High Intent
High Intent explores unfiltered lessons from the frontlines of modern marketing.
Hosted by Justin Rubner, High Intent features interviews with innovators in brand building, demand generation, sales conversion and beyond.
- Why are CMOs disappearing?
- What behavioral science principles can marketers employ to ensure better conversions?
- How should a new marketing leader scale an early-stage company?
- How can we get sales reps to follow through on cold MQLs?
... These are just some of the topics we explore.
About your host: Justin has spent his marketing career building and scaling marketing programs — for tech startups and Fortune 500 companies ranging from NCR to CoStar Group. He also was a brand strategist for the Air National Guard.
A former business reporter, Justin brings positioning and storytelling front and center to his approach to ensure clarity, pipeline growth and long-term brand equity.
High Intent
How is AI Impacting Market Research?
Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.
We're drowning in customer data, customer feedback, website analytics, social listening, surveys, reviews, and AI-generated insights. Why do so many companies still struggle to truly understand their customers?
After four decades studying human behavior, Mark Michelson has some ideas on how CMOs and CEOs should leverage AI to get better research results.
As founder of ThreadsMR, Mark has 42 years of marketing research experience with clients ranging from Google to Coca-Cola.
Today he bridges traditional research expertise with artificial intelligence, teaching professionals how to use AI tools without losing the human judgment that actually make insights useful.
He’s also the founder of AI Collective Atlanta and heads the Southeast region for the popular group, where he focuses on practical AI education.
Mark is next on High Intent, where we explore unfiltered lessons from the frontlines of modern marketing. I'm your host, Justin Rubner.
How exactly is AI impacting market research? Can composite digital personas round out your research project? Mark Michelson can answer these questions and more. As founder of ThreadsMR, Mark has 42 years of marketing research experience with clients ranging from Google to Coca-Cola. Today, he bridges traditional research expertise with artificial intelligence, teaching professionals how to use AI tools without losing the human judgment that actually makes insights impactful. He's also founder of AI Collective Atlanta and heads the Southeast region for the popular group, where he focuses on practical AI education. If that wasn't enough, Mark's a professional musician and at one point fronted a Pink Floyd-led Zeppel cover band, appropriately named Pink Zeppelin. He's the next guest on High Intent, where we offer unfiltered lessons from the front lines of modern marketing. I'm your host, Justin Rubner. Let's do this. Well, Mark, thanks for joining High Intent. Very happy to have you on the show.
SPEAKER_00My pleasure.
SPEAKER_01We're going to be talking about research today, and I'm very glad to have you on. You've been doing this for 40 years? 42. 42. Okay. Great. I want to get into the basics first. Your firm Threads MR conducts online and mobile qualitative and quantitative research. Let's get into the very basics here. Can we talk a little bit about the difference between qualitative and quantitative and their use cases as well as results?
SPEAKER_00Sure. I wrote an article on this once talking about qualitative and quantitative. And I'm going to share what the essence of that was. I think it demonstrates it well. A survey was done across the USA with all people. Let's say we had a thousand-something interviews, which gives you a confidence margin of plus or minus three. That's where most all your national polls are done. Okay. And I just asked the simple question what is your favorite color? Right? Nothing else. You'll find that purple will be in the top 10 easily. And among women, it'll be in the top five favorite colors, right? If I only use that information to make decisions as a manufacturer for things like cars, how come there's not more purple cars on the road? That's because we didn't ask qualitatively, what do you think of someone who drives a purple car? And how would you feel driving a purple car? So qualitative is about feelings. It's about in the schools of anthropology, sociology, and psychology. Whereas survey research or quantitative research is predictive by nature. It's sampling, it's not a full census, but it is sampling a market so you can then make predictions and feel confident about those answers being within a confidence range. That confidence range is determined largely by the sample and more importantly, how the sample's drawn. For it sits in Atlanta. If you're trying to be representative all of Atlanta, you wouldn't do everything in say Alpha Retta and say that applies to all of Atlanta. It just wouldn't be true. You have to sample all over Atlanta, right?
