Name: Make Better Personas. Do Better Marketing.
Uploaded: 2026-09-29T19:16:05.022Z
Duration: 1 h 4 min 24 s
Description: Make Better Personas. Do Better Marketing.

Transcript for "Make Better Personas. Do Better Marketing.": Welcome, friends. We're back. Welcome. Welcome. Oh, it's been a it's been a long time since we, we had a SparkToro office hours. It has been. It's great to see everybody. It's okay, Joseph. I do wanna know what kind of apples they were. I also wanna know, is it possible for attendees to broadcast their audio? No. Okay. Yeah. I mean, don't think that's a risk. yeah. I mean, in general, in gold in Goldcast, I believe you can. However, we have that disabled. So don't worry, Joseph. Chomp all you like. Go to town. Go to town. I hope everyone is excited and ready for today's chat about personas. I have put together a somewhat short deck. I don't wanna spend a ton of time, sort of being prescriptive and and forcing everyone to do what I like to do with personas and instead show you lots of this tie. This tie is not a tie. This is actually, I believe it's a. It's a, sort of a scarf turned into a neckwear, which which I have been loving lately. And so I just feel like more more things. I will tell you something sad, Amanda, though. I got some of my custom made clothing from Paris Yes. and, terrible news. Apparently I can't tell. They may have mistailored it, but, Oh, no. yeah, the the pants fit great. Wonderful. The jackets are too tight, and I'm wondering if I did I do too many pull ups? Is this a is, this a problem other gentlemen have experienced? Have you. too been hurt by big pull up? You got too swole, bro. Your your lateral muscles have now expanded and your, tailoring no longer fits in a way that you find compelling. This is, I believe they call these champagne champagna problems. Yeah. I. think so. Oh, no. Your steak is too juicy. Your lobster's too buttery. Ugh. It's great. I love it. Alright. I am going to present. Amanda, we always get some questions at the start of our webinars. We can't, remember. what they are. The questions we are often asked are, will we get the recording? Yes. Of course, friends. You will get it on this very link on this little website here very shortly after this call ends. And, also, the presentation. Yes. Don't worry. You will get a PDF of this presentation. And then the final question that we often get is, is that adventure time behind you, Rand? Most people ask that. Yes. Yes. It is. Yeah. Indeed. Alright. Let's see. Hopefully alright. I believe you can now see my screen. Looks beautiful. Wonderful. Wonderful. With with SparkToro's new branding. Hide this screen sharing. Great. And, today, what we wanna help you do is make better personas no matter what you use personas for, and there are a wide variety of things that people do with personas. In fact, Amanda ran a survey earlier this year. I'll show you some of the results of that survey, actually. But to get right into it, I wanna start with talking about template personas. These are these are the ones that at least I learned in in marketing, I don't know, 02/2001 or whatever it was at the University of Washington, sort of classic marketing personas that you were probably taught on the job or had to take over from a predecessor. And a lot of these traditional personas look something like this. You can find thousands of them on the Internet. I find these I struggle with these a lot. I think if you get value from them, that's great, but I am concerned that because they are primarily, instinct and intuition driven rather than data driven, it's it's a lot of vibes and guesses and fan fiction. In fact, I think, Amanda, that a ton of the whole personas process could literally be called customer fan fiction. It yeah. That's basically is. yeah. It's sort of like, I imagine this customer who does these things, who has this name, who balances schoolwork with a part time job. I think I think in the, in the best case scenario, right, if I imagine, like, a a truly wonderful marketer who's who's writing something like that under challenges for their persona, the reason they're doing it is because they want other people on their team customers as real people. And so by giving them these specific qualities, even though that says nothing about what marketing you should do or product decisions you should do or financial decisions you should do inside the business, It at least builds empathy for, hey, this is a real person outside our organization and not just, you know, faceless, nameless, make the sale. There are a lot of products on the web that will let you fill in the blanks on these. I have not found one that I love. Otherwise, I would have pointed it you to it. That is mostly because I think that the key areas that marketers should care about for marketing personas are almost never present. Sometimes some of them are. I think brands and influencers and preferred channels are actually pretty good. I think motivation, fear, growth, power, social, I think bio, I think personality, many of the demographics, not actually very useful. And I think this is this is missing a whole lot of detail around what we should do. Yeah. The most popular I always, sorry. feel like oh, sorry. I just wanted to say, like, the motivation piece here, the incentive, fear, growth, I've I I've always thought that's very superficially interesting, and that I've always thought I feel like, theoretically, I'd want to know that. But once you have that in your actual persona template, it doesn't actually change what you do. And, like, I mean, what does it mean if they rank super high on fear? Like, our customers are always scared. Like, what does that even mean? I I really I think that there is, very little that a marketing person can do sometimes in product strategy or even, like, company design. You might say, hey, we're going into a market. I don't know. For example, we're making antibiotics for kids. And so as a pharma producer, you're like, gosh, we have a huge hurdle to cross with customers and potential customers who are fearful of harm or secondary effects from our product. Okay. I can imagine that knowing that might be somewhat useful. But, yeah, for marketing purposes long after the company's been created, I struggle with it. Yeah. Thing that frustrates me about all kinds of data, which is if by collecting and knowing this data, you would not do anything differently. You should not bother doing it. You should not bother collecting it or storing it or reporting on it. It bothered me a ton in that thread the other day on LinkedIn where I was