Video: What 332M Searches Can Tell Us About How Google Really Works | Duration: 3800s | Summary: What 332M Searches Can Tell Us About How Google Really Works

Transcript for "What 332M Searches Can Tell Us About How Google Really Works": Oh, Eve. Just one of those days. Yep. Showbiz. Man, I feel like an elderly McDonald's patron ratted me out. Just playing. Just playing, people. I mean, a lot to unpack with Yeah. Stuff lately, but we don't have to unpack it. I really here's what I want. I wanna know what the monopoly money in the backpack was about. That's that's like a you know? I hadn't heard about that yet. Well, so he left he left his backpack in Central Park between two boulders, and inside, it was filled with monopoly money. And so the police, like, took it to a lab because they were worried it might have explosives or whatever, and then when they opened it up, the only thing that's in here Matt, this is a good good suggestion. Yeah. I the monopoly money thing, the weird weird gun from, like, the 19 sixties or something, the, the bullet casings etched, the getting caught by an old lady at McDonald's. It's all just too wild. Too wild. It really is. This season on America. It just doesn't sound believable. Nope. Nope. The script the you know, when the Hollywood editors are sort of going through your script, they're like, look. This is this is really exciting. It's quite good, but there's no way. There's you gotta make it more realistic. Right. The audience isn't gonna believe this. You know? Like, monsters coming down from space, they're in. But this, it's just it doesn't it's not gonna happen. No. I I do have to say, though, the the comments or, you know, the people kinda thirsting over this guy are have been pretty funny. Like, the casings didn't say married. Wow. Oh my god. That's Or the amount of Wow. The amount of men I've seen, no playing along and saying, I can fix him. Oh god. For those of you who might be joining us for the first time and are not aware, at SparkToro office hours, which welcome, by the way, Amanda and I typically have some, irrelevant banter to start off and give folks a chance to join. Many many folks we find join within the first 5 to 7 minutes, but not necessarily minute 1. But we do have an incredibly exciting webinar for you today, and it will help ignore all of the bad news that might be in your news feeds. So, Amanda, do you wanna tell, do you wanna answer our most commonly most frequently asked questions? Yes. So the most frequently asked question is usually, it's a joke that I'll say, but I won't. It is most people ask about the recording, which, of course, will be available to you very shortly after the webinar ends. And if you need to reaccess this recording more easily, you can even just go back to your, like, reminder email that says enter the event. That's a great way to to find us again. And then at some point in the next several weeks, this will also go up on YouTube. But just so you know, the recording will be available as soon as possible, and I'll also I will also email it to you usually later today, sometimes tomorrow. Awesome. Wonderful. Amanda, a long time ago, in the before times when things were okay and not terrible on Twitter, we would invite folks to use the Twitter hashtag for this event, which is SparkToro Hours. Where where where should they do that now? Where's the Well where's the action at? Good question. So now you can still use hashtag SparkToro Hours, ideally on LinkedIn, probably, also on Blue Sky and threads if that is your jam. Love it. Love it. Oh, I'm so glad you asked, Joe. This mug, which I can't remember the name of the woman's pottery service, but it's a a woman who does pottery in Ashland, Oregon. And when I walked by, I was stunned by how beautiful it was, and I picked it up. So let's see here. We have got, an incredible amount of data and information to go to through, and so I'm going to present my screen. And, when I do I love how they change where the screen share function is every time. Top left today. Yeah. Today. Tomorrow. Who knows? Alright. Whoo. Hide this guy. So, thanks to the generosity of our clickstream partners at Daedos. I'll talk a little bit about clickstream in here just so everyone has it, but, thanks to their generosity. We have some absolutely incredible data to share with you and, certainly, I invite you to, you know, whatever you want. You can take screenshots, you can share this stuff. I will we will also, make sure that the presentation is available, but if you'd like, we've got the blog post up which has the overwhelming majority of the graphs and graphics that you'll see. So you're welcome to steal the shots from there. I know a bunch of people have already mentioned to me that they're gonna be using them in their sort of next year strategy decks and presentations along with, hopefully, their SparkToro reports. So, Amanda, why don't we talk about some data sources real quick? For those of you who might not be familiar with what clickstream data is, so this is basically clickstream is one of the 3 things that powers, SparkToro, 3 or 4 things that power SparkToro. Essentially, what it is is it's, you know, millions of people, millions of devices around the world, including many millions here in the United States, both mobile and desktop where, the through some sort of partnership or data sharing agreement or, you know, like join and we'll give you free HBO for a month or, you know, whatever it was. Companies like comScore and, Nielsen and Datos, who we use, similar web as as well, they collect every URL that the device visits. If they open any browser, they see the URL that is visited. This is how they get things like, oh, we can tell, whatever, darn tough socks that 30% of their, Amazon visits to their Amazon pages end in a visit to darn tough's own website to do a price comparison. Right? And the Darn Tough people might be like, Wow, that's amazing. We really want that data. So Datto collects all this data, and then they can provide to us all sorts of amazing stuff, every URL visited, right, in order. They can tell us keywords that they search for, in Google. This is all real humans, no bot traffic because you don't have any, bot services or you have only real devices. There's no sample bias except for the panel itself, which admittedly, that's one of the limitations. There's no mobile app data. It's a little bit like being a Neil a Nielsen family, but for the Internet. Exactly that. Exactly that. It's like being a Nielsen TV family in the 19 sixties. There's no mobile app data. So if you for example, you know, if you use the, no. Yeah. Let me just share my passcode with everyone. You know, if you go down here and you use Google's search bar inside their search app, you will not get, data until the browser is opened. But if they search inside mobile Chrome or mobile Firefox or anything like that, you will. It won't include automated requests, which can be good or bad. Many people are interested in, like, how much search traffic comes from rank trackers and people crawling and SERP APIs and