So I am going to kick off this presentation, talk about, just very briefly, what we’re what we’re gonna go through today, and then I’m gonna kick it over to you, Eli, and let you take us away. Tell us about Datos. Tell us about Clickstream data broadly. I think many folks are gonna have a lot of questions. I certainly, have learned a tremendous amount about Clickstream, but still love to learn more. So alright. I I think Amanda might fairly ask and did ask, how the heck do you measure who sends traffic on the web? I know. I need more shots of you, Amanda, in various poses so that I can insert them. It’s it took a long time to find a good photo for this. But the the question of how do we measure things at scale that happen on the web and try not to get bogged down by individual bias. For example, trying to calculate who sends traffic on the Internet by just looking at a few hundred or a few thousand people’s Google Analytics, which would not tell you because anything that doesn’t contain a referral string is not gonna be in there. The bias that comes from the the websites you might analyze. So what you really need is a huge group of people who are willing to give you this data from their, you know, their desktops and their laptops and their phones. And that is where clickstream panels come into play. So if you want high quality analyses of millions of people, what what millions or tens of millions of people, real people do on the web, the the URLs that they visit and don’t visit, the things that send traffic to other places, what those sessions look like, path to purchase. The the only real way to get at this is clickstream panels, and I will do a poor job compared to Eli of answering what this is. So, Eli, will you will you, kick us off, take us away? Oh, I should. What should I do here? Should I stop well No. No. You are in charge. You push the button. Click through the be easy. I was telling Amanda earlier, I had to switch to a Mac recently. And if I had to start figuring this out on my own as a long time PC user, it would probably mess up the flow. Right? So I’m be your I’m gonna be your your hype man in the background. Yes. You just give me you just give me, like, the hand signal for for go to the next slide, and boom. It’s gonna go there. You got it. You could be my my my speaking emanuensis. Right? You could be my support. So so thank you, Rand, and thank you for the introduction, and Amanda and everybody for joining. I’m going to keep my portion relatively light today, and I say that because I joke with Rand that he’s the real draw. Right? That’s why people come. That said, when it talks about data, I’ve been in the data business since the late nineties, and ClickStream, in particular, for, call it, fifteen plus years. So, Rand, if you go to the next slide, I really just like to establish I love to talk about methodology. Right? Where does data come from and why does it matter? And in this instance, what is ClickStream? And think of Datos as a a provider of ClickStream that meets everybody’s compliance criteria. But the net of it is that we focus on, at scale, anonymized data. Right? That’s what we’re after, and we do it in an opt in capacity. So if you think about our panel, it’s fifteen million total people worldwide spread out across, you know, you know, call it two hundred plus countries. It’s about seventy percent desktop and thirty percent mobile. Alright? So when you think about what our particular split is, I do note this in one of my later bullet points is that it’s we are just focused on browser activity. So clickstream, by the way, by definition, is when you are on a desktop or, like, a web browser or on a web browser when you are on your mobile phone. We don’t have app data, but it is quite literally the stream of clicks that you are making, which is effectively a massive online intelligence database, which is really about the ability to find relevant information that is for pattern recognition, which I’ll talk about a little bit further. I think that that’s really about where we focus on these things. So the key about it is think about the journey that anybody here takes when you’re on the web. It effectively is what websites do I visit, what search terms do I use, what videos do I watch, what products do I buy, and then what order do I do those things in? Right? That’s really and we’ll talk later about consumer journey. And then as I had referenced already on the next bullet point there, that slide you know, it’s desktop and mobile browsers, no app data. Right? Can I can I ask, is is mobile app data eventually possible? Is that something you guys are interested in? Or are you sort of like, well, maybe long term, but short term, we’re focused on these other things? It is. Like, if you could imagine, that’s probably one of my number one questions that I get all the time is about, can you tell me what’s happening inside of apps? But from a methodology perspective, that is a much more difficult thing to gather up at scale. Right? To go out and recruit a panel, and to be able to have people that offer up that information and are willing to do it. It’s just a very expensive proposition, to be honest. So it is something that is on our road map to be able to add. But, you know, we are not in the business of, like, dropping SDKs behind you know, onto apps and things like that. It’s not what we do. So we deal at the the user level as opposed to, like, going directly to the brand level, if you will. So it’s there. I would love to have it. It definitely if you think about ClickStream, it’s not the entire picture, and that’s an important part, whether it be Clickstream or any dataset, is remembering what silo or collection of silos am I crosstabbing. Right? Totally. So it’s important to think about we talk in terms of, like, classic population and, you know, enumeration and projection and representation, you know, all these different types of, like, statistics, one zero one words, and different things. So I just I always say caution everybody, but I do like to define, and that’s what this slide is about. What are we and what aren’t we? Right? So when we go through and you think about the the output of, you know, the insights, if you will, you know what we’re talking about. Totally. Yep. Got it. So there was there was a question that I I think is actually maybe pertinent right now, more so than later. Can can you tell us a little bit about Datos’ market, sort of uniqueness versus something like SimilarWeb? I think, Kate had asked about that. Yeah. Actually so if you go one of the next slide, we we do actually talk a little bit about what is different about Datos versus other Clickstream data providers. So, Rand, if you click four times for me, let’s fill out all of these because I will speak Okay. Okay. So they’re really, like, the way that I would sum each of these up. One is size. Right? That goes back to the point earlier is about scale. So, again, I can’t you could talk to similar web about what their numbers are, what they do, or what they don’t, but I could talk about what we do and what we do well. And one is we certainly have an incredibly large panel. Right? Like, is fifteen million people worldwide. Really is a lot of people to be able to, like, passively observe, which I’ll talk about in a second. The second part is about richness. Right? That’s something that we focus on. There are many panels you talk to, and it’s about mile wide, inch deep, or, you know, mile deep, inch wide, etcetera. Like, what if if I said the same thing twice, but you see what I’m getting at. So, again, it’s about breadth and depth. And I think that that’s also a very important thing about clickstream data and ours in particular. It’s a focus on all of the parts of the funnel. Right? So I had mentioned those before, I imagine, from search all the way down to, purchase. And there are different types of methodology that are great