SPEAKER_01Yeah. So, Mark, let's say we're doing some research on brand perception. What would be an example of a qualitative result and a quantitative result in that research?
SPEAKER_00Okay, so one has to be assume they're already familiar with the brand to have a perception. So it's not an awareness study. You'd probably be dealing with someone who's in the category, buys on the category, and is familiar enough with the category to have a perception. Okay. So I might start off with qualitative and just ask what their feelings are. And a lot of times we'll use projective techniques and qualitative. In other words, it looks like this. If this brand were an animal, what kind of animal would it be and why? Right? Or if this, you know, if this brand were a town or a city, what would describe that city? So you describe something else other than the thing you're talking about. Subconsciously, you're talking about that thing, right? Right. Once I understand some of the base perceptions, and I'm looking for the high end to the low end. So it's, you know, it's a little tiny town to nothing to it's a giant metropolitan, right? Once I understand that, then I can kind of quantify those things better. I can describe them more clearly to take to a survey and then find out how many people agree that it's one of these five perceptions. Okay. And maybe even measure to the degree, like on a scale of one to ten, how well does this fit? Right. So you can get that kind of scaling going on as well, like your scales.
SPEAKER_01What are the different types of quantitative research?
SPEAKER_00In primary research, which is where we go out and do the questioning, all right. So we ask the questions, the primary method that we use is survey research. And survey research can and used to be a bunch of different types, like door-to-door was a thing when I was a kid. Then you had like mall intercepts and then telephone, and then that was uh wiped out by the internet. But you know, back in the 90s, we had these huge debates over whether or not telephone and internet were going to give similar results. You know, could we trust the internet to deliver reality or at least something that is close to reality? So these big debates happened, and it turned out there was really no differences.
SPEAKER_01Sure. People are people, no matter what how they express it.
SPEAKER_00Right. And and we could do other things with online that we couldn't really do in person. And that would be like conjoint analyses, where you like if this, that you know, which of these two do you like best? You know, well, okay, then trade that out. Now, taking that one, look at this one. Which of these two do you like best?
SPEAKER_01That's conjoint.
SPEAKER_00Yeah, it's it's called trade-off analysis. So consider jointly is where it comes from. Conjoint. Consider jointly between these two or three things, which one do you like and why? And we might just change features. That kind of stuff gets real complex when you're trying to analyze it and design a study in person. So internet allows us to do even more than we could before. Also in showing videos at scale. You know, if you want people to look at a scale, a video, and give a score on it of some type, be it a Likert scale or scale one to five or whatever. Now, the other thing that happens in quantitative is not only a survey, but in customer experience, a lot of these thumbs up and thumbs downs you see after every transaction, that's quantitative research as well. Okay. So that's that's on a customer experience measure and the touch points, whereas a survey of opinions, you know, is a little bit different than hey, I felt that this product was delivered correctly. Yeah, it was complete. I like the product, I'm happy. That's a different kind of measure. And there's a lot of different, smaller nuances within the actual question types. You might have heard of things like an NPS question, you know, net promoter score, zero to 10, and that has its own analytics, you know, where you take the top eight through tens and then you subtract the ones that are fives and below, come up with a score, right? And there's a lot of those types of question types that we come into play as well. The main advice I would ever give anyone is don't overload every question you can think of and do a survey. Sure. Do them over time. Don't have more than about maybe 10 or 15 questions in any survey because people just burn out.
SPEAKER_01Do you find a lot of clients want to pack a gazillion questions in there?
SPEAKER_00Every one of them.
unknownYeah.
SPEAKER_00You know, they're like, oh, what about this? And then you get some other department involved. All of a sudden you got a 70 question survey. You're like, hey, come on.
SPEAKER_01Okay. All right. So 42 years you've been doing this. What is the biggest change that you've seen in how companies understand their customers?