that kind of stuff. Well, unfortunately, DATOS has no information about them. These are all real humans. And there's no demographic breakdown provided. They do have demographic data that they can provide if you pay for it and something, but for this, no demographic data. So today's webinar is about these 4 big questions that I had around how people search. Now for for those of you who don't know, SparkToro is not an SEO company. My my previous life was in SEO professional life. But like many people in digital marketing overall, we recognize that Google sends an overwhelming majority of all the traffic on the web and that knowing what happens inside Google can give us incredible insights into how people behave on the web. And so Google search, very interesting at the macro level even for us at SparkToro. You probably saw that this year, SparkToro started providing, keyword search data on audiences. So, you know, if you search for an audience like electrical engineers or, VPs of product or people who work in HR or people who search for drones, whatever it is, SparkTor will tell you things that they search for in addition to all the other, data. Once in SEO, always in SEO. I don't know. I I still have I have beef with every time someone puts me on a top SEOs list. I'm, like, you're that's not a credible list. I haven't optimized a website in 7 years. Okay. Big question number 1. Are large language models challenging Google's market share? Well, the DATOS thanks to DATOS, we are actually able to get pretty darn close to this. So this is, from a webinar I did with DATOS earlier this year looking at perplexity, which has grown rapidly. It is it's it's not big. Right? I think it's in the, maybe top few 1,000 sites, but not top few hundred. So it's it's popular, but it is not what I would call mainstream popular. It's much more popular in tech world, people on LinkedIn, that kind of group. You can see that ChatGPT, on the other hand, is globally popular. More visits than Bing by a considerable amount, and Bing is quite a popular website. And in terms of web page visits per month, you can see that these LLMs are getting, you know, not super dissimilar, pages per These are web page visits per month. This is essentially saying that people who search on perplexity on average will visit 16 and a half perplexity pages. There is growth happening there too. In chat gpt, it's 13.5. In Google, you know, full disclosure, it's like 200. So many more, you know, searches per month for a searcher. In terms of people who use AI, tools, you know, you can see here that, well, almost every single person who uses an AI that that DATOS can see visiting whatever chat gpt or Gemini or Perplexity or, Claude or anything like that. 99.8% of them also used a traditional search engine. And I tried to graph this out. This is a graph that combines several data sources, so I I wanna be a little careful. I wouldn't I wouldn't fully quote this, but if you were to treat every single prompt in chat gpt as a search, this would be approximately what you would get in terms of the monthly searches. Right? So chat gpt would be technically larger than Bing if you believe that every single prompt put into chat gpt is the equivalent of a search. I don't believe that at all. I don't think that's a reasonable assumption, but this would be the maximum size that you could possibly get. In that, equation, right, if you were to this graph unfortunately made its way around the web, in a frustrating way, at least frustrating for me, which is I wish that folks had been more careful when they shared it, and I should have been more careful when I shared it. See this little note at the bottom where it says, Assumes all LLMs all LLM requests or searches. I really I really, really, really should have taken that out. By the way, YouTube is still almost twice as big as Chat gpt even if you think every single, prompt is a search. So, we can start to get to the answer here by looking at Google searches per searcher because if this number is going down and LLMs are going up, we could reasonably say LLMs are cannibalizing web search. That's not what's happening. It is staying stable or seasonally adjusted going up. Right? These numbers are rising. And, by the way, for those of you who are thinking, oh, there's so much opportunity in LLMs, well, there might be some. I think I saw that the New York Times grew the number of visits it received from, chat gpt and perplexity from, like, 10,000 last August to over a 100,000 this August. Now if you're in the New York Times, a 100,000 visits, probably not all that meaningful, but it's 10 x what it was last year. If it's 10 x again next year, maybe maybe it starts to be meaningful. Llms. Sorry. Llms, large language models, AI tools. I I don't personally love calling them AI tools because they're not AI, but they are large language models and the press likes to call them AI and whatever. So we I use those interchangeably. Apologies. So our verdict on our first question, nope. Doesn't look like it. So far, LLMs growing fast. There's lots of use of them. Not all of it is for search purposes, but they are not making a dent in Google search usage. We are not seeing a decline in Google. We are seeing, an increase in in 2 key metrics, searches per searcher and total number of searches per month seasonally adjusted. So there you go. Number 2, Amanda and I talk about this a lot. Amanda, for for those of you who don't know, coined the phrase zero click marketing, and zero click content. And Google search was was sort of the first place where we noticed this. I think it was actually Gabriel Weinberg from DuckDuckGo, like, a decade ago coined the term zero click searches. And, you know, you can see in here essentially that traffic referral when it comes to traffic referral, Google is the overwhelming majority. Right? So 63% of all this is from Datos' data. We looked at, like, millions of sites where their traffic comes from, all the referral data traffic that they get. Where does it all come from? The answer is in here. Even if you were to exclude the top 200 sites, the sites that get the most traffic on the web and just look at the longer tail of all of our businesses and websites, people that you might be helping out, not even small businesses, but but big companies. Google is 72 a half percent of all those referrals, and then it kinda drops rapidly. 75% of all traffic is coming from search, 14% social, 11% sort of everything else. This is this is just traffic referrals. This is not where people spend time. This is not where people consume content. We are not talking about where influence happens, which is not just in these places. We're just talking about traffic here. Alright. So Google is still sending an overwhelming majority and we looked just a couple months ago at what happens after people search and some really interesting data in here. So for Americans, 41.5% of us make 1 or more clicks on our on our Google search results. Right? We perform a Google search, then we click something. Maybe we click 2 somethings. I think on average, we click 1.15 times or something on on different results. And you can see that 70% of those clicks go to organic results, 28, almost 30%, go to Google themselves. It's it's good to be the king, and 1% to a paid ad. Those of you who are in paid search who are listening to this right now, you might be deeply frustrated. You might be like, there's no way this data is right. I regularly get 5 or 6% click through rate. Well, hang on. Just hang on a second, and we'll get there. 