at other things. There’s web scraping for what’s going on on a page or know? So you could certainly go through different methodologies about what is great or not great, but that is something that Clickstream and what the way that we do it, it comes in quite handy. I think the third one there, and I say in the bottom left, passively observed data. This is really about what I would what I did not make up this term, but it’s a physics term called the observer effect. Right? And the idea being is that you inherently change something the moment that you begin observing it and interacting with it. Right? So think about classic panels the way that and this is like Schrodinger’s cat, for example. Right? But if you were to think of, you know, classic panels, you would go out. This is pre Internet. You, like, you recruit people. Right? Think of, like, random digit dial on your phone, and now we know who’s gonna be the next president president or what have you, is that if you think about the recruitment process, the incentivization process is that going out and finding people that are incentivized by whatever recruitment, you know, upticks there are is you inherently change it. It becomes difficult mathematically to figure out how to do it. So when it comes to what it is that we do, it’s about passively observing. Right? So that we are not interacting with the observers. We just watch what they do online. Right? And then limit ourselves to only the things that And they have opted into and are appropriate. Yeah. Just just for clarity, I think there there might be some confusion among some folks in here. Can you can you talk about, like, how a panel is is constructed? This is not Sure. Someone was like, is this third party cookie data? Like, are you collecting this? But no. Right? This is this is more like Nielsen set top boxes in the nineteen sixties. Yeah. That would be without listen. I could talk about this all day. Right? So just to go directly to the question is we go out and we recruit and focus on consumer facing software that meets any number of opt in and compliance criteria. And we recruit directly at the user themselves so that the user is opting in, in exchange for a free software or something to that effect. And then we only and we’ll talk about this later. We only deal on anonymized data. Right? Like, again, we’re an intelligence database, not an ad tech company in any way. So that’s what happens is that it’s not about cookies or AAIDs or MAIDs or IDFAs. You have to geek out, just list every single ad targeting acronym. But you can imagine things like Apple ATT and Google cookie deprecation and so on and so forth. These are separate from what we do. Right? Because it’s opt in directly from panelists themselves. So that’s where the information’s coming from. It’s passively observed, from when they are, you know, browsing the Internet and the software is seeing what they’re doing on the Internet. So if that answers that question and helps clarify Yeah. What we’re talking about. It’s generally the thing where, like, as an example, I might get a, I don’t know, an offer on my phone that’s like, hey. You know, do you want to share data with Datos? And in exchange, you can have whatever, like a free subscription to HBO Max and watch your TV shows. And I’d be like, yeah. Cool. And then they say, hey. Datos is gonna anonymize and aggregate your data, and, you know, that that data will be never tied to you personally, but it will be collected for every URL you visit. And I’m like, yep. Awesome. Cool. Give me that free HBO or whatever. And and like what the value exchange is. Right? But even to be more specific, and this is a very important factor, is that we only work with software that have a defined business purpose for actually, like, observing URLs. Right? Like, know, so it isn’t like it doesn’t work, like, just call it compliance wise, then it would be like, you’re a flashlight app. So then the background are also harvesting URLs. Right? So it needs to be like, this is a retail couponing software that, like, specifically needs to see your activity in order to deliver what it is that you downloaded it for. Right? So just to sort of get into that point. So I could talk further about this, and anybody is certainly happy to email me later or hit me up on LinkedIn. But I will answer one question that did pop in. One thing I did wanna talk is that diversification point is important, Rand. And that is going back to why we focus on size is that since we are not out there recruiting per se, like in a classic, you know, again, call it pre Internet panel sense, the size of a panel being spread out across a couple hundred countries tends to smooth out a lot of those wrinkles, right, when it comes to bias, if you will. Like, if you have a large enough sample, people tend to you tend to cover enough diversification by having different software and different people. It helps take care of that representativeness, if you will. But, identify outliers and sort of, like, you know, count them less or decide not to include them in the panel if it doesn’t make sense to do so. That is correct. Like, of our clients, they have you know, we will center in on different types of projects that will say what represents a stable panel of people for a particular project. Right? Like by demographic or behavior or whatnot. Yeah. So I was gonna show I’m not sure if if folks have seen, but this is, I believe, a couple years ago when your panel was a little smaller than it is today. But, you know, we did this big report at SparkToro where I tried to measure third party traffic estimates, And this was kinda my first exposure to the quality that you get with with Datos. Oh, actually, let me blow that up that graph up a little bit. But, yeah, if you you can see right here that essentially, Datos’ monthly uniques was the second best at the time correlated and very, very close to the number one with, Google Analytics traffic from all the people that we, all the sites that we analyzed and who opted into our collection methodology. And so this was the first thing that made me go, oh, wow. That is really impressive, for a relatively new company. I think Datos was only a couple you guys are only five years old, six years old? No. Actually, even less than that. You go back. We were founded in late twenty twenty. This was, like, from COVID on the couch. Right? And Wow. Yeah. Now we’re fifty people and doing our thing. But you’re hitting on it, Rand, and I think that’s an important factor for all of the data scientists at home. A clickstream panel is like gathering up data, but it’s really about the mathematics behind, like, the projection methodology when you talk about how do you take a sample and project out to a population and then extract what are going to be insights that make sense? And I think that this is where as Rand goes on to how we apply clickstream data and to be you know, later on when Rand’s going through it, you’ll see, like, actual insights. Our business is in, like, the derived data business. Right? So that we’re here in order to be able to help with insights. We are ingredients, right, to be able to contribute towards what our results I’m not in the results business. I’m in the ingredients business. Right? I have ingredients. You bake cakes. So I’ll quickly say, you know, we have a lot of different clients and a lot of different industries, but we break them effectively into what we call enterprise clients and then institutional finance, like hedge funds and places like that. But, I mean, listen. I I’m not gonna read every bullet point. I’m sure everybody here can read. The idea going back to the ingredients is that, you know, in the right hands, right, our data can weave flax into gold. Right? It’s there to be very helpful as intelligence towards, you know, building products, being predicting what’s going to happen in the future, helping me