpenAI and others, but then Google can also use this technology to its benefit. So I'm wondering where you guys land with this question. Our colleague and technology analyst, Grace Harmon, writing that Google's grip on search weakens as AI tools gain ground and reshape the path to product discovery. She notes that the share of consumers using AI search tools on a daily basis dropped, oh, sorry, not dropped, doubled to nearly 30% from February to August. Meanwhile, Google's share of general information queries fell from 73% to 67.

(15:21):

On the other side of the coin, Asa Fitch of the Wall Street Journal asks, is AI killing Google Search? It might be doing the opposite, suggesting the world's dominant search engine is so far adapting well to the AI age. Its AI overview tool, he says, has over two billion monthly users, and Google has AI mode that competes more directly with chatbots.

(15:41):

So, Jacob, is Google in a better or worse position today because of AI search's fast rise?

Jacob Bourne (15:46):

Yeah, I'd argue that actually Google is in a stronger position today because of AI search. Now, it didn't immediately seem that way. There was a lot of concern when ChatGPT launched in 2022 that this would be the end of Google's ad revenue. But I think what happened is just that it accelerated the timeline of what Google was going to do anyway. I mean, Google had the transformer, which is the technical foundation for generative AI, was invented at Google in 2018. Google kind of sat on that technology and didn't commercialize it. And so ChatGPT, OpenAI/Microsoft kind of caught Google a little bit flatfooted in 2022, but I mean, I think Google's structural advantages are now surfacing as sort of levers to really help it catch up. Its cloud revenue, based on AI tools being used by the enterprise, AI and infrastructure for startups, that revenue is growing. Its ad revenue is still growing. I think that Google has sort of an AI innovation and big data advantage over players like Microsoft.

(16:58):

And then on the cloud front, even if OpenAI does well or any other competitor does well, it's still going to get cloud revenue. So I think it's leveraging its existing massive search user base effectively so far with AI, while also growing its AI cloud. Its ad revenue is still growing. So I think that this is ultimately sort of just an accelerated vision for what Google was ultimately going to do anyway. I mean, back in the year 2000 even, Google co-founder, Larry Page, essentially said AI is going to be the ultimate search engine. So it took over two decades to see that happen, but I think it's on a good footing right now.

Marcus Johnson (17:50):

Nate, do you agree?

Nate Elliott (17:51):

I do. Yeah. First of all, I think Google's in a lot stronger position in consumer-facing AI than most people give it credit for. We're at a point where Gemini's got 650 or so million monthly users, where AI mode probably has even more than that. As of this recording, the last time they talked about numbers for AI mode, they had 100 million monthly users, but that was back in July when it had existed for two months in the US, for less than one month in English-only in India, and that was the sum total of people who had access to AI mode.

(18:28):

Right now, it is available in over 200 countries and a whole bunch of languages, and if the take rate is the same in all those countries and all those languages, we could be looking at 800 or 900 million AI mode users per month right now, in addition to partially overlapping with the 650 or so million monthly Gemini users. So we're talking about really big numbers. We're talking about Google potentially being at or near a billion total AI users at this point every month. And frustratingly, OpenAI reports their numbers weekly, while Google reports theirs monthly. It makes it really hard to compare them. But whatever the number is, Google's a lot closer to OpenAI right now than I think a lot of people give it credit for. So I think that's part of the story of why I think they're in a really good position.

(19:18):

Add to that the fact there are billions of people around the world who go to Google Search every day, and they have these incredible distribution channels. They have the world's largest mobile operating system, the world's largest browser, the world's largest web mail, the world's largest maps, the world's largest on and on and on, and they know how to take those tools, they know how to take that distribution and leverage it into consumer adoption for products, even when those products are late to market. Add to the fact that in blind taste tests, we see that people actually tend to prefer Gemini over ChatGPT for consumer AI behaviors, like information search and image creation and video creation.

(20:00):

And finally, you get to the fact that Google already has this enormous advertising engine in place. It has the ability to actually make money from these tools whenever they want to. And while Sam Altman continues to make noises about not liking advertising and needing to reinvent the model, Google can literally flip a switch and make billions off this whenever they want. So I think for all those reasons, this is shaping up really nicely for Google right now.

Marcus Johnson (20:27):

It's certainly an argument for Google is fine. They've made over half a trillion dollars in search ad revenue since ChatGPT became a thing. And, Nate, you write in your new research that, and I think this really sums it up quite well, Google's hold is 90% of global search engine share according to September StatCounter data, a number that's not fallen below, you say, 89% in the last 15 years. No other search engine claimed even 5% of global search at any point during that time. However, you do say that AI poses a real threat to Google Search monopoly. Tell us what you mean by that. Is that looking kind of further out into next year?

Nate Elliott (21:15):

Well, what I mean by it is this is the first time there's been any real doubt about whether Google can continue to exert this virtual stranglehold on how people find information online. I mean, again, their share of global search hasn't fallen below 89% for at least 15 years, as you just said. It's actually longer than that. And at no point in that 15, 20 years leading up till now has there been a technology or a company or anything that's made it say, "Oh, wow, Google's got to think about this and respond to this and have some good tools to show people why they should keep going to Google." I think generative AI is the first time in a couple of decades that some people, a meaningful number of people, have found a tool they think is as good as or better than Google for getting them the information that they're looking for. And if Google didn't already have such a strong position in terms of more generative AI users than people think they have and a number of different tools that do well in these blind taste tests, then I would be more concerned about their ability to keep that stronghold.

(22:25):

I think it is a threat to Google. Google's not going anywhere. They're still going to be the primary way people find information five years from now, and I think they're likely to be, in fact in my AI trends report dropping this week, I'm predicting that Google will overtake OpenAI in terms of gen AI users in 2026. So I think they're not just going to be the number one way people find information online five years from now. I think they're very likely to be the top AI tool set five years from now.