37% of all Google searches lead to nothing. People are satisfied with the answer they get. This is combining mobile and desktop. It's higher on mobile than it is on desktop, but, we're we're combining those here. 20%, go for another search, 21%. That is what we call 0 click searches. No one got a click from that query. For every 1,000 US Google searches, 360 clicks go to what we call the open web, like not a Google property, not something owned by Alphabet, but somewhere else on the Internet, paid or organic. Paid ads, according to Google, appear on 20% of searches. So if 1% of clicks go to a paid ad, that means paid ad click through rate on average is, hey. It's that same number that you said you regularly get. Right? 5%, 6%, 7%. Okay. When PPC ads are present, it is probable that 5%, 6% of, searchers click on a paid search result. Okay. What about Europeans? Really similar story. And this is interesting to me because the EU has much more regulation around self preferencing. Right? If you've been in Europe recently and you did a search for hotel in Madrid, you will not get Google Hotels right at the top. They they do not allow it. They're not allowed to put themselves above everything else, and yet it barely affects the percent that goes to a property, a Google owned property. So if you're hoping like, well, you know, maybe regulation could help here, I I don't know. At least not the regulation that's been put in place so far. Google is really good at capturing Google's own search traffic. So here's that comparison, which you can see. And, unfortunately, the bad news is that this was done before the big AI overview rollout. So we did this study through to June of this year, 2024. We released it in the summer, and then Google was like, AI overviews are back, and they're more dominant than ever. So when we do this again next year with Datos and I've talked to Eli, about doing it again next year. My expectation is these numbers will look worse. You might also be aware that a few companies are dominating the top results. That is also shrinking the traffic pie, making it harder. This data is quite interesting. This is a graph looking at referrals to the most popular sites on the web, which which I've, highlighted in pink here, and then the long tail of the web, every everything outside of those sort of top traffic earning sites. And you can see that over time, over just a a course of oh, this is supposed to be 2022. Sorry. I'll fix that in the in the graph. Over just a year's time, the percent of traffic going to the fat head, sort of the dominant websites, has risen 3, 4% already. So big sites are capturing more and more of the pie of Google's share. And I think this is a trend, you know, we're gonna see for a long time. I just made this chart over the weekend. This is looking at the query volume from our new panel, the panel that I'm gonna present data on for the rest of this presentation. These are a 130,000 US searchers US devices, mobile and desktop, that where Datto's provided me with every keyword that they searched for, over the last, I guess, 21 months. Right? For this year period, August to August, you can see that searches seasonally you know, there's a lot of seasonality in search volume. So it tends to be the case that around, you know, holidays, people search more. October always sees a search spike, which I always think is Halloween, but I'm not actually sure if that's true. I, I would have to confirm that. But yeah. Oh, hey, Amanda. We have the data. We could go confirm that. Should we go look? Well, I can't look live, but I want to see February March, are those pretty typical dips? Like, is that a typical seasonal thing that we see? February for sure because it has a shortened month. Right? It's, like, 3 days shorter. But, March, I'm not sure about. March is a little surprising to me. I'd I wanna go back and look at more Marches to see if that seasonality is there. So the verdict on this is searches per searcher, like we talked about, is probably growing. Total number of searches, also growing, but total searchers is not growing. There there are no more new searchers for to be captured. Right? Like, 97% of the United States population is online now. We just there's just not that much more. And so zero click searches and the bias to the top sites is probably what's driving down that opportunity. And, certainly, for anyone, you know, if you care about your own traffic, what I would encourage you to do is look at things you know, look at your sector, look at your searches, See where those zero click elements are taking share from you. I know lots of people have been affected by that. Alright. Big question number 3. What percent of searchers already know what they want? So this would be people who basically they're not going to Google and they're searching for audience research tools. They just go to Google and they search for SparkToro. And so other other brands don't really have an opportunity to capture them, Or even people who go and search for, you know, they're not going to Google and searching for, you know, which streaming service should I subscribe to or best streaming services 2024. They just go netflix.com in the Google search bar which which a lot of people search for, by the way. Yeah. That seems really hard. Does that seem really hard to you? It was if I knew how difficult it was gonna be when I started this project, there's no way I would have done it. I cannot. I poured so many hours and weekends into this. It was just, so here's what we what you can't do what I wish what you could do, and the simple way to do it is is is a way some people have done it, in the search marketing field before, which is essentially to take giant corpuses of keywords or websites and then use those. But but that's not what you want what you need. What you need are people. You need individual people and a big panel of that. So what I asked DATOS for was every Google search from every US device in their panel over the last 18 months. They said no. They were like, what? No. You wouldn't even be able to handle you wouldn't be able to handle the data. I was like, that's that's fair. You guys are probably right. I would not 100 of millions of rows would really, really impact me. Okay. But what they did give me instead was all keywords that had more than a 100 searches in those, it was actually 21 months. I got 3 extra months from them after this. From From a 130,000 active US devices, both mobile and desktop, like I mentioned, and Datos did this nice thing. It's quite nice for the keyword analysis. It's a little frustrating