segment. Like, going into the days of one to one targeting and segmentation, I believe, and we believe are dying. Right? Like, I think we’re all just moving away from that. And that’s been part of our supposition from the beginning. Right? It’s been that I am I’m an old school marketer. Right? If I go back, I’m probably older than I would care to admit right now. But, like, guess what? Advertising works before the Internet. Right? Like, so and and part of that is remembering that segmentation and contextual delivery of relevant messages, it still works. Right? Like, classic branding legacy, you know, legacy impact that something may have. So, again, clickstream data is it’s about pattern recognition across all of these things since we’re not all unique snowflakes. We think we are. We feel like we are. But, like, every single one of us could be broken down into one of, like, I don’t know, remember the number, but sixty six Claritas prism segments, shotguns and pickups or whatever. And it’s like, you know, you we all we all move and make buying decisions, you know, we’re not we’re not that individually special. So I remember when I got put in the shotguns and pickups category, and I thought, I think there might be data accuracy issues here. Right. There may be a couple issues. Well, I mean, that was just based on your ZIP code plus four. Right? So that was also, like, you know, you start there and be like, who are my neighbors? Right? Like, what are they comparing it against? So, anyway, just to keep moving. So, again, I I know everybody wants to get to Rand, and we’ll talk about, you know, the study and about referral traffic. But if you even click once more, Rand, you’ll see a little pop up bubble. But the net of it is that what can Clickstream data do, it really goes back to that intelligence database. This is a very I wanna call it like a blunt force instrument example. But as you get closer to, you know, Christmas day on Amazon, guess what? Searches for gift cards start to spike. Right? So I I also wanted to call out this one. What’s happening here? Right. A cleaning brush, my feeling is that that’s could be an example of, like, I got a lot of people that are coming over and, like, I have to deal with this. Like, there’s that’s in a million different permutations of what we could go with here, but it is funny, and you see things start to spike, within the data. And then you’re sort of like, oh, that’s kind of like a funny that’s kind of a funny outcome. So, yes, depending upon how large the the group is that’s coming over for the holidays. Yeah. Like, if you move on to the next slide, Rand, I don’t wanna take up too much time. You know, also and this is really where we’re gonna talk about this later, identifying new marketing opportunities. Right? And this is looking at over the past couple of years. Reddit has been on the rise for the five brands that we’re gonna analyze within or at least call out in some slides later on. So there’s some interesting stuff that has, you know, been happening and, you know, there’s any number of takeaways from it. And the last one that I’ll not you know, or a second to last one. Sorry. I forgot I had additional slide. Like, traffic and interest. It’s really just like a nice, call it, like, you know, guideline to be able to say, is something growing? Right? And in fact, if you can imagine, anybody have any idea what happened in Major League Soccer in twenty twenty three that maybe there was one of the most famous best of all time soccer players that showed up. Right? Oh my god. Major league soccer. I thought this was the multiple listing service for real estate. No. Sorry. This is a Major League Soccer interest. But I just I I just liken it to, there are little things that could be indicators to help quantify what feel like very qualitative events. Right? Like, you’re like, well, how big is that? Or how much does Apple, like, pay the MLS and Messi for showing up and so on and so forth? I mean, clickstream data could be very helpful, right, in being able to define that. And, of course, this is really where we shine, right, as we work with builders. Right? And you can see we are proud to work with Rand and SparkToro. But going back to that ingredients point or, again, all of my, I’m from the south metaphor, so I use them all the time. Like, I’m in the concrete business, and, like, you build houses. Right? And that’s certainly something that we look at, which is breaking down Clickstream into its component parts, and turning it into something that is going to be useful in the right hands. Yeah. And so, I mean, we get to do amazing things here, right, where we can say, hey. People who search for garden furniture end up on these websites even if those aren’t the sites that rank in Google. And that is, in my opinion, so much more valuable than just knowing which sites rank for which keywords. Right? Which anybody can perform in a Google search. Well, that’s really it. Right? It’s a it’s like a fourth dimensional analysis. Right? You could determine what people are doing, like the quantitative. But then doing it over time and connecting those dots in a neural network fashion. And I think that’s really when I think of SparkToro and the tools that you’ve built, that’s what it’s about is that is there a pattern to be recognized in a nonlinear fashion outside of, like, last click attribution. I clicked on search, now I bought you know what I mean? Like, we could talk about that all day. So with that, I should probably hand it over to Rand as I’m sure everybody wants to hear about the bulk of this. Yeah. Yeah. And and, Dino, by the way, good to see you here, my friend. Yes. Datos is one of the sort of four primary data sources that we use inside SparkToro. So it’s really powering a lot of the what do people who fit this audience profile, what do they visit? And then that helps us obviously get lots of data about, other things as well. So we Datos is one of our big, components after we moved off of Twitter with v two. Alright. So let’s talk about today’s hot topic. So hot right now. For a long time for a long time, I really have wanted to get data about who sends traffic on the Internet besides Google. Right? Like, SEO was my focus for my the first seventeen years of my career. Right? Like, that that’s what I worried about all day every day at Moz, and I really focused on it. But as I was leaving that company and starting SparkToro, I was like, gosh. You know, can we can we move beyond that? So in two thousand nineteen, the year after I started SparkToro, I worked with, Jumpshot, which which used to provide clickstream data, and published some data. And one of the one of the pieces of data that I got was this piece of data, and it received a whole mess of criticism because of methodology and style and what we were measuring and not measuring. And so and and then, obviously, jump shot got got shut down or whatever. And and I was so frustrated that I couldn’t go back and get high quality data about who really refers traffic on the Internet. I don’t think this study was fully wrong. I don’t think it was as, problematic as a lot of the people who who pilloried it online made it out to be, especially on Twitter. You know, if if you’ve ever spent time arguing with SEO bros on Twitter, it’s Amanda knows. It’s just a joy, you know, and a privilege. I yeah. Anyway, what what I will say is when we got to do this with the Datos data, which I which I would argue is even higher quality and, prepared in such a way that that we really worked around a lot of the critiques. You know, I knew going in what some of the problems were. And so I worked with, Eli’s team over at Datas to sort of, like, put all these things together. And it was it was great. Like, it was amazing to be able to get that quality at scale. So this is what that data looks like today, specifically for January. I believe this is January of twenty twenty