Jacob Bourne (22:57):

What I would add to that is just that prior to ChatGPT's launch, there was the concern that Gen Z was going to social media more, social media sites like TikTok, for example, for search. So there was some talk of concern over Google due to that. I think that it's much, much smaller of a concern than generative AI has been because, in part, it was a smaller demographic that was using social media for that purpose. But I think on both fronts, the threat is still ultimately probably small long-term for Google.

Marcus Johnson (23:36):

Yeah.

Nate Elliott (23:37):

I mean, Jacob, one of the things in your recent report that you talk about, though, is sort of the generational aspect of this and how we could see a generational change in how people seek information. And some of the data you presented on the school calendar year and gen AI usage I thought was really notable there.

Jacob Bourne (23:55):

Yeah. And going back to Nate on that point on just the whole behavioral change aspect of that, we have to also consider that Gen Alpha is growing up gen AI natives. So their search behavior is going to be heavily influenced, I think, by AI and it's going to be more second-nature for them to just search with AI versus traditional search.

Marcus Johnson (24:18):

Yeah. I like this idea of using different platforms to search differently as well. Grace, our colleague, Grace Harmon, who is writing different search platforms have their own specialties, making multi-platform search strategies more common, saying that Amazon leads for direct product queries, social media, Jacob, as you were saying, growing importance in discovery-based shopping, Google still dominates in early-stage or product research, and AI gaining ground in comparison shopping and personalized recommendations. So it's kind of similar to how different folks use different social media platforms differently. They don't have the same experience, even though they all copycatted each other. People use TikTok for certain things, they use Snapchat for certain things, they use YouTube for certain things, and it seems as though it may be similarly with search, they're going to certain platforms for certain ways of searching.

Nate Elliott (25:09):

I'm not sure how much appetite people have for that, though. I mean, I understand Grace's point. I think when people want to find information, though, they go to one place; they go to Google. They have for decades. One of the things Google found along the way was that even though they've got more than a half dozen different types of search, they've got news search and product search and map search and everything else, people didn't want to think about which of the Google varietals they needed. They just wanted to go to google.com and use one search box. And so rather than offering different tabs on the homepage, Google says, "Yeah, just search one time. We're going to figure out what it is you're looking for and we're going to default you into the type of results that best match what you're looking for." And they do a really good job of that in general, saying, "Oh, we think you're looking for maps information right now." And so instead of just 10 blue links, most of the screens filled with map information or product information or news information or whatever they think you're looking for.

(26:12):

I'm pretty sure they're going to start defaulting people into a full AI mode LLM chatbot for longer and more complex queries, maybe 5 or 10 or 15% of the time, at some point in 2026, but I also think that's part of Google's advantage here is people really prefer, when they can, to just go to a single place and trust that they're going to get the information that they need. I mean, if you've ever taken a corporate AI training, one of the first things they go through is what you use different models for, use Claude for text creation, and use Perplexity for detailed reasoning, and use Gemini for deep research. And I don't think anyone has the time or the space in their head to remember all of this and make those choices, and I think the fact that Google can do all of this in a single search box is a big part of the reason why I think they're likely to win in the consumer-facing side of AI.

Marcus Johnson (27:12):

Jacob, do you agree with that?

Jacob Bourne (27:14):

Yeah, I pretty much 100% agree Google will likely win the consumer-facing AI... That's not the entire AI race, though, so I think that there's still opportunity for niche players in the market. Probably not very many of them, but it's mostly going to be Google, and then you're going to have other opportunities in different enterprise sectors, different academia, research, things like that for others.

Marcus Johnson (27:44):

Yeah. Yeah. I wanted to cover... Nate, you've got a ton of recommendations for brands and agencies in your report, but we're almost at time. Maybe we can give folks a quick one, a quick recommendation for brands and agencies in terms of what they should be thinking about when it comes to AI and search right now.

Nate Elliott (28:01):

Yeah, I'll combine two of them into one real quick, which is you're not late, but start now or you will be. A lot of what I hear from brands and marketers and agencies and retailers is a real sense of concern that because they don't already have expertise and lots of experience and case studies in successfully using AI to market and to reach consumers and to drive sales, that they're far behind. A lot of people are under the impression that AI has already replaced Google and fundamentally changed things already. I mean, these are all things that may happen in the future. Again, we don't think ChatGPT will replace Google, but we do see AI having a really big impact on how people are discovering and finding and buying products and services. But I'll go back to that stat, 3.3% of general search time in the US is spent with AI, and 96.7% is spent with general search.

(28:55):

So you're not late. You're not in crisis mode if you haven't started yet. Having said that, I think you will be late soon, and if you start right now, you're still an early mover. You're still an early adopter. And that's a good place to be.

Marcus Johnson (29:10):

Well, your full report is called AI Will Dominate Search, and then you go into explaining when and how. You go to emarketer.com, you can find it, or of course there is a link in the show notes. Thank you so much, gents, for hanging out with me today for this episode. Thank you first to Jacob.

Jacob Bourne (29:28):

Thanks for having me today, Marcus.

Marcus Johnson (29:31):

Yes, indeed. And thank you to Nate.

Nate Elliott (29:33):

Thanks. It was a pleasure to be here.

Marcus Johnson (29:35):

Yes, sir. And thank you to the whole production crew and to everyone for listening in to Behind the Numbers, an eMarketer video podcast made possible by Viasat Ads. Make sure you subscribe and follow, leave a rating and review if you can. We'll be back on Monday. Happiest of weekends.