SPEAKER_00Wow, there's a lot of them. I would say that uh coming to terms with using the web to understand people at scale. Okay. That's gotta be the biggest change I've seen in my life so far, right? So being able to reach people around the world and engage them in a long-term ethnographic diary study using mobile phones, we were doing that, you know, the minute mobile phones came out, right? Because we had people that we needed to understand their daily rituals and habits regarding, say, skincare, and particularly women, you can't go live with them very easily. You know, they get routines all day long from the moment they wake up till the time they go to sleep. What products are they using? When? Why? These aren't beauty products, these are skin care or prevention products. Now, huge differences between Japan and Brazil, even between Italy and Sweden. So if you're doing this in 30 countries over a period of, say, a month, and you've got maybe 30 or 40 people in each country at scale, we couldn't do that before. Mobile phones were available, we couldn't do that before we had the analytical capability to interpret all of these videos coming in and all these photos and comments. So that kind of stuff really revolutionized the marketing research industry, particularly in the qualitative. Quantitative, you know, you used to have phone rooms of people sit around and dial people all the time, telephone rooms, call sounders. Okay. So they would do outbound telephone calls to get people to answer surveys. You can imagine the amount of labor just in not being able to connect with someone. Sure. And auto dialers and all this stuff going on, that whole almost entirely that industry is very small now. And it's targeted to really hard-to-reach geographic areas. Like if I have to do Wilmington Island or something, yeah, then I know I can just call in that area, right? But even now with mobile phones, it's hard to reach those people. The second coming, which we're still experiencing, is the AI revolution. Within AI, you know, knowledge work in general is made a lot easier. I mean, yeah, I'm talking about doing your analytics around with quantitative, using Python and all these types of tools. And we're used to using uh, you know, Tableau and some of the other analytic tools. They're all built in within AI now. Yeah. And dealing with unstructured data, which is the nature of qualitative, it's a lot easier. Put all that stuff in a notebook and define trends automatically. Well, you know, however I want to slice and dice, you know, is up to me. Now the thing is, is with qualitative, it's not predictive. You know, it's not intended to be, it's more directional because it is emotional. It is what one or 10 people think, right? I wouldn't use that as a basis for a multi-million dollar decision. I would find out what they think and then roll it out to survey to see how many people agree with those thoughts, right? And then get ready for a decision time.
SPEAKER_01Well, let's just get down to it. How is AI changing how companies do market research?
SPEAKER_00So the first thing that's happened is, you know, when AI came around to the market in 23 and more widely adopted in 24, was a lot of experimentation on it. But the first thing that happened really in the research world was a lot of the existing platforms that we use as tools embedded AI in different areas to give you help with, you know, editing some sort of comment or statement to actually designing a whole study. Right. So these tools are have been out there. There's a whole world of research tech and technology. A lot of companies well invested into things like recruiting, uh hosting asynchronous focus groups, uh bulletin boards, mobile diaries, an AI survey room. Yeah, there's that too. I mean, you could have an AI survey room. And one of the more interesting developments that I took on really early back in 23 and 24 was the idea of synthetic respondents. Okay, so synthetic respondents come across on multiple layers. There's quantitative synthetic respondents where you're basically filling in missing surveys. So if I have a sample I need to reach of, say, 500 people, and there's really only 300 that are reachable. The other in try and try, yeah, at some point you just exhaust your try, it doesn't match the juice ain't worth the squeeze, right?
SPEAKER_01Of finding the extra people.
SPEAKER_00Yeah, right. So there's a way to model the data around the answers so you don't change any of the answers, but you can add in the additional people based on synthetic persona modeling or synthetic response modeling, right? Those two are thinly sliced. One means let's look at all the responses and then build our models around that individual profiles, right? Because you don't want to amplify extremes on that kind of thing. And the other one is building a qualitative digital individual, and that comes in two flavors. That sounds really weird, Mark. It is, it is. There's two flavors of that. One would be the composite digital persona. That's where we go out and find all the people we can that have a footprint, a digital footprint, that are in that category. And it's usually hard to reach people. Now, I'll create that model based on, say, five or ten people who are known in that industry. If, for instance, I just did something with countertops, right? So this is high-end countertops. We needed to talk to installers, to architects, to interior designers, and all those types of people, right? They're not easy to reach. Yeah. Yeah. So I could create a digital persona that's a composite of the top 10 architects that use marble countertops, right? They're known for it. They've got articles on it. And so I do a search for those people first, and then I take everything they ever wrote and put it into the digital persona.