four. Right? And so, you know, I bet you can guess what some of these are. But if you wanna if you wanna play the game, what is the third largest refer of traffic on the Internet? YouTube. Fourth, Facebook. Fifth, Reddit. It has been rising, by the way. Definitely rising. Twitter has, moved down a few slots. There’s no surprise there. Yahoo, then DuckDuckGo. Duckduckgo is small. It’s so much tinier than Instagram. But because it’s a search engine, it refers way more traffic than Instagram. As anybody who’s, you know, been on Instagram knows, they make it really hard to get to another website from Instagram, whether we’re talking about desktop or mobile web or mobile app. Then Amazon. Amazon does send traffic out, which is kinda nice. In fact, they sent a little bit of traffic to, to Geraldine’s website recently because she was a best selling author over there, and you can you can click on her author profile and get over there. LinkedIn, right behind them, and then Wikipedia. And then if you go to, like, the next hundred and fifty five sites, if you’re at the end of sort of the traffic refers that we looked at, you’re talking about, hundreds of a percent of traffic referring. So, you know, if we were to take the next ten thousand sites that refer traffic on the web, they would not equal, probably combined, they wouldn’t equal what Yahoo or Twitter sends. And that’s that’s just the way it is. Right? Like, the the top dominates. So if you’re interested, this this graph is not meant to be screenshotted or to look pretty, but there is I do have the full list of a hundred and seventy traffic referring domains in the blog post. Amanda, could you could you share that with everyone in the chat so they can get access to that if they’re interested? So you can download this and play with it and, you know, make make your presentations for clients, use the data however you’d like. We really like to open open up that stuff and so does Datos, which is great. And then I I decided that I wanted to break out the data in another fashion that I think is really important, and that is to look at the head of the sort of all the websites that receive traffic on the web and the the long tail, right, which which is is these these guys for for you. And so what I did is I took anything that’s outside of that top hundred and seventy, right, like, top domains. It’s actually a little more than that because we removed adult content and some, like, CDNs and and other stuff like that that that didn’t really belong here. Interstitials, right, that are advertising based things. Like, when you click on a Google ad or something, it goes to a a redirect server. We remove those kinds of things from here. But the long tail, basically, all of our websites get the majority of their traffic, overall majority traffic from Google and then Bing and Microsoft sites. I I put Bing and Microsoft sites together because a lot of the searches that happen on, you know, whatever, live dot com or Microsoft online dot com or Microsoft dot com or SharePoint or a bunch of these other places use Bing’s search engine, and so you will get a bunch of referral traffic that comes out via Bing, which is why they’re grouped together. And then Facebook and Reddit, YouTube, and DuckDuckGo, etcetera. So so you know, this is this is a a breakdown of that. The ones that surprised me most, to be honest, are, Reddit, YouTube, LinkedIn, and Pinterest. And the reason that they surprise me is because I sort of thought I especially of, Pinterest, probably more than any of these, I thought Pinterest was really sending traffic to just a handful of the really big sites, especially places like, you know, whatever, DeviantArt or ArtStation or, a couple of image hosting places. They send out ton of traffic to, Google Images itself. But turns out Pinterest is actually a a decent traffic referrer. And I was frustrated quite frustrated by Discord and Twitch, which are which are a little more traffic hoardy than I was hoping. OpenAI just barely made it on this list. That’s the host for ChatGPT. So if you’re wondering, hey. Does ChatGPT like send traffic? They do, but it’s a very small amount. Then I put together this list. These are the most stingy traffic hoarders on the web. And what this means when when you’re looking at these in order is essentially that WebMD for every thousand devices, right, like a a unique device that visited it in a month, WebMD only sent fifty eight clicks to other websites, which is pathetically small. Like, in my opinion, problematically small. If I were an antitrust person, I’d be like, hey. What’s going on? Well, probably not with WebMD, but maybe with some of these others. And the most generous traffic refers on the web, no surprise, it it really is Google. Right? This is basically saying that, you know, even as bad as Facebook’s gotten about hoarding traffic, they do refer essentially, you know, five point eight clicks for every thousand monthly uniques. Now that’s quite small. Right? Like, in my opinion, that’s very small. That means a thousand devices visit them. Only fifty eight hundred visits are sent out. You know, Facebook’s obviously keeping the overwhelming majority of every visit, especially when you consider that they’re the most still the most popular social network behind YouTube if you consider that a social network. Then we looked at changes. So, Eli, we grabbed data, I think, going back to early like, January twenty twenty three to to January twenty twenty four. And so Right. Yeah. You can see in here, right, that essentially there’s a few interesting ones. I mentioned HBO or max dot com. Right? And you can sort of see them growing their referral share a little bit. But this threads is the one that I thought was gonna be particularly interesting to watch because you can sort of see it, you know, over the last couple of quarters. Remember, it only launched in, what was that, July of last year, and it is growing fast, both in terms of traffic and in terms of traffic sent out. And remember too that we’re not measuring the mobile app here. So this is looking at essentially just Threads desktop and mobile web experience. I think Threads is really one to watch given what we see in here. Seems like if they continue on their current growth trajectory, they will be bigger than Twitter, in two years, I think, two and a half years, something like that. Also, Twitter shrinking, so that kinda helps. Significant changes in referral share of some of the larger traffic sending sites. If you remember, last year, Elon turned off. He was like, hey. We’re I I don’t like how rich links are appearing, and so I’m gonna sort of turn those off. And you can see him doing that in q three, which brought down the share of traffic that they sent out. And then he was then apparently, the metrics didn’t look good from that, and so they turned them back on in, I think, was November or December. And now you can see that they rose again, not not quite to where they were, last summer, and that’s just because Twitter in general is shrinking. But there you go. Desktop and mobile. Mobile Reddit. Isn’t that crazy? You know why I think this is? Here’s my theory for why Mobile Reddit. I first off, I think the mobile the mobile Reddit experience is more engaging and enjoyable than the desktop one. I think the desktop one’s still awkward. They haven’t totally figured it out. Yeah. Garrett’s right. They own I kind of am with you. I only really use Reddit on mobile. It’s it’s like a, you know, just it’s a scrolly, you know, entertain me experience, and Reddit contains enough links that when I’m really interested in an article or something or someone posts an article in the comments, I will click those. And so I think this number is because of exactly that behavior. It was one of those, like, surprised me initially. And then when I thought about my own behavior and how this works, I I was like, yeah. That