for the volume analysis, but they took so for example, Amanda, if you searched for, I don't know, let's say, headphones, best headphones, 6 times in a day, like You kept going back to Google and searching for it again and again, which many of us do. There's a lot of duplicate searching. I do this all the time for a recipe. I'm like, oh, I'm making whatever stuffing recipe. I search for it 5 times. I search for it when I go to the grocery store in the morning, when I come back. Those are all treated as one search, one device searching for that thing. Mhmm. Duplicates basically shrunk down to a single one. Then step 2 was that I took a thousand of those keywords, and I manually classified them into different buckets, brand versus generic. Right? So this is human. Like, I am classifying. I'm saying Netflix is a brand. I'm saying best streaming services is generic. Navigational, informational, commercial, transactional. So the traditional sort of search intents. Right? Navigation, I want to go somewhere. Informational, I want information about this. Commercial, I'm interested in this product or service. And transactional is take me to a place where I can buy this thing right now. I'll show you some examples of those. Then categories like vehicles, lifestyle, travel, and tourism, and also volume buckets. I didn't have to do that one manually. Then, I try to see if an LLM can accurately classify these, and this is what took me just forever. I'm happy to give you all the prompting and stuff so you could do this yourself. You could essentially classify using chat gpt. And I use a product called GPT for sheets, which I think is excellent. You can use it inside Excel or Google, Google Sheets, and I was able to get to 96% accuracy. Thanks to the help, well, of me banging my head against the wall for forever, and Brittany Mueller, who had a call with me and then helped me over email. She runs an AI, class for marketers. I would highly recommend her work. She saved my toughest on this, and, I was really only able to do it because of her help. I think I would have given up if not for her. But at last at last, we can answer branded versus generic. Also, I'll I'll show you examples. Right? Yahoo Fantasy Football gets 15,823 searches, over these 21 months from these 130,000 searchers. Fantasy Football gets 5,000. Darn tough socks and socks, pretty similar. Navy Federal Credit Union and is almost 3 times the size of Credit Union. Rick Steves, much more popular than flights to Italy, which kind of blows my mind. Yeah. Incredible. Amazing. Okay. So then, by taking all of these millions of keywords that are in here and classifying them all, I can see that as a percent of keywords, branded keywords are 31%. People searching for brands, that's about 1 third of all the search terms, but not volume. Volume, it is almost half. Almost half of all searches, 45% of all searches are for a brand, not a generic thing. This is in every field. Right? So we're we're talking about, like, in arts and entertainment, in vehicles, in, you know, outdoor equipment, in what you name it. B to b, b to c, all of it. 44 per point 19 percent brands, the rest generic, which is wild. Woah. Britney's in the comments. Oh my god. Yeah. Oh, it's like having a rock star at your concert. It is. I like You wanna come off on screen? In my head. I was like, oh my god. If I say anything wrong, Britney, you just you just correct me. Okay? So, 44% of Google searches are for specific brands, which is which is quite remarkable. Then I was able to classify this was the hardest one. This is the one where, actually, Britney, you helped me a a tremendous amount. The key here, by the way, is not to play with the prompt itself. It's to play with the way that you name the classifications. If you just say classify into navigational informational, that doesn't work as well. But if you tell the LLM, is this search navigational meaning that people are just trying to reach a specific website or informational people are searching for information about this topic they are interested in it. Commercial they have a potential interest in a product that they might buy or research or compare. You know? Then that works really well. It's the sort of specificity of how you name the categorizations rather than the prompt itself, which, again, wouldn't have figured out without you. So here here we go. Navigation will be something like TikTok. That's people just going to TikTok. Carl Jung, that's informational. They're searching for information about the the philosopher. Verizon Business, well, they're not necessarily gonna buy something, but they're almost certainly interested in the Verizon Business Services. You could you could potentially treat that as navigational, arguably. I think that that's a reasonable thing. I classified it as commercial, and so did the LLM in this case, so that worked. And then, pha near me. That's someone who wants pha right now. They're gonna buy it. That's a transactional search. Quotex login, obviously, synonym. Right? Looking for a synonym for obviously. So You get the sense of what these kinds of searches are, and here's the distribution. Transactional, a very small percent of all the keywords. Informational is an overwhelming percent of the keywords. But check out what happens when we instead, and I think this is the more interesting one, go by volume. So not volume of keywords but volume of demand. There you go. Navigational almost triples and becomes a third of the pie, and informational shrinks to about half the pie. So navigational, a third of all activity on Google, things that people search for is just take me from google.com to this other place that I wanna go. Right? It's people searching for Netflix. It's people searching for TikTok. It's people searching for Navy Federal Credit Union. That's a navigational search, and it is almost a third of all Google searches. If you look at the top 100 Google searches, the ones that were most searched for by these 130,000 people, they're overwhelmingly navigational. Just, you know, no no contest. People go to Google to get to somewhere else. Got it. And this makes sense because our browser bars are also Google. So unless you type in, you know, correctly sparkporo.com and you instead just type, you know, sparktorospacecom or space sparktoro.com or sparktorocom, Google will give you a search result that will take you to the website. Many people also use this to be safe, right, so that they don't mistype a URL and get fooled and and go to the wrong place. So the verdict here actually is mixed. I think, you know, half of Google searches over brands, a third are navigational, but there's still 1,000,000,000 and 1,000,000,000 of searches that are sort of informational, transactional, commercial. Well, transactional is is quite small. That's where the the most value is, but quite small. And so I honestly was surprised. I was expecting navigational and brand to be even bigger. I know many people who are in search marketing expected it to be much, much smaller, so this is surprising on on both sides. Last