SPEAKER_01All right, I did not know that this was a thing. Uh, what uh what tool are you using to synthesize this?
SPEAKER_00You can use any of them. I like to use notebook personally, right? I use the combination of perplexity for my uh internal research. So the first, you know, getting the persona built. I use perplexity because it is sourced. The key to a good synthetic persona is having grounded data. So you know that that data is is vetted, you know, it's not just made up stuff. Unlike the kind of thing where you go and say, You are an architect, you know, and you ask your AI to pretend it's an architect, right?
SPEAKER_01But it's it's actually grabbing stuff from architect.
SPEAKER_00Correct. It's it's going back and it can bring that forth and say, as cited here in this article, you know, this is what they would believe. Now that's composite synthetics personas. Now, there's nothing like Whoa, whoa.
SPEAKER_01What do customers think about this when you say that, oh, I'm not interviewing real people, I'm interviewing a synthetic.
SPEAKER_00Well, the you do a combination, so you don't really blind, you know. I wouldn't count on all synthetics to give you truth at this point, okay? But I would say it's a good supplement to other research. And a lot of times when we're doing the research, we may take the actual interview transcripts and build them into the models as well. Okay. So what happens in reality is we may do an interview with a real person at some point and say, hey, what do you think of this? This is. We might spend an hour or two with them and capture the whole transcript. Now, that person's hard to reach. We can't go back to them after we have a new through our concept or five ideas to change around, right? Like, what do we call it? You know, what are your big concerns? We already know some of that. So we can then ask that persona those smaller questions along the way. There's another type I want to cover, and that's the digital twin. Digital twins in general, the the phrase is misused or it's used a lot of ways. A digital twin is often used to just describe like a copy of myself, right? And I might even put a video on there, it looks just like me. It's made on Hey Gen, or even use my voice from 11 Labs. That's not the type of digital twin I'm talking about. There's another one that just says, Hey, pretend you are a, you know, a buyer at Best Buy. You know, and that's not very specific. Uh let's just kind of pretend you are that, and the model might pretend it. But if I wanted to go out, let's say I'm gonna make a call to Best Buy, and I wanted to prepare myself for that. And I and I don't even know who exactly to talk to, right? I'm just making this up. But let's say, let's say I want to sell mystery shopping services to Best Buy. Well, first thing I want to do is who is in charge of the programs of mystery shopping? How are the decisions made? What is their history with mystery shopping? What would they think if you know mystery shoppers awarded their salespeople for saying the magic phrase, right?
SPEAKER_01In instead of just getting them in trouble.
SPEAKER_00Well, yeah, or doing an audit, right? So the thing is, is I created a whole board of directors of Best Buy, the real people. I just like here's here's so and so. She's the CEO. Get everything she's ever written or done, compile it into a large document, right? A markdown document for that. I might then say, okay, let me have the marketing director, let me have the operations director, and go drill down to where it gets to the VP level or whoever's in charge of this program, right? I found out just by asking the persona, just asking these digital twins, I said, Hey, what would you think if we had shoppers reward your employees for promoting a product or service? They said, if we found out that a vendor was doing that, we would cut off all relations with them immediately. I was like, whoa, why? We don't want that happening in our stores. We don't want them to every vendor out there sending in mystery shoppers under yeah, it's no control. There's chaos ensues, right? And now no longer the employees, they're just trying to win. Right? So and I was like, okay, well, what if, and I'm just talking with the twins, I'm like, what if we had a program where you're in control and the vendor of that month, whoever the featured company is, let's say it's Sony, I'm just making it up. Let's say it's Sony, right? And they got a new virtual headset. And so everyone in the in the company is trained to say, Hey, have you tried the new virtual headset? Right. And if they just say that one thing, then they're rewarded. Now you're in control, Best Buy, of that, right? And we're gonna send the reward to your manager to deliver to you. They're like, Oh, that's a great idea. We love that so fine twists like that will prepare you for your eventual real-world call with them. Uh this your pitch.