makes sense. So fascinating to see those differences. You can see, by the way, this has been studied for a long time. I’ve published data about this that Google on mobile sends a lot fewer visits than Google on desktop. And this is, in my opinion, because on desktop, when you are researching things in Google, you’re going to lots of websites, visiting all sorts of things, popping it open. Google on mobile, you’re often like, jeez, how old is Idris Elba? He looks great. And then you you type that in and, like, you know, Google tells you that he’s disturbingly fine for his age, and it’s it’s really bothersome for the rest of us. I we we broke things out into social media sites. So these are basically all the social sites in the top one seventy. You can see Telegram and Nextdoor, you know, send very little injure right. Threads despite their tiny size and the fact that they just launched, growing. OnlyFans was interesting to me. I I talked to Geraldine, my wife, about, like, hey. Should I sign up for OnlyFans just to see what the experience is like to figure out what where they send traffic and how they send traffic? And we decided it wasn’t worth it. So I don’t know how OnlyFans interface works. Sorry, friends. If someone is familiar with them, probably don’t leave that in the comments with your real name. But Yeah. We totally believe you, Ren. Got it. Sure. No idea. No idea what OnlyFans is. I also I also did not search for cleaning brush on December twenty fourth. Fair enough. I I do think that it was particularly notable that how how hoarding how traffic hoarding TikTok is. TikTok is the number four or five, depending on your interpretation of what is a social site, most used in US. But look at that. They they are way down the list. They’re not even in the top ten for sending traffic out, and that is because TikTok to get out of TikTok’s interface, you have to do a lot, And TikTok does not inspire the sort of behavior of, oh, yeah, that whatever person making that cool painting or filming that adorable frog playing with a monkey. Probably it’s the other way around. Or whatever it is, you know, those videos don’t inspire people to click on a profile and then go to that person’s website, unfortunately. And and I think, frankly, as a marketer, I worry a lot about TikTok. I think people put a ton of effort into building audiences there that are largely valueless. Like, outside outside of the money that TikTok themselves give you, which they can turn off anytime, I get nervous. Whenever someone tells me they wanna be a TikToker, I’m like, oh, please. Anything but that. Like, build a newsletter. Go you know, even Instagram better. Search engine traffic, I looked at this one too. I thought this was particularly interesting. Right? Like, all the tech thought leaders, like, every time I post about, you know, AI or or skepticism around it, people are like, oh my god. You know, OpenAI ChatGPT is crushing Google. Like, I already use them ten times more than I use Google to solve all my searches. I I don’t see this. This is actually kind of a question I have for you, Eli, because sure you must get this from clients and, you know, I know we’ve done we did that, ChatGPT sort of what are people using it for. What what’s your two cents here? Well, I think one, there’s an adoption curve. Right? Like, you know, there’s like anything, people get very enamored, like maybe OpenAI or ChatGPT when it first came along. The second is, as we’ve noted here, this is purely about browser activity and not about what’s happening in apps. Right? So people are downloading the app, so I can’t speak to that. But I think in the end, you just forget Google is not just it’s like a portal. Right? It’s about how you enter the Internet on your devices these days. Right? It’s like the default search engine or on any number of browsers as well as, like, when you open up your device. And it’s just that type of behavior doesn’t change overnight just because there’s something new. Like, look at the size of that blue slice. Right? Like, that is that doesn’t change overnight. So that’s what I would say is that the embedded the dogmatic behavior, if you will, the inertia of that is, is very important. Someone brought up Melissa, yes. Very smart comment. This is referral traffic. So this is traffic to other websites, not total traffic. However Right. However, if you look at the total traffic for OpenAI, which I believe is back here all the way back here, it is it is teeny tiny. It’s zero point two one percent compared to Google’s sixty four point, you know, two seven. Oh, sorry. No. I do have the traffic data. Where do I have that? Oh, I I look at that next. Sorry. We’re gonna we’re gonna go But, Rand, and I’ll jump in, and Melissa makes a very good point. Right? Is that even putting aside, do you like, does one go on the web in order to be, like, refer me somewhere? Or you start with, like, a Google to be, like, answer you know, riddle me this. Right? Like, answer a question, and ChatGPT is sort of, like, answers a lot of questions, right, when you have them versus Google being, like, a big portion of your navigation. Right? Which is again, we’ll probably talk about that later, but the sheer number of navigational searches that happen on Google that people type in Google. Take me to Google. Right? So it’s you can’t forget just about, like, the flow of that about it’s not just answering questions. It’s like straight Divi directions, right, as well. Yeah. A little bit of bad news on this slide. Essentially, what’s happening here is the dominant players are getting a larger and larger share over time of all the referral traffic that, you know, everyone from Google to OpenAI to Instagram to Facebook is sending, which is sad. You know, if you map this out for for five or ten years, things look things look real bleak for anyone on the web who’s not in that sort of top few hundred domains. That would make me quite sad. But hey. I mean, I I think I have some conclusions around this, when we get to the end. So this analysis also included data. It was not just about referrals sent. It was also about traffic received, and that is what I’m I’m gonna dive into because referrals are not the only way to create influence. We are here, in fact, with the progenitor, the creator of, the concept of zero click content, my colleague Amanda. And, even though lots of lots of tech brows on LinkedIn will try and take credit for that, they they they did not, invent it. She did. So, alright. Here what I did, I I took essentially every site in the study, and then I manually categorized them into thirteen broad classifications. Right? So Macy’s and Nvidia and Samsung are all in ecommerce. Capital One and Chase and Citi are all in financial. Epic Games and GameSpot and Fandom are all in games. Cool. And then I broke this out into a who receives traffic. Like, where are people spending their allocation of visits on their desktop and mobile, devices when they’re in browser. Here you go. Right? It is social is number one, which should be no surprise to anyone. Social would be even bigger if I had included YouTube, but I broke YouTube into video and audio, because I really feel that YouTube is not the same as a social platform like Facebook or Reddit or Instagram. In search might you know, is obviously from our study driving the overwhelming majority overwhelming majority of of referral traffic, but it’s only the fifth most trafficked category. Right? So this is not where everyone is spending all their time. Google when I go to Google, I’m there to leave Google. Like, I I’m not there to, like, spend time inside of Google. I wanna get in and go to where I’m going to. But, you know, productivity tools. Right? Microsoft and Canva and Adobe and Open AI and GitHub and AWS and dozens of others. They’re all included in productivity, and that’s actually a big, big category, especially on desktop. I think