like, hey, these post purchase surveys, like, we all know they don't work. Here's all the research about it. And people were like, I still like it as a data point. But if you're not gonna do anything different, I don't know. Some maybe people dislike things. People like their fan fiction. People like their fan fiction. Yeah. Wait. Last little joke I wanna call I do wanna call out a funny joke I saw here in the chat from from our friend, Spencer. He says, persona customers have such perfect lives, but they have solved everything except that one thing. except your one thing. The AI this AI persona, generator from HubSpot, they still offer, by the way, the legacy app. You can see the link down there. This is currently the most popular, AI tool on the web, sort of the number one thing that will come up in Google or or, the thing that gets the most visits. Obviously, HubSpot has had a persona generator for a long time, so no surprise. I I thought we should give it a try together. So I filled it out for Alert Mouse, right, which is my, one of my other companies that that makes, like, a Google alerts competitor. And so I try to describe my ideal customer. It asks a few questions. It has, like, three or four different prompts in here, and then it gives you this. So here is my persona, Weber Wiggins, who is 25 to 34 years of age, a web creator with a bachelor's degree who uses Facebook, Instagram, Twitter, LinkedIn, and Pinterest, and is in the technology industry. I I found this even less impressive than a lot of the template ones. No offense meant to HubSpot. I think they're a great organization. I think a lot of things that they build are great. I suspect this was vibe coded by someone in the marketing org and could use some serious polish. So despite being very popular, I I would I would sort of suggest against this. This doesn't it doesn't tell me key information, and it is all hallucinated data. Like, literally hallucinated, meaning that there it's not based on anything. There's no, you know, corpus of data that from which it is drawing, unfortunately. And I I like HubSpot's products a lot, but I think this free tool could use some help. You can click on it, by the way, if you wanna if you wanna check out the live version and scroll down and see all the stuff that's in there. You'll get the vibe that I do, which which is essentially that this is just stereotyping. And this is my biggest fear about personas in general is that they become even more than fan fiction, they become stereotypes, and those stereotypes can bias you to make really bad marketing decisions, as opposed to knowing knowing what else is out there. Of course, my my favorite you know, you are all are familiar with the the current SEO, AI, EIO trend of, hey. Let's make a listicle. This one ranks quite well in Google. Here's a a listicle of all the AI persona tools out there. Of course, the number one tool is Joomla as ranked by Joomla on the website, jooma.com. It's a great new technique, works really well. In my opinion, all of this, the whole process of filling out a template or using the AI generator or or following, you know, your company's history of what you've done with personas is the cart before the horse. And that is because does anyone you remember the scene from The Incredibles, Geraldine and I quoted to each other all the time? With the following line being, why do you need to know? Why do you need to know? I think this is a fantastic question from Frozone's partner, girlfriend, wife, not not clear. And she is asking a question we we should all be asking ourselves, which is anything you're putting in a persona, what are you gonna do with that? Once you have the answer to this bit of persona information, what will it help you accomplish? The reason I encourage you to do that is because before you create the persona, if you know that you only need these five pieces of data, you can dramatically reduce the lift in the persona process, and you can focus on getting the highest quality and best data you can for the items that you really need. There are some bad answers to this question. It's part of our process. That's a terrible answer. That is just that is just we do it because we used to do it. They're part of annual reports. That's also we do it because we used to do it. Doesn't doesn't tell me. We use them as one of several inputs to determine our strategy. I want more specificity. Maybe you do, but you gotta you gotta tell me specifically. If you're using, for example, gender data in demographics to make your ads more appealing to one gender or the other, I I wanna know that. And I would also caution you unless you have, like, ninety ten split, that can be a dangerous proposition. They help the team stay connected to the customer. I think this is what I was talking about at the at the very start of the presentation, and, I understand it, but I also think it's a it's a poor answer because it does not get to what are we gonna make actionable out about this data. Good answers. We stress test creative against this. Right? Like, we look at the at the matches for demographics and psychographics and behavioral data, and then we stress test our ads. And we're like, wait. Wait. Wait. Wait. Wait. This this does not match, with who our target customer is. This is not going to work well. We should design this or reconceptualize it before we start testing and launching. Great use of personas. We use them to prioritize new tactics and channels of the mix. Excellent use. Wonderful marketing use case. Absolutely, you should do this. If you are considering which channels tactics to invest in, absolutely, you should figure out your it doesn't have to be persona data. Right? You could just do raw audience research for this, but very, very helpful. And finally, something like we compare similarities and differences between our different audiences, our different personas to know where we need unique approaches and where we can do overlap. That's fine too. Right? So if you're talking about content marketing, you're talking about social media marketing, or you're talking about advertising or video creation, great. This all makes sense to me. Even if you're just talking about, hey, we're going to make different landing pages for different customer sets, all very reasonable. I also wanna mention this is not just you. When, Amanda, when you ran this survey earlier this year, for SparkToro, we got a 150 respondents saying that personas are really pretty challenging. Oh, I apologize. The screenshot didn't come out quite as clear as I hoped. But you can still see I I actually