SPEAKER_01So, how has AI impacted what you do in the best way and the worst way?
SPEAKER_00Okay, I'll start with the best way. It's made analysis and design of studies a heck of a lot faster.
SPEAKER_01I can totally imagine.
SPEAKER_00Yeah, yeah, a lot faster. I mean a hundred times faster. Like you know, there's no way at scale I could do what it's doing, right? So that's the best. There's some interesting new tools too. I find that people will talk to an AI more truthfully often than they will to a human. That's surprising. It is surprising. But if they know it's an AI and they know it's anonymized, you know, then they'll be like, okay, yeah, I've got some deep secrets or whatever to share.
SPEAKER_01Yeah.
SPEAKER_00It's it's odd, but that doesn't happen in real world. I'm really a serial killer. Yeah, well, you know, for instance, I do a lot of work with cancer patients, right? And I have a lot of big history with it. When you're dying and you know you're dying in the next three months, or your family knows you're dying in the next three months, you can't just talk about cancer. You can't ask, hey, what do you feel like you're dying? You know, it doesn't work that way. Instead, we use these projective techniques, right? So I might use pictures of landscapes and say, hey, before you were diagnosed, which one of these represents your, you know, how you felt on an ongoing basis? Just pictures of landscape. Everything from volcanoes to to islands, you know. Yeah. So they can pick that out. Now tell me what you see in this picture. They can talk about that all day long. They're really assessing their own emotions under underneath that, how they felt before. Now, the moment you were diagnosed, you know, where were you? You know, it's usually the fiery or murky, foggy kind of you know, thing, right? Talk about that. What do you see? And they talk about how they feel if they were there.
SPEAKER_01But generally, people are more open to an AI chatbot.
SPEAKER_00Yeah, just like they are with projective techniques. You know, they can't access and talk about what they can't really talk about, yeah, without breaking down entirely. But that's the good stuff. Now, the bad stuff is that a lot of clients are believing everything AI says. And there's a lot of hallucinations that happen if you're not dealing with better controls. For instance, I do all of my analytics in rag systems, where I'm in control of all data. Okay. I don't let it just go out to the web. What is a rag system? Uh retrieval augmented generation. So, notebook is a great rag. Gotcha. Because you can control the inputs and say, only deal with these documents I gave you. Don't go to the web. No hallucinations. It minimizes all that. Okay. It's grounded data. And I may put first party data in there, like my energy transcripts, right?
SPEAKER_01But you can't go out to the web.
SPEAKER_00Yeah, there's a way to, but I don't need to. I don't want to, right? I want to deal with the data that I feed it. So that's vetted data, right? It's not just like made up something. And the AI will make up stuff and confidently lie to you all the time. Um, so understanding what happens with that is key. And there's a lot of people that are working at companies that may go with whatever the AI says without understanding.
SPEAKER_01Could you give us an example of a lie?
SPEAKER_00Yeah, sure. What's the market opportunity for X? Okay. And there's a lot of people that are right now. I just saw something on, I think it was 60 Minutes or one of these. It was talking about this guy who had an idea for a business. He put it into AI. The AI told him it was brilliant. No one else in the world is doing what you're doing. Yeah, whatever it is, right? Exactly. Yeah, that sounds like a good one. So, you know, he like literally dug into it, and the AI kept telling him this is fantastic. There's no one doing it. The market opportunities, gazillion dollars, you know, and all this stuff, right? Everyone's gonna want your thing. Well, he gets a second and third mortgage on this thing, spends a year digging into it, and then finally he's like, he's getting no sales, no traction whatsoever. Nothing. Everyone's laughing at him. Like, you know, dogs don't take Ubers. You know, it's not where we're interested in, right? And uh, and so you know, he finds out eventually he goes back and says, Hey, you know, someone told him, say, ask ask your AI to be honest with you and tell you, you know, what honestly, you know, where are you getting this information from? And the EI came back and said, Oh, I thought you were being hypothetical.