people often minimize the importance of, like, we’re all at work all day doing work things, and those work things are in the productivity category. Right? Your email device. Right? All all all or your email application, Excel on the web, right, your Microsoft Office, all of your Google spreadsheets, whatever. So I did a comparison. I basically took every category, you know, travel, social, security, search, etcetera. And then the pink bar is sent referral traffic, and the blue one is visits, like like traffic. And you can see that with the exception of search, everyone gets a lot more visits than they refer out traffic, which which makes a ton of sense. Right? You don’t go to a financial website. Like, I’m not going to chase dot com to be redirected to somewhere else. Sometimes they will send me to other sites, but, you know, there you go. So I think this is hugely useful when you’re thinking about how do I create influence in places that are not gonna send me referral traffic. How should I be thinking about zero click content? Why is it so important? Yeah. Search, the only category that sends out more traffic than it takes in. Okay. So some conclusions here that I came to from this. I I think, look, smart folks like yourselves can form your own, conclusions. But if you are exclusively focused on traffic acquisition, like, if you if it has to be shown in your referral, then you’re gonna be stuck with search. That’s that’s kinda your only place to play. No wonder SEO and PPC are so big because if you wanna see in your analytics who sent you traffic, it’s always gonna be search. And that is not the same as where people spend time online, and that’s not the same as where they’re influenced. And this is a tough one. There’s just not channel diversity here. Right? Like, of the top ten, four of them are search engines, and and the overwhelming majority Google, of course. Alright. If you’re trying to prove acquisition via referral strings. Right? So the referral string is what passes from, in in the browser, passes to your web analytics system, whether you’re using Google Analytics or something else, and tells you where a visit came from. And that is gonna it’s gonna make you a chump for Google. You know? Not to mince words here, but essentially, almost everything that happens in all these categories, all these other categories, productivity and social and news and video and audio, it’s going to result in people doing this. And who gets the credit? Who gets the credit when that happens? It’s Google. And it does they they don’t even have to search for your brand. Right? They can be searching for something that happened. Hey. I saw this video, this great video about this thing. I don’t remember exactly who it was, but I’m gonna I’m gonna look for it. Oh, yeah. It was whatever. Chris Savage from Wistia. Oh, yeah. It was Eli and Rand when they did that webinar together. Whatever. Just it just kills me that all the attention that’s paid to all these other categories is not going to show up in your referral data, and therefore, it gets underinvested and especially by a lot of organic and content and SEO marketers. I think that’s frustrating. Okay. Number three, if you wanna win at social, if you wanna influence people through that, if you want that channel to be valuable to your marketing, you have to embrace, Amanda’s invention. Right? You’ve gotta do zero click because these sites don’t send referral traffic. They all these sites combined send about a seventh of what Google does. So you could be fantastic across, you know, fifty different social networks, and it’s still gonna look like Bubkis. I didn’t include it here, but in our in our studies on dark social referrals, a lot of these don’t pass referral strings too. Right? So anything that drives traffic from WhatsApp, a bunch of the traffic from LinkedIn, a bunch of the traffic from Facebook is not getting counted because of dark social. Alright. Last one. If you’re trying to figure out platform prioritization in in terms of social networks in particular, we have this feature in SparkToro that I would urge you to use or check out. You can if you’re like, I don’t have the money to sign up for SparkToro, fine. Email me or Amanda or support, and we will, like, run the report for you. But, essentially, you know, this is showing me that people who have the word graphic designer in their profile are, you know, on average using Instagram a little bit more than than the average web user and LinkedIn a lot more, Pinterest and Behance a ton more. And so I would prioritize those places where my audience is much more active. Behance Behance doesn’t doesn’t usually show up in the top twenty, but for graphic designers, it’s a great network. And in fact, I I did my job posting for, for my other company for Snackbar on on Behance. Got a lot of, illustrators applying for that. Alright. So, Eli, do wanna, in our last few minutes here, tell us a bit more about some remarkable uses for clickstream data and other things people can check out if they want? Yeah. Just a few pieces. And listen. You could definitely go to the dados dot live. That’s our website to our blog, and you’ll find a variety of content. And our content and I’m not gonna run through all of these individually, but the idea being is what are practical applications for clickstream data and going back to it as, like, an online intelligence database. Search is something that lends itself very well. Right? Understanding what people search for is a direct line of sight into, call it, like, you know, the thought process of the populace and how that’s different around the world or at different times of years or what it’s going to be. So I definitely recommend checking out some of our items related to Google and search because it’s just fascinating how people utilize search, and what it means. The next piece and and Rand talked a bit about Reddit. You know, we’ve also seen Reddit I mean, a, this is a blog post we published just last Friday, right, in relation to their, you know, their recent IPO, and they’ve been in the news quite a bit. But Reddit’s a very interesting type of website simply because it it has a different sense of, like, what represents, how it’s moderated, how the data is generated, how do people group together, and then upvote or downvote. It’s not necessarily just like, here is, like, an algorithm that’s, like, blasting you in the face, right, which is sort of like the the the TikTok style. It’s just like it’s gonna keep grinding on what’s most popular, but really having the marketplace make some votes. And and part of this, and again, you’ll see in the blog post is about referral traffic. Right? Like, how does Reddit play? You gotta remove, like, Google, obviously, like, as as we saw earlier, but different ways with which that you can start to think of getting your brand or your company or your client for that matter into the conversation in an organic way. Right? That isn’t just like, paid. Here are the ads. Right? So, again, I I recommend checking out some of that analysis. And if you go on to the next slide, Rand, you know, we also have been talking a lot about AI over here. Right? And I don’t wanna just jump on the bandwagon with it and be like, AI. Every company’s an AI thing. You know? Like, we have a you know? Here, my my my golf clubs are AI created or something. But in the end, if you think about clickstream data going back to it being what is, like, a lot of what’s going on out there in the world, like, again, that intelligence database, there’s a variety of ways that AI developers, both large and small, you know, could look at clickstream data to be able to help them feed into what are the patterns that their AI should be understanding and mimicking. Right? Whether it be a chatbot or product recommendations or whatnot. Because the context of things matters. Right? And clickstream data is definitely rich in