I kind of love this. We I broke this down or Amanda broke this down by free customers. So so free users of SparkToro and folks who are paying customers. I love that the top answer for everyone is the same top answer that I have, which is that there's no reliable or credible data. All those processes that I showed you at the start of this presentation for personas, in my opinion, are not reliable or credible data. If someone said prove it to me, you don't have proof and that I find that very, very frustrating. There's no shared definition of good. Their personas are built but never used in decisions. Yep. Why do we do it? Because we used to do it. It's too time consuming. Yep. Hard to keep updated. Stakeholders don't see value. The last one, great. If the stakeholders don't see value, I think hopefully, hopefully, this process that we're gonna talk about today will be more useful. And then we asked folks, in this in the same survey we asked for, what are the things that are most important to you in a persona? Pain points and challenges, number one, keywords and questions they search for, objections and barriers to conversion, decision criteria. Some of these, they fall into two different categories. They fall into the category of you can get these really well by talking to people, through surveys or interviews, or you can get them you need to get them through passively collected, behavioral data. Right? And, I'll I'll I'll show the breakdown here. You can get some of this data for free. I for those of you who are budget strapped, but have the time and bandwidth to do this, I wanna show you how you can get many of these items without spending a dime, without any use any product, any paid product. So customer interviews, that can get you pain points and challenges, decision criteria. Basically, all of these sorts of things where people are pretty good at describing, self describing, their needs, their wants, their behaviors, goals, and aspirations, sure, skills and previous experience. This is that's a rare area where people are quite good at saying what they have and don't have, unlike a lot of, you know, like, with, like, the the the post purchase survey thing. Right? Asking people where they have been on the Internet prior to making a purchase, you just you cannot expect good results from that, anything reliable. But asking people, where do you work? What do you do? What skills do you have are required in your job? Yes. Yes. People will give it to you. Will some people embellish? Absolutely. I'm not saying every answer to every one of these is gonna be perfect. I don't but you don't need perfect data. If you get literally 15 or 20 answers to these questions, you can create a median and average that'll give you enough to go on. That that is that is the point of this. So it's imperfect data, but that doesn't that doesn't matter. The rest of these all need passively collected data. So if you need to know keywords people search for, you can't ask if you interview 10 people who've, recently purchased running shoes. They they will not be able to remember what their Google search history contains. Maybe they can pull it up for you, but, content consumption patterns or preferences, websites they visit, user journeys, press and media platforms, accounts they fall. I'm thrilled that people want to know these for personas. I think this is a step in the right direction. You know, it's embarrassing though. Those templates, the AI builders, they contain almost none of these. Platform preferences is one of the only ones I found on those, that was consistently there. The rest of these are not even present in those templates, and so I I suspect a lot of people are missing out on them. Alright. How do we manually get this stuff? So Google keyword data, I didn't take screenshots of everything and I didn't throw in all the logos, but you're familiar with the usual culprits, SEMrush, aahrefs, Moz, the Google the Google keyword tool, and 500 other, SEO and keyword research tools are out there in the market. You can use your favorite one. I have not done a deep dive. Amanda, I don't know if you have in any time in the last couple of years into, like, what's the best keyword research tool? I suspect some people in the comments. If you have a favorite keyword research tool, maybe leave your, fellow attendees a favorite keyword research tool and why you love it. That would be super helpful. Almost all keyword research tools are pulling from one of two data sources when they show you this volume and trend data. Either they are quoting exactly what you see here, which is a Google Keyword Planner, which is not perfect but can be directionally, useful, or they're getting it from click stream data that is then modeled against sort of the full, you know, web usage. Similar web, data switch powers, Semrush. I think who else? Someone else has a, a big clickstream panel that they buy from I can't remember. I think it's one of the one of the, big European platforms, and they also put that keyword data in there. AI prompt topics, there are two ways to get it. One is the expensive way. Clickstream data. You can you can custom buy from someone like Daedos or or SimilarWeb, or the other, comScore, Nielsen, the other guys, or the cheap way. The cheap way is how almost every single tool in fact, I can't I don't know. Someone Will Reynolds told me that, peak a I, p e I think it's p e e c dot a I, that they model using Clickstream panel data, the quantity of prompts that are about a particular topic. But my understanding is that almost every other AI tool gets it from doing something like this. They send an API request with a prompt to the AI tool and they're like, hey, this many people search Google for this model for me what prompt topics they're going to use in the AI tool and then they give volume data. I think this is low credibility. There's no data source that I'm aware of that gives you a real sense of, like, if lots of people search Google for trail running shoes and a few people search for, let's say, let's say, a hundredth, one one hundredth of the number of people who search for children issues, search Google for, Google alerts alternatives. Is that same ratio held up in AI tools for prompt topics? I don't think so. I'd be pretty sure that it's not, but I have no data to suggest that and and nor am I aware of a credible source that would provide that for you. Also, there's a lot of problems with clickstream data inside AI tools. They can only see ever the first prompt sent. They can't see inside API requests. They can't see inside the desktop or