question, the distribution of topics that people search for. This one's quite fascinating, and I have a bonus section for you at the end here, but, things that people search for the most. So I essentially used a very variation of the Wikipedia and the similar web classification for domains on the Internet. You know, everything from home and garden to law and government. I can tell you, the LLMs were much better at this than I was. They are not only more efficient, but they are also more accurate. Anytime I hand classified one of these, and then the, you know, chat gbt was like, no. No. This is science and education. I'd go to the search, and I'd look at the Google search results and be like, oh, yeah. I guess they are looking for science and education. Hey. No. That was impressive. So here's the distribution. Arts and entertainment dominant, like, absolutely dominant. This is, percent of keywords. Again, so I'll show you percent of keywords first and then percent of volume second. And you can see arts entertainment. Computers and technology, which, to be fair, I think the reason this one is so high is because so many people search for Internet related things. Like, people are going to their Slack, and they're searching for Microsoft Office and Google Sheets and Google Translate and Netflix and you know, like, well, Netflix is probably an arts and entertainment, but, many of those thing many of those websites that you go to are computers and technology, so not a lot of navigation in there. Volume, actually pretty similar. Right? So arts and entertainment, computers technology, ecommerce and shopping, and and on down the list, I was quite surprised. I thought there would be more recipe searches, and so I thought that down here in food and drink would be smaller than 2 a half percent, but there you go. I don't think as many people cook at home as I think cook at home could be. So I tried to be like, well, what percent of these are of high commercial value versus lower commercial value? And outside of, you know, essentially, advertising, like like CPM display advertising, which is most of these sectors, around around a third. Yeah. I I would say around a third is relatively low value. And, you know, this is this is the arts and entertainment category, right, where people are at the bar, and they're like, did you know Paul Rudd is 55? And they're like, no way. He's 55. That dude looks 35 at most, and they all search for him. Unbelievable. He's 55. By the way, in case you were wondering, this is Geraldine's favorite statistic from the study. 0.00184 percent of all Google searches are for Paul Rudd related things. That includes Paul Rudd age? Yes. Yes, it does. Okay. Now we know. Yeah. Which is a very popular search. People are very confused about how old Paul Rudd is. They can't believe it. There's lots of variations on that search term, And since we have the data, I actually put together a bunch of distributions of keywords and volume. Will Critchlow emailed me, last at the end of last week, the the founder of Distilled and now, SearchPilot. And Will was like, hey. I think you should put these, you know, all the, volume buckets into equally sized buckets and display it this way. So I made a bunch of graphs for him, and now y'all get to benefit as well. So we took the 332,000,000 Google searches across 21 months from our 130,000 panel, which which this comprises 334,000 unique keywords. So 332,000,000 searches, 334,000 keywords. And then the combined volume of those searches, I made buckets so that everything is about 42,000,000 in terms of, how much demand, right, how many times they were all searched. The top 100 keywords are responsible for one of those buckets. Top 100, 42,000,000. Next 1000, 42,000,000. Next 5000, 42,000,000. So this is essentially showing that while the long tail of Google still big, still exists, the fat head Like, the the head is really dominant. The top 100 queries have as much volume as the bottom 200,000 combined. That is, I strongly suspect based on, you know, my years in SEO previous to this, that if you were to run this this this exact same analysis in 2005 or 2007 or even 2012, you would see a very different story. I think what has happened is that the the US economy and brands and just the world has kind of gravitated toward these big powerful domains, and and we are less unique on the web than we once once were. I'll also I I also took this, and I broke them into buckets of 10,000. So each of the 334,000 keywords, I I took the most popular 10,000, the second most popular 10,000, etcetera, and then I graph that, and you can see that the top 10,000 keywords are almost half of all searches, and 75% of demand is just the top 20% of search query terms. So anything that was searched for more than a 1000 times by the 130,000 device panel. It's the head. The head of the demand curve is absolutely huge. And I think if you're if you're someone who's trying to grow a small brand or build a brand, Google might not be the opportunity that it once was. This is this is my big takeaway from this. I also graphed it in a few other different ways. This one in particular is interesting because I mentioned that these this this analysis and this one are both the 332,000,000 Google searches over 21 months. But for September, only September of 2024. Right? Just September 2024. Daedos, I think, maybe by accident, but I didn't question it, and they didn't give me a hard time about it. They gave me every search, including searches that were only performed one time. I I am not sharing those. I deleted them. Like, I was responsible. I didn't wanna have any whatever issues, but but before I did that I was able to make this graph and and do this categorization which is super cool. So these 772,000 terms on the right here, those terms are searched for only once or twice. That's essentially that bucket and then, you know, going up and up and up. So the, this distribution looks a little bit different, but single search keywords are 59%, almost 60% of all the unique query terms, but only 2% of demand. The long tail, which, you know, is historically like, we've always thought of Google as being having a huge long tail, Yes. It's huge in terms of its length. Like, the number of terms in here is massive. It dwarfs everything else, but the volume, the combined volume of those is is minuscule. Right? Alright. So let's, let's talk about some some takeaways here, and then we'll do some q and a. For people who focus exclusively on traffic. Right? If your boss, your team, your client is like, I don't care about influence. I don't care about growing our brand. I care about you getting visitors to our website. You're kinda stuck with search. Right? There's not channel diversity for you. Search is still referring the overwhelm majority of things. But I am gonna point out this is from Will Reynolds at SERE Interactive who who presented this graph on stage at Brighton SEO, and I thought it was very brave because he stood up there and he said, I took over our marketing efforts. I I will took over our marketing