context. And if you move on to the final, you know, slide and if you wanna keep up with any of the data’s reports, you click through this rant, you know, again, you could sign up. We have, a very low touch, you know, newsletter that happens once every couple of months. So feel free. Please sign up for our Datos newsletter or come to our website and sign up to be able to receive whatever it is you need. Awesome. Yeah. I mean, I am a subscriber because I I love this stuff, I care about it. And, also, I need to slip a twenty dollar bill, actually, with inflation, maybe a fifty to Carrie because I agree with you, Carrie. Right? Like, lot of the time when I look at Clearstream data, I’m really interested from an industry perspective, what’s happening, answering, like, big picture questions. But, also, I mostly care about my audience, And that’s, you know, that’s a little different. So alright. We do have, I think, a little bit of extra time for folks who wanna stick around and get some questions answered. I know there were some awesome ones that we got to during the presentation, but I think there’s a whole bunch more. And, Amanda, maybe you can take us through those over the next next ten minutes. Yes. Alright. Sorry about that. Okay. So we we actually did get a couple questions about Substack specifically. I think I saw some in the chat, and I saw some in the q and a. Does subset does Substack send any traffic? I’m sure they do. I don’t think let me check the I’ve got the Excel sheet here. No. I don’t think they’re in the top ones a hundred and seventy traffic refers on the web. So But, you know, Amanda, my opinion there is I think Substack gets consumed in people’s inboxes. Yeah. But unless somebody has, like, links embedded, right, in their newsletter, like, much traffic is that sending out? Yeah. No. I think I think I mean, this is true. I think for for, like, Mailchimp and, you know, the other big email providers as well is that that that content is hugely influential, and people consume a ton of it, but it doesn’t send a lot of referral traffic. And they they are also probably not I don’t know. Eli, do you know if Substack is in the top, like, couple hundred most traffic domains? I wouldn’t think so. I haven’t seen it. It’s been it’s more of a specialized place. Right? Like, I think that’s part of the idea behind Substack is the ability to be able to make money, right, like, your content, but that’s not going to have the same type of a, like, mass appeal, if you will. So there’s a what are you getting versus what are you giving up when it comes to particularly in referral traffic. We do have the data’s estimate of, traffic that they receive. Yeah. That’s not gonna be in the top. I don’t think that’s gonna be in the top five hundred most traffic domains. Makes sense. Okay. A question about data versus data. How diversified are the opt in users? How much is US versus other countries? And do you find that it skews younger or more in line with the general population? You know, diversification is really, like, an important word here. And I say that because it’s not as if we have one data source. Right? We work with dozens upon dozens of different types of data sources. And if there’s something that, you know, we think about this goes back to, like, the call it the mathematics. Right? And, like, the statistics that go about building panels. It’s about finding not only as many data sources as possible, but actually analyzing them for fidelity. Right? So that we could understand that there are going to be different datasets in different countries that are better or worse for different things. Right? Because we evaluate data sources all the time. Their their data too skewed in one direction, too young, too old, too specific, whatever it may be. So we have a particular way that we go about determining if the data is not only representative, but I’ll to be frank, has a commercial application as in would this data actually apply to what you know, into our algorithms and then the types of products that we build on behalf of the clients. So in the end is that if you think of us as an eater of world of data, what that means is that we are out there. We have a never ending appetite to go out and find new datasets. Right? And going back to that point about size and then diversification, and I had mentioned the data spread out across two hundred plus countries. And let’s call it, you know, probably the US is a very valuable one, so we focus there. Right? Because there’s so many Internet users in the US. So you’re talking, like, you know, several million, you know, people that wouldn’t be able to make that up versus others. But we have large sample sizes for pretty much any tier one, tier two, and even tier three country. And I say that tiers is in, like, commercial value of the data. Right? Like, you know, as far as that’s concerned. So it’s something that we focus on, right, is continually fighting new datasets to feed into that diversification. Are you, like, in terms of a demographic makeup, like, you might take a look at a group of people and say, oh, this data source contains more older demographics. So we are going to when when we’re trying to do, like, a, an estimate of overall behavior, we’re gonna sort of shrink that dataset’s, impact on the on this particular analysis, or we’re gonna weight it differently in order to get, like, a you know, what we think is the truth for for a a region or a country. Well, I’ll tell you what. Right? There is the answer is yes when it comes to understanding bias and then what is, like, how to weight the scale, if you will, and when it comes to, like, datasets or subportions of datasets. But here’s the thing. We are one hundred percent anonymized. I have no demographics on this tool. Right? So I don’t even have a way to be able to do that. What happens is that we actually are able there’s a variety of public sources that you could, like, compare data against and then we can just sort of, like, you know, gut check it, if you will, to be able to determine what’s what. But we focus on anonymized behaviors. Right? So I don’t even care. It’s more about I think of things as people that buy gardening tools on Amazon. Right? Again, Rand, you guys have a product that sort of thinks in those terms. And that’s really how we think of things as well. So that’s what I’m getting at is when we talk about offsetting bias or diversification and talking about our fidelity metric. Stability is also an important one too. Right? It’s the ability to understand panel tenure and, like, churn rates and things like that. And then as a subset of any given panel, which users or which groups of users more to the point are more representative. So that goes back to your point around the the waiting, if you will, to be able to determine, what is most predictive. So, yes, we will work on different subsets of different types of people because we our clients will come to us and say, I need this type. I’ll be like, hey. Let me go see if I have those. Right? Like, in some way. But it’s gotta be anonymized based on behaviors because I just don’t Yeah. And so this is a good this is a good point, right, that I mentioned Datos is one of the primary data suppliers for SparkToro, but the you guys are not behind our demographics because you’re anonymized. The demographics data comes from matching up data and Google search data and social profile data with LinkedIn. And LinkedIn does provide on public profiles if people choose to specify their their gender and age. That is where this data comes from. And so when we create these that you see, for example, for Substack, oops, for Substack, this is not Datos. Just just to be clear, Datos is anonymized. Right. Alright. We have another question here. Right. Can you explain dark social slash traffic, and can you list a few of the common dark sources? Yes. I can. Although, I will tell you, I’m very