the mobile app versions of these, only the desktop web version. You're losing a lot. Websites. There's expensive ones. The expensive way to go is similar web. I believe I think their minimum is somewhere in the $500 a month per range right now, but they've been testing a whole bunch of pricing. So I'm not exactly sure. However, the data is very good. When, Amanda, when you and I did our our big study of, you know, Google searches and and zero click search, we use SimilarWeb. They they have outstanding data. They have the most coverage of, in particular, the mobile web. The cheap version is you go to Google and you surf. Right? You run all the keywords. You could you could set up an AI process to do this, or an API process if you wanna use somebody like SERP or SERP API or data for SEO, and you crawl all of these and you find all the websites that come up and how how much they come up in Google. This will not give you a great sense of websites that don't rank well but are still trusted on a topic, but it will give you, a pretty good start. User journeys, there's there's no cheap option for this. By the way, I think these are fascinating. In my opinion, user journeys are the flipping coolest things to see. This is a, this is an amalgam of a few people who, Eli Goodman, the the CEO of Deidos, who's a friend of ours, basically did this analysis looking at someone who bought in several people who bought a TV from Samsung and then bought a wall mount and and what that purchase process looked like. And you can see just how complex and wild it is. Will Reynolds and I geek out about, you know, a clickstream path to purchase all the time. I we just think it's the coolest thing. It really I don't think there's any better way to form empathy for how people actually buy than to look at it and be like, oh, oh, this purchase came from Reddit and then went to Google and then went to perplexity but wasn't happy with the answer there and then went to ChatGPT, but then directly typed into the browser the URL for the Samsung website because Chatibiti didn't provide a link, and then they added it to the cart. But then they left the cart, and then they went to Target, and then they checked alternatives, and then they went back and purchased it at Samsung. Woah. Like, just what the heck? Crazy. Press and media Press and media. Thankfully, this this is cheap, but not a very good way to do it. So I you can search for the keywords and then click on the news tab inside Google and get essentially press and publications that do coverage of this sector space. I don't love it. It's it's hard to figure out. There is also another process. John Shahada runs a company that helps you. I believe it's called, oh, shoot. What's John Shahata's company? It's it's a Google Discover I'm blanking, product. guys. But it, like, news something. yeah. So I'm just blanking on the name. But his product will let you basically say, hey. Show me publications that are popular inside of Google Discover, Oh, it's g g dash or news dash? g dash. Yes. Great. G dash. G dash, which will show you sites that are popular platforms that are popular inside of Google Discover like this. But you can run that similar process that you would for keyword research. Platforms? I don't think there's a cheap option. Again, you have to go with something like SimilarWeb's probably your cheapest, non SparkToro option for this. So if you want if you wanna get platforms, you don't wanna use SparkToro, I would recommend SimilarWeb. I'm not aware of another public source. You the expensive version, of course, is buying with clickstream data yourself. And then accounts they follow accounts they follow, you know, you can do this manually. So I went and I looked at Samantha Rose who's the CEO at endless commerce and and basically looked at all the sources that the top voices that she follows on LinkedIn. You can do the same. Here I did it for Amanda on Instagram. I'm very proud to be on that list. That's really sweet. I suspect Instagram automatically puts me at the top because because it's me looking, so I just wanna see myself in there. Alright. Is there is there an easy button for this? Well, kind of. The good news, friends, is that, Casey basically took all of that entire list of everything that people wanted and built every one of them. I'm I'm warning you right now. I feel deeply uncomfortable at this. You can. Anyone who's known me for a while knows that I hate saying nice things about products at companies I work on. But I'm gonna try to swallow my discomfort and, and tell you about this. So this, is the SparkToro persona generator. I'll walk you through exactly how to do it. In fact, we can run some test use cases together. That's what I hope to reserve the majority of our time here for. But I wanna point out two things here. One is some data comes directly from click stream or collected social data, crawled social, or crawled keyword data, crawled website, crawled YouTube data. And some is that data source plus an AI interpretation layer, and we, unlike every other tool on the Internet, which pisses me off to know, we will tell you when something is AI interpreted. So, this means that these pain points and goals, we essentially feed the relevant behaviors and, demographics into an AI model. We use a prompt. In fact, Christy and I worked on these prompts together until we got something we were quite satisfied with. And then this is the output of that prompting plus data process. In my opinion, that is far more credible than, prompting alone. But is it as good as potentially perfect as in as a customer interview? Probably not. So I just want you to as you're going through here and you see this AI interpreted data, know that this is going to be stereotyping. It's gonna be reducing some rough edges. You would definitely get even more value if you added customer, interviews into this process. But it's about as good as you can get when you add sort of technology available to us and data that you could potentially get. Let's go do this live. I have got my SparkToro workspace right here. Is that showing? Doesn't appear to be showing. Not yet. It's still your presentation. What if I close the presentation? Nope. Doesn't care. It likes the presentation better. Alright. I'm gonna stop sharing and reshare. Screen. Screen. I'm, gonna go window. in our chat asks, apologies, Fred, if I'm saying your name wrong, Joao. He asks, for the AI interpreted stuff, Mhmm. can we see the rationale that AI made to write that? I bet behind the scenes, we could conceivably expose it. You know what? It depends on what