efforts. I will not imitate his voice for many reasons. He, he said, I took over our marketing efforts and traffic fell almost 40% over a year and a half, and revenue and new clients went up 20% because they're not the same thing. They're not the same thing, and I worry a lot that people who think about traffic as their whole job are ignoring the fact that traffic and revenue are different. If you're choosing channel investments based on referral data, I think you'll become Google's fool, and this is why. This is a pie chart from that same study we did earlier this year showing all the websites that people visit and, put into categories, so search, social, productivity tools, news websites, media, right, e commerce sites, like like Amazon and Etsy and Walmart. And almost everything that happens in these categories where you consume something on another site leads to one of these. And who gets credit in the analytics? If you're listening to this webinar right now and you're like, oh, SparkToro has all this cool data. I wonder if they can tell me, you know, which AI tools are popular with my audience or whatever it is. You might go to SparkToro, but but what's gonna happen is we if we were looking exclusively at referral traffic, we would just see, oh, Google sent us another visitor. How nice. Google must be the channel we should invest in. Dumb. Dumb. Dumb. That's a terrible takeaway. Right? What what's really happening is, you know, the webinar was influential or or a LinkedIn post was influential or something else happened. Right? Maybe somebody else talked about us, and that caused someone to go to Google. And so I get I get real worried that what's happening is the traffic data is biasing us to invest in the wrong places. If you're gonna try and determine prioritization, I would not look at where your traffic comes from according to analytics. I would look at where your audience pays attention. Right? Like, if I wanna reach graphic designers, I wanna be present in the places where they are being influenced, not just looking at my existing referral traffic and saying, Oh, that must be where I should put dollars and time and energy. I made this graphic, thanks to Dana DeTomaso, who's a genius marketer and also a genius data visualizer. She she was out to dinner with us, a couple weeks ago in San Diego, and she was like, Rand, try this. Take a pie chart of, sort of, where influence happens and then layer over that where you spend budget and energy trying to influence your audience. And so this is where our audience is influenced. You can see, like, oh, a 110,000 people, are influenced through YouTube channels in in this audience, and a 100,000 through email newsletters, and a 140,000 through social feeds, and whatever. And then when you add that, you see the mismatch right away. You're like, oh. Oh, shit. I'm putting all my energy into the into the wrong place. Oh my god. I'm so over investing in Google search, and, god, I'm totally slacking on YouTube, and, oh, what, you know, what am I doing on, Discover Google Discover and Apple News? Like, that if I think if you can visualize this to your boss, team, client, you know, c suite, CFO, they'll they'll grok it, right, once once you explain. I do recommend showing 1 and then the other. It it kinda works quite well. You You can find this. I I wrote about this, and I have a blog post, about it on SparkToro from last week. So if you can get your content, ads, brand into the sites where your audience pays attention, I am suggesting that might be more valuable than trying to drive traffic on your own. And what you have to then do is look at the lift in terms of branded search, direct visits because many most of these places don't directly send traffic. If someone consumes content in these places, that will not lead to a direct referral the way it does with Google. SimilarWeb has this nice, tool that can help you out here. So this is essentially showing you competitor they call it competitors, but what it really means is people who visit x, people who visit abduzito.com also visit from up north and Behance and DesignInstruction. Right? And, I think this is very nice. They'll let you run 2 or 3 free searches like this, every day or every week, and then they they sort of flag your IP and you have to you have to sign in and buy. John Shahada from Newsdash, and gddash released a tool just this is, 2 weeks ago he released this tool. It is called DiscoverPulse. You can you can find it. He linked to it on LinkedIn. You can get the link here. DiscoverPulse will show you places where people are essentially getting in their news feed, their their Google Discover feed, sources that are, presenting information on that topic to that audience. I think this is awesome. Like, this is something obviously, we care a lot about this as SparkToro. It's something we try to do. I love that John is doing this for Google Discover because I think that's an absolutely great channel. Alright. Number 3 takeaway. From 97 to 2017 for 20 years, search Google Search was an excellent place to start doing your marketing. Absolutely excellent. It built my career. You know, I know tens of thousands of people who SEO helped and built. Today, not so much. Search is really a reward for doing marketing right everywhere else, and I would argue that it is probably a better bet for many many marketers to build your brand on other people's platforms, which is a recommendation I used to I used to fight against all the time. Like, I think that Amanda and I have built our platform over the last 6 or 7 years on YouTube, on Twitter until until the new owner took over, now on Blue Sky and threads, definitely on LinkedIn. I'm starting to see a ton of traction on Reddit. These places have our audience, and bringing them to our site from those places by earning attention in those places is far more effective than trying to drive direct traffic. Number 4, brand wins. Right? So I when I look at this, what I see is if you can be the brand that your audience searches for rather than trying to compete on the thousands of, you know, lower volume terms that make up a an information space, that's that's where opportunity comes from. Okay. I think we've got what do we have? Like, 10 minutes for q and a, and I believe we've got some, questions already loaded up. So, Amanda, will you give me a sec? I need some water, and then I'll be Yeah. BRB. Alright. I'm gonna leave that. Alright. We have a couple questions here. Just wanted to add about the whole rented land thing. I mean, for my personal brand, I've approached it like building exclusively really on rented land, but then I have my personal site, amandannett.com, which is really just to capture that navigational search where that's kind of my reward. If anybody cares, they can search for for me or my website. Alright. So some questions here. And friends, by the way, if you have other questions, you can put them in the chat, but it's easier if you put them in that q and a section, and if you also upvote some of those existing questions, if you, you know, want those