frustrated that, SparkToro doesn’t rank well for for dark social traffic. Okay. So here is here is the blog post that you should check out if you’re interested in dark social, but, essentially, this is the graph that you really care about. And what it’s showing is each let me blow that up a little bit. Each different source and whether it passes referral data, we tested these, across hundreds of clicks with a small group of people, a lot of people like yourselves, right, like the people on this webinar. And we were like, hey. Will you go to this TikTok profile from essentially an an email and then, please, you know, click on the link in the profile, and then we’ll see if we record the visit. We did record hundreds of visits to the TikTok linked URL, and no TikTok referrer string was included. So a hundred percent of the time, every visit was attributed to direct in the analytics software. Same is true for Slack. Same is true for Discord. Same is true for Mastodon. Same is true for WhatsApp. Facebook Messenger, weirdly, sometimes occasionally sent a referring string, sometimes Instagram messages, sometimes LinkedIn posted, sometimes Reddit, sometimes Pinterest. So quite frustrating, but you can you can get a sense that, essentially, everybody except a Facebook post and a Twitter post had some percent of dark social. It’s variable. It changes over time. I’ve seen times when YouTube passed no refer string. I’ve seen times when certain, Google the Google app I’m not sure if I can show you this on my phone, but so this app down here, this, Google search app is essentially, sometimes completely referral list, which is infuriating. So and that one’s called dark search. But dark social, big problem. You know, if if you’re getting a bunch of, for example, lots of b to b software companies, Datos, I’m sure, gets talked about all the time in Slack communities. So does SparkToro. And when people click that link, it passes no refer string, and we never know where that visit came from. It gets marked as direct. There you go. Alright. I think we have two more questions. Alright. Do you think citations in Google SGE will generate traffic? What an outstanding question for doctor Pete from Maz. I don’t know I don’t know if you have an opinion on that, Eli. Oh, I do not. That is beyond my technical capabilities today, to be honest. I don’t wanna front that I won’t be able answer that. Let’s let’s hit up doctor Pete on threads and ask him, and, and we’ll we’ll see if we can get a a reply. My opinion is maybe? I don’t know. Yeah. Question mark. You know, what what would what would be interesting, Eli, that we could do long term is if search generative experience comes to Google, we would be able to, in the data status set, see whether, we could take a look at a bunch of keywords for which SGE comes up. Right? Because because one of the data sources for SparkTorre is Market Muse, and then we could send that to you and say, hey. Will you tell us if these citations in SGE led to referral traffic or led to clicks or searches? Yeah. If it shows up as a parameter, right, like, know, within the dataset, that’s a question mark that I have right now since there’s pretty much an unlimited unlimited number of permutations of clickstream format. But, yes, that is in theory if there’s a way to be able to break it out like that. You know, we’d love to be able to analyze it. Yeah. So possible. It it’s really cool what you can do with the clickstream data and other data sources to answer questions like that at a at a broad scale. Alright. A last question. This might be kind of fun. What do we think about WebMD’s referral stinginess? Think it’s intentional? I’m just curious what everyone’s theories here might be. Okay. I’m gonna I’ll tell you what I broadly think, which is when you look at this list, what I see are a couple of I’m gonna call them made for ads websites. GameRant, in particular, is hated by the video games community because they sort of intentionally send no traffic. If you click on any link from a GameRant article, it will take you to another GameRant article. My my my cofounders and I at Snack Bar Studio have complained about this bitterly. But Healthline and and WebMD and Mayo Clinic are all in this group, and I think part of it is a sort of a health safety thing. Like, they don’t want to send traffic to anyone else who might have any advice that is not a hundred percent medically safe and sound. And that that that is my theory. I think there’s, like, a legal aspect to these health companies hoarding traffic and not referring out. But I don’t know if you have different opinions about these these big traffic hoarders, Eli. No. I mean, I would absolutely agree on that. And I think also, though, when we talk about traffic hoarding, think of it that a lot of these places that you see on here are, like, the end of the line. Right? Like, you know, it’s in you know, that is the end of your session. Right? And maybe for oftentimes when people are looking for things that are, you know, health related, they get the answer that they were looking for, right, good or bad, and then also probably stay within that universe, right, within those walls, like researching other things. So it probably I imagine it links to itself quite a lot. Right? Like, this is related to this is related to this. But my feeling on it behaviorally speaking is that people that are going to those websites are pretty motivated to be able to get answers to something that they’re thinking about that is either, you know, like, on their mind, whatever it may be. So it tends to be that’s why I’m calling it, like, the end of a journey, like, probably a special specialized one as opposed to one that is, okay. I’m gonna do this, and then I’m gonna go shopping. Right? Like, you know? Yeah. I mean, Airbnb. Right? Airbnb is in here, and that’s clearly the end of a journey. Not you know, you’re not going to Airbnb to go find another hotel or, you know, a a third party website. Yeah. That makes sense. And then you also think about the kind of people who would be getting their information on WebMD or Healthline, for instance. Like, using myself as an example, I think I I would have probably pretty average consumer behavior here, which is I might search for, oh, what does L theanine do? And I’ll find a Healthline article, and I’m like, oh, cool. Didn’t know that. Like, I’m not a researcher in health that I’m gonna, like, make decisions and use that as my research. It’s, no. I’m a consumer who wants to know this thing. I’m not the researcher or the clinician who is kind of, like, you know, too advanced for that type of website. So yeah. Yeah. That’s a good point. Yeah. The end of journey theory, I think, is is really smart on a bunch of these. I mean, UPS and FedEx are that’s like a place I go to to whatever print out my shipping label, and then I’m done with my journey. So yeah. Yeah. Well, friends, I just wanna say, first off, a a huge, huge thank you to Eli and everyone at Datos. They didn’t charge us a dime to produce any of the research in the study. They they gifted their team’s time, and I know it was a lot of time and energy that went into putting this together. There were there were, like, half a dozen folks from the Datos team that contributed this report, in addition to myself, and just extraordinary, their generosity. So, Eli, Datos, thank you so much. I I can’t wait to do this again with you, and I’m thrilled that so many people were excited about this as we were. Amanda oh, thank you for for your contributions and for hosting us and putting this all together. And where where can people get the recording? The recording will be available on this very page as soon as we leave, and then we will also email it to you along with some useful resources. Thank you, everyone. Thank you so much, Eli and Datos team. Eli. Been amazing. Everybody. It was great. Awesome. Bye, friends. Take care.
By 