model we're making the API call to, and it depends on what you believe about rationale. Here's what I mean by that. When AI rationalizes things, our my understanding is that very often that is post hoc rationalization, not actual. right? So it it's essentially saying, like, I created this output. Now someone is questioning me about it. I must spicy auto complete put words and phrases that credibly sound like they would rationalize a reason for why I did why I produced this. But it is not actually telling you how it got to that result. So just not take with a grain of salt, but maybe take with a tablespoon. You know? Like, it's a, it's a little bit nuanced. But, yes, I can. ask Casey about that. I think I don't know if Casey's on the webinar with us today, but I can definitely ask about potentially exposing, rationale depending on the, model that we call. That should be possible. So, yeah, we can we can look at whole bunch of these. I will let's see. Why don't we look at knitters who are seeking yarn? So this is a I I ran a report on this group inside of SparkToro. From those reports, you can say generate a persona. Those, purse that persona generation process produces a document like this, and you can see all of the items. I'm gonna blow this up just a little bit so that it's, a little easier to see on everyone's screen. So you can see we it's basically in order of the things people told us they used and wanted, which, to be fair, I I I actually almost exactly agree with that the that these are the most important things. And then you can see the, you know, SparkToro data that's in these AI interpretation layer, SparkToro data around media habits, around brands and influencers, press and media, all that kind of stuff. This is obviously exportable. I encourage you to play with it. And platform behavior is also in here. Although, I will say the platform behavior, I think, is stronger inside the SparkToro report than it is in here. Like, you can you can get a couple of, like, nuggets. What the AI is doing when it produces this report is we basically say, show the most interesting, things. And so in this case, it's it's just this heavy Pinterest usage. But, for example, if we look at interior designers actually, that might be Pinterest too. It wouldn't surprise me. ChatGPT, Pinterest, Instagram, Pinterest. Yeah. So, there you go. This this data, if you have, basically, the the questions that you're trying to align, I find this quite useful. I've already used it for alert mouse, and our marketing team at SnackBar Studio is using it. I've helped lots of people apply this data. I found it quite good, which is why I felt comfortable showing it off. But I think it would be awesome if we maybe did some live runs of this stuff, and I can show you exactly how the process works and what it produces. So if you've got an audience that you wanna see, please let me know in the comments. And in a in a world's first, Amanda, because I'm under the weather, I'm gonna have to briefly bio break, and I will return in just a. second. So, do you want me to turn off my screensaver so you could pull up yours? Sure. Okay. Alright, Cool. friend. Okay, I'll let. you I'll let you pull up SparkTor. And in the meantime, please give me your audiences that you want me to analyze. Hang on just Yeah. Yeah. So I'm gonna pull up my SparkToro first. I love this. You know what? I haven't shared my screen in a long time. How do I do that? Alright. Okay. My screen is being shared. Okay, friends. Lots of comments in the chat. Thank you. Okay. I am going to start with I'm just gonna go first. HR professionals fed up with our current payroll provider. Okay. Just gonna copy and paste that. Do do do. Alright. I know there's a question here from Lauren Keller. How often is the data updated, especially for media habits and brands and influencers as those things can change pretty quickly? You know what? I haven't thought about this question in a while. I believe it's about every, like, once every couple months. So we kind of say quarterly, but I'm going to put a pin in this question and revisit it if we need to. Thank you for the question. Okay. So we have our there we go. This is our audience of HR folks. What's pretty cool now is that we can we now show you your total addressable market, which we estimate, you know, the estimated audience size, how it compares to the prior year, and their estimated market value. I think this is really cool to see for, you know, when you're doing general market research, or this is also really helpful sometimes in, like, finance decks. I know that might when I worked corporate, my team would have loved this. Okay. Let me see. There's some product makers with revenues. Oh, okay. But I'm gonna try and I'm gonna try and dig through some of these comments and find a great audience too. Which one are you analyzing? Oh, HR professionals? Okay. Yeah. HR professionals seeking better payroll providers. Alright. I lost track. Am I going to, personas now? Yeah. Yeah. Okay. Alright. So let's go to take action. Generate personas. Let's go. Yeah. And if you if you saw there, right, Amanda was able to to check one box, but, conceivably, you can check multiple reports and get persona data on, Angel. multiple groups. So this is also for example, if you're analyzing multiple competitors' audiences at the same time, you're analyzing, you know, four or five different breakdowns of your audience. That's a great way to get different data on different groups, and then you can see the the overlap comparison as well. Yeah. Oh, while this generates, we got a couple questions on our TAM, Yes. estimates. So how comfortable are we? Where are we getting the TAM from? Yep. Tam is coming from, I believe, if I understand correctly, three places now. So it is number one, we use if it's relevant, we use the count of LinkedIn profiles scaled to the population of the country. So, for example, if you say, whatever chemical engineers at companies with more than 500 employees, we will go look in our LinkedIn dataset for exactly that group. We'll make a count, then we'll go, oh, okay. Well, we have whatever it is, you know, a 100,000,000 or 80,000,000 US LinkedIn profiles. We know that there's a 180,000,000, you know, working adults in The US, so we're going to scale that accordingly. The second piece is we make an API request to the AIs to go out and fetch any publicly available, data sources that have market sizing estimates from large scale surveys, from a number of trusted providers. So, like, a, a Gartner or Forrester type of report, and that also is inclusive of US