to get answered. So in the spirit of that, the one that has an upvote is from Russell. Would you mind elaborating on your statement that small or new brands, Google might not be the place to begin? That for smaller new brands, Google might not be the place to begin. Yeah. Yeah. Absolutely. So let's see. This, Russell, this is based on it's based on really a a collection of data sources. One is the fact that Google is punishingly brutally harsh on new websites and new platforms. So as an example, I'll I'll show you. I think I find this quite, quite frustrating. Best food and drink gift guide. Have you seen this, Amanda? For the first time ever, we are on page 1. I didn't see this. Never happened before. How many years did we run this before we got anywhere? Is it 6, 5, or 6? Or was it 6 years? Yeah. So Yeah. Essentially, even though we were the only people who had a best damn food and drink gift guide, if you searched for exactly that for the last 5 years until this year when Amanda published this one and and, like, finally broke through, we were nowhere. Like, we couldn't we couldn't rank for it. I'll show you another frustrating one. Zero click zero click searches in Google. This is another example of where this is a phrase that that not only we we coined, but we we were the only ones who presented the research for years years. Every single one of these refers to a SparkToro study, but we are not in the top because these big brands dominate. You you can see this for for all sorts of of queries. I think we're finally well, I say we're finally. I'm logged into my account, so, really, we should use, like, Brave or something and do a new how do you do those private window with Tor? So this is this is, like, how I see what unbiased for those of you who need to get, like, an unbiased view, I recommend a privacy centric browser like Brave, and then you can use an an IP spoofer, well, IP masker like Tor, which takes a while, and then you can see what unbiased Google results would Google doesn't like me hitting them. Have you have you guys seen this before? Where they're like, oh, come on. You you're sketch. We don't trust you. Like, alright. Yeah. Yeah. Yeah. Yeah. Fine. I don't care. But, anyway, that that's a that's a way to do that. And and, essentially, what I'm the point I'm trying to make here is these types of, searches where people are obviously looking for SparkToro stuff, you know, things that we have coined, things that only we have created for years, we could not rank in Google for them despite the fact that, you know, I'm a relatively well, was a relatively talented SEO and and well known person for this. You know, we didn't do anything wrong. It's just that Google has a bias against new sites and a bias toward big established brands, and big established brands are pretty smart nowadays. So when SEMrush and, search engine land and, you know, media post see data that's published on our site, they're like, oh, shit. We should write about that. We can rank for that. We can outrank the original source. That happens all the time. You see it. I I people complain about it with, like, Forbes. Right? Mhmm. Alright. The other question I saw was By the way, I agree with Trona. Local search still has lots of potential. That's where you don't see big brands dominating local. On local. Oh, that makes sense. In the spirit of that, actually, the next upvoted question is from Craig. What are your insights on local search here? Actually, I think this study, because of how it works, because this what we're looking at here, is from mobile web and desktop and excludes searches that are on Google Maps specifically and excludes the Google Maps app, I don't take away a lot of local insights from this. I, in fact, would say this applies to web search broad web search and brands and ecommerce and all that kind of stuff, and I would not take much from this study at all when it comes to local. I tried to do a classification of local searches. I believe the LLMs, they're not great at it. Right? Because if you search for Navy Federal Credit Union on the web, is that a local search? If you're near one and you're a member, I think it is a local search. Yeah. But but the LLM is not classifying it that way because as a broad web search, it's not. Right? It's like I'm investigating. It's a commercial search. So local, I think, was significantly undercounted in here, and because there's no Google Maps data, I just I just wouldn't apply this to to, local. Got it. That makes sense. Okay. This question interesting question from Carolyn. So are you essentially saying it's not possible to market online without social media going forward? No. Not possible. Not wise in most cases, I would say. I think it is I think it is extraordinarily difficult. Right? When we're talking about you know, if you look at the the latest Pew data about, you know, new social media use 2024, which is which is something I pay a lot of attention to. So this is essentially, like, what percent of Americans are using which platforms, and you can see in there, Again, oh my god. You know, if you don't use any social media, no YouTube, no Facebook, no Insta, no Pinterest, you're really you're really excluding a lot of your, potential opportunity. Mhmm. That's not to say there's there's no way to do it. Just, I don't know. It would be like saying, hey. I'm gonna do SEO only on Bing. No Google. Like, it'd be a little weird. Alright. Lance asks, what skill sets should SEO learn to future proof themselves? Okay. I don't know a ton about what the SEO world and the skill set looks like the last 5, 6 years, but my understanding of what it was in the 17 years prior to that is extremely transferable to almost all the recommendations that I'm that I'm having for folks in here and that that we generally have, which is, like, you know, the whole practice of link building for 20 years was, how do I go get these websites to link to my site? And that skill is super transferable to how do I get this website to talk about me, promote me in their newsletter, feature me on their site, get a guest contribution. You don't have to care about the link. I don't care about links anymore, and SparkToro is, you know, built into north of a $1,000,000 a year SaaS brand by essentially through essentially being talked about in the places where our audience pays attention. Mhmm. I I think that is the future of a lot of digital marketing is being present in the places where your audience pays attention, and and the link building skills that you have in that's in SEO are eminently transferable. The same thing, I think, is true for a lot of your analysis and and, sort of analytics skills. So if you see, for example, like, oh, gosh. You know, we are seeing a lift in traffic when we do these things in social media and on other other branded sites. Awesome. Fantastic. Build the correlation analysis that shows your boss, your team, your client. Hey. When we post a LinkedIn video that gets more than 10,000 impressions at least once a week, we see this level of lift in terms of our branded searches