government data, or Canadian or UK Government data in those cases. So those are also, involved in there. And then we will also do scaling based on demographic filters. So if if you say whatever, oh, I only want women chemical engineers, we will attempt to find, like, what percent of LinkedIn profiles with the with who are chemical engineers are are women, and then we will scale up or down accordingly, by the percent. That is I Casey may have put some other fancy math and calls in there, but that should be I believe that's mostly comprehensive. It's not perfect, but Mhmm. I would I would also argue neither is Forrester or Gartner's estimate Right. or or USGS data. So I've I found it relatively relatively good in most sectors, but not perfect in everyone. Right. And in this case too, I mean, any TAM estimate is estimated. And now it's commit. know? Yeah. Like, Yeah. it is interpretations of various sources of data. It's usually just done manually. Yes. And what it. is what I can say is the SparkToro one, Mhmm. Mhmm. almost always will be better for, getting that estimate than saying to raw chat g p t, how many chemical engineers are there at companies over 500 people? Right. Like, when you feed quality data into a model, you're just gonna get so much better stuff. So much so Yeah. Cool. Alright. Okay. You know, I'm gonna download this. yeah. So the I I was gonna say with with this audience, like, one of the this is the HR audience. One of the things that I find really valuable is taking this persona template and then backing it up with the required, the requisite like graphs and charts from the report Mhmm. and then putting that into a presentation. If you are, for example, saying like, hey, these people have these content preferences, this is a nice short bullet point list and then you can take the graph of, for example, you know, social networks that they visit and pop that right on the side. And these two things together can really help make the case if you're pitching a client. Right? You're in an agency and you're pitching a client or you're in house and you're trying to pitch your boss or your team on approving budget or running an experiment, which, you know, I think I think experimentation is the sort of the only option that we have in an attributionless world, for marketing. That that's really where I think this stuff can shine. Okay. I think we have bunch more questions. We do. you have a lot. do you want me to take over and you read q and a, or do you wanna keep controlling and I will take q and a? You take q and a. I'm taking q and a. Right. Help. Yeah. By the way, what I did was this persona you all see on on the screen, I just went ahead and clicked on our where's my screen? Okay. This fancy download PDF button, and I plopped it into Google Drive. And I shared it with you all there in the chat, so you can see for yourself how it looks like when you download these reports as PDFs. Alright. Ronnie asked, are there any plans to include insights about buying committees for b to b products and services? Oh, this is fascinating. So, essentially, trying to build a persona for a group of people. I'm trying to think how one might do that in a data backed way. Ronnie, I think it's a great question. My suspicion is that will be challenging in the short term, but potentially in the long term. If you wanna shoot us an email about it, I can talk to Casey, and we can look into it, a little more. But it's a it's a very fair point. One thing that I could see, doing oh, yeah. Amanda, we can't see your PDF, I don't think. No. I I dropped it in the chat Ah, great. Okay. Excellent. so that people can download it. I'll drop it again too. Perfect. Perfect. But yeah. Ronnie, fascinating question, and I agree that b two b buying committee data or or understanding a group of people making a decision, which can include lots of different individuals is very interesting. Let's see. Someone's saying it's not in the chat. I see it in the chat with two thumbs up at the very bottom. You might. need to scroll to the bottom. Let's see. Can you spare can you parse data by primary language? So Spanish speaking audiences versus in English speaking. Yes. But only in the three countries where we have data. So US, UK, Canada. I would love and we are Casey and I have been working on plans in 2027 to be able to cover other countries. The biggest challenge that we have is that, SparkToro's data is by, purchased by country. In order for example, all the all the, like, amazing insights that you see, all these numbers and data, right, that comes from us basically, buying datasets of of LinkedIn profile data, buying datasets of, click stream data, and then doing a bunch of crawling. We can do the bunch of crawling ourselves, but we have to have that clickstream dataset that is quite expensive. Every new country we go to is many tens of thousands of dollars and so essentially SparkToro has to grow before we can expand. Let's see. Are there plans to include oh, yep. We got that one. Any best practices for creating audience prompts? Fantastic question. Excellent question, Joseph. I realized that that takes us a little bit back, but that is something where, I have I've got a video in the resources help hub about this. My strong recommendation is make the tiniest slice of your audience that you can that is distinct in behavior from the other groups. And and if you can run four, five, six, you know, even two or three rather than one giant one. So if you, for example, are looking for people who work in HR in The US, my god, that's a huge group. Is it the HR folks who work in recruiting? Are they at larger or smaller companies? Do they work at consulting firms and agencies, or are they in house? All of those things. And then, yeah, you can do what Amanda's doing here where you make a persona for each of those different groups and then see the results. So, yes, excellent excellent question. That resource, I can try and pull up for you and put in the, chat. What do you know? And I'm I'm wearing another in that video, so, you know, we got that going for us. Alright. Great question. Two questions. What are the best practices when your audience is very gated? Oh, very gated or lacks volume of insight, I. E. There's not a lot of publicly accessible data about them, and most of it is gated content. Interesting. What think? I might need, yeah, I might need examples of that, Kevin. But okay, for for example well, I'll give you a a spicy example. So, for example, let let let's imagine that you needed to reach OnlyFans creators. This is this is a group with with a lot of market purchasing power, but relatively challenging to get data on because the public profiles that they might have are very disconnected from the sort of profiles that one might be able to easily crawl and access. So quite a gated group, and there's a lot of privacy issues around this and, you know, multiple personas, not personas in the sense that we're talking about here, but, like, people using different online, accounts. Still, I would say because of Clickstream, SparkToro is usually able to get very good data about this group. You know, you say something like, gated content, meaning but, like, for example, Google wouldn't be able to see it or access it. Clickstream can still show the URL. Right? A clickstream path will still show the URL that was visited by the device. So my my suspicion is most gated content stuff, we're still pretty good at, or gated communities. I would say, Kevin, if you find a community where you're like, gosh. It doesn't seem like SparkToro has data about this. Can you help me maybe describe or or get insight? Shoot us an email, support at SparkToro or Rand at SparkToro. It goes to all of us. That works great. Okay. Second, when looking at personas, should you include a how this persona uses AI to find answers section, or should there be a persona specifically for AI assistant models? You don't need one. So, for example, you if you wanted to, you could say when you're describing your audience, you could say HR directors who use AI to find software solutions for their management or, you know, whatever it is. But the only thing that would really do is limit the group that we look at to people who, like, don't, who are very heavy into those AI platforms. So you you'll just see them be very high in usage. If that's of interest to you, you can do it. My suggestion would not be to start there, because you'll exclude a significant group. But according to the DATOS data, like the reports that we put together with them, only 20% of Americans or 30% of Europeans are heavy AI users, meaning they they visit and use those tools 10 times or more a month. So you'd you'd just be cutting off a significant portion of the group. But if you're interested in that group, you can see it. Nathan, yes. It is recorded. This is being recorded, and we will send you the recording afterward. Oh, look at that. Amanda's answering. Rand's answering. We're all there. Okay. Oh, Emily, what a question. If you have your own customer interview documents, can we upload those into the SparkToro? Okay. First off, I would like to officially petition to add the article the in front of SparkToro. It makes me makes it sound way cooler. We can drop the. the. It's cleaner. It's you can you can upload your documents, but but not directly through the UI. The way I would do this is the MCP. So if you haven't used the SparkToro MCP, it's badass. Like, Casey really did a lovely job of integrating. I I use it with Claude. It's like a one button click to add SparkToro to your, to your Claude instance, and then it can basically connect. You can upload those docs into Claude and say, hey. Go connect, like, our interview documents to the persona or the audience research data from SparkToro, and I want you to give me whatever output you want. I think the MCP has been kind of a game changer. I believe I can't remember, but I think, like, something like, some significant percent of SparkToro's users are MCP users primarily now. Cool. Have you built audiences using an LLM and GSC data? Oh, Joe. No. We have not. I am I I don't know. I'm just skeptical of the stereotyping that LLMs do when they build audiences. So, like, what you see here, what you see inside the product, that data, because it comes from real people, I feel a lot more trust around it. It can't give you everything. And and so, you know, like, for example, the the AI prompt topics, right, will say that, hey, we use AI plus these models and and send the data and then get that or the, you know, AI powered label on the personas. So you're you're getting some LLM use inside the product, but the core base of it, we believe always has to be real data. So have not used LMS. GSC data is interesting to us. We've thought about it. We haven't collected it. So it's not something we have now, but we've thought about building a repository of that and anonymizing and aggregating, of course, like like an AA trust does. I really like some possibilities there, but it would have to be so many sites. My god. Just just so many. Okay. Our last question on here, is there a way is there will there ever be a way to input, add qualitative qualitative audience research? I I think yeah. Ronnie, you mentioned, yep, the MCP is the way. So great. Alright. Then I see Sarah asked, can I get super niche with my audience? Oh my god. Sarah, you can get so niche. It's crazy. You can be like, I only want queer friendly coffee shop owners in Wallingford, Seattle, which which is a a neighborhood. There can't be more than four of them. But but, like, because I think we only have, like, six coffee shops in one. You can get very, very detailed. Here's the only thing. Once you start going way, way, way down deep and getting incredibly niche in the audience description, we have an instruction that basically that interpretation layer that turns that that goes and turns into the data fetch can't show any data unless we have at least a 100 profiles that match because of privacy. Right? Essentially, what we don't want you to do is say, like, I wanna find a fullard tie wearing Jewish man in his mid forties who broadcast webinars from his office in Seattle, Washington. Right? And then, like, you get data about just me. That would be creepy. We we don't want that. There there there are all sorts of problems in there. And so there have to be a 100 people who match the description for us to show it, in the in the product. No way. I'm falling for there's, like, 10. It's true. Amazing. Awesome. I don't remember what you said, Rand, about the I think it was queer only coffee shops. I'm gonna try this super niche audience. Anything we wanna, add. to this? Queer friendly queer friendly coffee shops. Queer friendly. I thought it was only. I'm like, I mean, that's probably There there probably are some, but I think, yeah, mhmm. more more friendly than only. And those I bet I bet there's a 100 there's more than a 100 queer friendly coffee shops in Seattle, so I bet we could we could do it. Yeah. Kevin, go run to your heart's content. Okay. Was there one more? Okay. What about Brad asked, what about for lower consideration purchases? CPG, toothpaste, ice