for the next 6 weeks, and we see those searches turn into this many new sign ups, and we see those new sign ups turn into this many customers. Right? Like, you're great at that. You're already good at it. Just apply it to not just search traffic. Mhmm. Yeah. I might I might suggest, like, you know, as an SEO, you're learning how to optimize for the search engines as a top of funnel channel. I might recommend trying to learn more about how to optimize on other top of funnel channels, like social media. And I don't know. I also I don't wanna at the same time, I don't wanna be so prescriptive as to say, like, you must learn social media algorithms because I don't know. You know, it's up to you. Right? But I might suggest that it's a good idea if you can if you see yourselves as like, I understand Google's algorithm. Well, some other algorithms that are important to understand as a marketer is the social media algorithms. They're all pretty different, but they all have I think in broad strokes, they're quite similar even though, you know, the nitty gritty is all different. So I might suggest that. For I I will tell you right now for the video game that that we're making, right, Almost everything that I think about from a marketing perspective is how do I get here? This is the Steam homepage for people who are are not familiar. The Steam homepage, which is customized to each person's account, and there are 100 of millions of people who are members of Steam, and there are tens of millions who according to Datos' data are sitting there refreshing their favorite categories every day waiting for a game to come out in the genre and with the tags that they like. Steam determines your visibility. And just like, you know, if you're a YouTuber, YouTube determines your visibility or TikTok determines your visibility or Reddit or LinkedIn. And and I would say your job, right, is to find what these are, find out what these channels that influence your audience are more more so than where you're spending time and energy now, and then go, oh, man. You know, this is a big portion of what we're doing. I need to get great at that. Google's not the only thing usually that you need to be good at. Okay. I think we let's do maybe 2 more questions. Alright. Joel asks, how do we improve brand search? How do we get to brand search improvement? Yeah. So almost everything that you do that puts your brand in front of people who are likely to, resonate with it in some way. So they they like what you're doing. They like your messaging. They like the product that they see. They are interested in the service, and they've heard about you in multiple places. That is what leads to brand lift. For example, one of the I think one of the most frustrating things that a lot of PRs find is that they will get, placement in a big publication. They'll be like, hey. You know, one of one of these industry websites and blogs that's really important to your audience, We got you an article in there. It talks about how great you guys are, and you'll see barely any brand lift. And you'll be like, oh, that's so frustrating. Why is that? And then you do some conferences and events, and then you're also present on LinkedIn. And then some people are talking about you on Reddit, and you get featured on some YouTube channels, and you get podcasts. And then the brand lift starts because one mention tends not to resonate. It happens. It's one of those things where you have to get multiple brand touches. And this is something marketers knew for, you know, a 150 years. Right? If you looked at 20th century marketing, it was all about how do we get 7 brand touches in front of a potential customer through billboards, through radio, through television, through, sponsorships, through events and conferences. Right? That was that was, like, all of what 20th century advertising and and marketing was all about. Those 7 brand touch 7 is the number that I think Ogilvy and Mather came up with that was, like, so important. And the same thing is true for the web. I don't know if it's the number 7. There's a number. For your audience, for your sector, there's a number. You get in front of those people in the places where they pay attention in ways that resonate multiple times, you you will win. It could all be on the same channel. It could be just LinkedIn LinkedIn LinkedIn. It could be Reddit Reddit Reddit Reddit Reddit. It could be steam steam steam steam. If you're in the right place where people pay attention multiple times, you will you will start to become familiar, and they'll go check you out. Cool. A last question here that I think will be interesting to end on from Peter. How different do you suspect the slope and fatness of a long tail to be for b to b? The hypothesis is that there are way more unique searches in each niche category, especially at markets not dominated by 2 to 3 players. I I think that is an excellent theory and possible to confirm. One of the things I want to do next year is work with DATOS to do kind of a, like, a custom analysis of a 1b2b and 1b2c field and show the distribution. SOX is one of the ones I was thinking about in b2c. Right? So, essentially, like, looking at what's the demand for all the brands in the world of socks and people who buy socks online, and what's the demand for all the generic terms, and where do people go before they search for branded terms. Right? Like, how do people hear about darn tough socks? Is it is it all word-of-mouth? Is there something going on online? Is it YouTubers? Is it TikTok? Is it Reddit? Is it something else? You know, the the r buy it for life subreddit is at the top of a lot of the sock searches, and I sort of wonder whether that's a very, influential channel for where people learn about that brand. And so I don't have an answer to that, Peter, but that is an answerable question. And and, actually, I I will say this, which is if you or your brand has, you know, some money to spend in 2025 or some extra year end budget in 2024 and you wanna confirm this for exactly your sector, you can talk to the Datos folks. I'm happy to make an intro if you wanna email me, Randesparktora, or you can go to their website, which is, datos.live. And they can do this. Right? They can do this, like, specific analysis for one field for, like, all the people who visit your site and 3 of your competitors. What else do they visit on the web and where do they go? All that kind of stuff. I think you're you're probably right that in if you take a sector where no one's dominant, where there's no big brand, the long tail is gonna look like much more of the demand than it does here. Cool. Thank you for that, Rand. Thank you, friends, for joining us today. We had some really great questions and considerations in the chat as always. And as a reminder, I will be sending the follow-up email later today or tomorrow, which has a recording to this, the slides, and any and all relevant links. Thank you all for joining. We'll see you probably in the new year. Woo. Thanks, Amanda. Thanks, everybody. Thanks, Duran. Alright. Take care.