cream, fast food? I think this is lovely for b two b. I will say if it's very low consideration, like a lot of impulse purchasing, I think the SparkToro data is interesting but not quite as actionable as it is for, I would say, like, niche consumer interests or moderately considered purchases inside consumer. So for example, you know, dishwasher or television or or neckties or, you know, video games or, like, that type of stuff in consumer, all excellent. Ice cream, maybe. Ice cream, it depends on the type of ice cream. If we're talking about, ice cream in the in the frozen food section at the supermarket, there's there's some use cases maybe if you're, like, luxury high end or very you you you have a certain niche. I think the SparkToro data could be pretty useful for creating some brand demand through, you know, trying to find the media outlets, the sources of influence, the influencers who you want to try your ice cream, that kind of thing. But if you're Briars, meh. This this data, I would say you're you're you probably want more of the, like, market research stuff, and you're probably throwing a ton of money into brand advertising anyway. For b two b, obviously, that, you know, this data is priceless. And if you're if you're an ice cream store, I believe by the way, I don't know if I'm allowed to reveal this. I'm gonna say that an ice cream shop that was very popular that started in Portland was is a customer or was a customer of SparkToro. I don't know if, they've canceled. I haven't looked. Yeah. bud. You're probably thinking of the same one that I am. And they yeah, they were using SparkToro to do some research into their audience and customers and comparing themselves against some other looking at some other, ice cream sort of premium ice cream vendors, like people with with multiple shops. So it can be useful. Alright. Woah. Lots more questions. Do you have to specify the country? Yes. There's a drop down on the audience research, page when you when you first input your audience where you will select United States, UK, or Canada. Those are the only three countries we cover right now. And by I think by default, it'll be set to whichever of those three countries you are in, but you can you can change it. Is there a best practice if you have personas that have variance? Man, Kevin, I really wanna show you a thing that Casey built that we haven't launched yet. Do you think he'd be okay if I if we previewed a feature that what do you think, Amanda? I think he'll be okay with it. I mean, speak now in the comments or forever hold your peace, Casey. I would. love for him to, like, go video on randomly just frowning at jump in and be like, don't show it. you. Alright. So basically, what Casey built is a when you describe your audience, we will go dig into that audience first and try and figure out if there is a are sub audiences of that group with separate behaviors that probably deserve to be analyzed differently. So let me find that. Oh, I like this persona for the queer friendly Seattle coffee shops. I like that it's Casey the conscious coffee explorer because I like that Casey is a gender neutral name. Oh, yeah. There you? go. I like Yeah. Oh, great. Oh, great. that. Shoot. What did he call that that product? Is it he called it a quiz maybe? Oh, yeah. The v three user quiz. Okay. I will show you. Alright. Let me take over the presentation briefly Yeah. if I can. Oh, do I have to hit stop? Okay. I think you have to hit stop. Yeah. Alright. Window, audience quiz. Alright. We're technically at the end of time. Thank you everyone for coming. If you wanna see a sneak preview of a thing Casey built that we're probably not to show supposed to show you yet, here it is. So this is in in answer to Brad's excellent question. So this is you can basically say, like, hey. I'm a whatever. I'll I'll do me for alertmouse@b2b. My website is alertmouse.com. Google alerts alternative that tracks all your mentions. Who do I wanna reach? I'm looking for entrepreneurs, founders, creators, writers, and anyone who needs to track their brand name on the web. And we're gonna say United States. And then, basically, what this is doing is it's going to SparkToro and then trying to figure out if there are multiple different groups. There probably will be since I specified multiple different groups and what those groups are, and then it will break them down and show me, for example, solo brand builders versus content content hustle crew. Alright. Yeah. What this is doing is it's basically once it's got the data, it pings an a pings a, an LLM and says, like, hey. Describe this group based on the behavior, and then that's what you're seeing the description, but the data behind it is SparkToro data. So then I can get these multiple audiences, I think. So in this case, five. There's five different audiences that we sort of identified that have unique behaviors for this particular group. And, yeah, I mean, I'm I'm pumped to see that come to life. I think it, wouldn't surprise me if in the next few weeks we run some tests with that, and and you can see it inside the I I'm gonna guess that Amanda and I will probably talk about it on LinkedIn before you see it anywhere else. Yeah. I don't think what happens. Alright. Sorry. I wasn't quiet. I not good at multitasking, No. but we great. have I saw you in the background, just running everybody's reports and showing them all the stuff, and that was killer. fun. It's. fun. Yeah. Yeah. And those all have the open share access now. It's all good. Enjoy the PDFs, and check it out, friends. You know. what? Can you get this to spin for yourself? I I'm gonna say it. Buy a subscription. You're getting access to a ton of data that is verified. It's vetted by us. There there were also AI interpreted pieces here, but this is really high quality data. And I know that, not everyone has budget for these things, but we we really do keep this pretty reasonably priced for the data that you're getting. So hopefully, you'll give it a try. If if I mean, I love doing these custom buys from Daedos and, like, seeing the full click stream, you know, like like Will's doing. But but we're talking about tens or hundreds of thousands of dollars to spend. So, yeah, your your mileage may vary. Thank you everyone for coming. We truly appreciate it. This recording, will go out to people in the next few hours, Amanda. Yeah. But it's it'll also be available here, like, on this very window through your magic link in, like, five minutes or so. Amazing. But Alright. yeah. Thanks, everybody. Alright, friends. Take care. Bye. Ciao. Ciao.