SPEAKER_01Well, if you were choosing Chat GPT, chat GPT loves everything you say.
SPEAKER_00So right, exactly. The whole sycophant thing led with the whole, you know, confident lying. Yeah, so that kind of stuff can get companies in trouble. And I think most have figured that out by now if they're dealing with it. Just bad things, yeah, just bad things in general. There's there's other things that'll do, like it'll, you know, you can get one answer on one AI with the exact same question, but in another AI, you get a different output. Sure. A lot of it has to do with your own personal settings you have in place in the exact same prompt, but also how the AI is trained, right? If it's trained on Grok data or on X data, you know, it's gonna be quite different than if it's trained on, let's say, encyclopedia data.
SPEAKER_01Yeah, what will always require humans in the world of research? I think I know you're gonna how you're gonna answer this, but uh food work, you know, tasting stuff, touching stuff, that is not yet feasible with AIs.
SPEAKER_00Okay. So I I just did some taste testing and concept to fit to brand for a large chain of restaurants across the country, right? So having people in a room, they can smell it, they can taste it, they can look at it, they can in real time. That's not going away anytime soon.
SPEAKER_01Can we say it's not going away at all?
SPEAKER_00It's not going away at all. Um now I don't know that I don't know that I can cockhorn say that because you can model the tone just like they've modeled neurons for AI. AI is basically a neuron model, right? So if you've got specialized neurons for things like sight and sound, which they have, right? Uh it's just a matter of time, I think, before they have touch and smell, and smell is related to taste, right? So if they train it on these smells, which translates to taste, and it tastes a little different only in the sweetness and spicy areas, but they're connected really close to each other, smell like tongue and the nose, right? But the idea is that having those senses, those other senses, is is there. The other thing I think that it's never going to be able to approach is the idea of emotions. Okay. We don't even understand our emotions sometimes. They just hit us like the weather. You know, even having a gray day, all of a sudden, oh yeah, whatever it is, throws out of whack, right? Like I heard this one talk about an AI is like a baby, right? But when the baby falls out of the crib, it cries and there's consequence, right? When the AI falls out of the crib, there's no consequence. It doesn't feel it, it doesn't cry, it just keeps on going. Right? That's a huge difference and a big gap for training AI models to be able to understand human emotion as well as the senses of touch and smell and taste. Now, I've seen some of the models trying to get to touch, right? Yeah, trying to fake skin and all this kind of stuff. Um, it's interesting. You know, I can deal hot and cold, you can do wet and dry, right? But you know, when it gets the nuances like a silky, smooth hair versus rough hair, you know, it's it gets nuanced within those areas. And that those are areas I can't say it won't ever, because there's a lot of people trying to do that, but particularly it's gonna be helpful with robotics. If you're dealing with robotics and they have to be able to detect certain smells on, say a battlefield or something, like a poisonous gas. Then it has to be trained on that, and then it can warn you know troops behind it that that's happening.
SPEAKER_01The X is coming, yeah.
SPEAKER_00Yeah, yeah. So that's another example.
SPEAKER_01Very cool. I'm gonna avoid the ethical questions around human emotions because I think everyone's talking about that, but it had some pop in my head. Let's go to all right, if I'm a CMO, how should AI change how I gather customer insights?
SPEAKER_00It would change only because it's gonna be ordered a certain way. When I say ordered, the output is gonna be structured a certain way, right? And it's kind of AI is kind of training us to think in terms of clean output, right? Sure. What is good structured output, right? I think it will go down an avenue and uncover some of the stuff that maybe is not being considered as far as different media channels and messaging, right? So if I'm a CMO, I'm looking at media, I'm looking at messaging, I'm looking at budget. You know, what's my reach? What's my cost per person? And what's my message that that's going to change someone's feeling about us somehow or another, right? Or create a feeling if no one exists. Those areas, AI can really dig in and uncover the what I call the ends on the on the bell curve, right? So everyone can get the middle of the bell curve, well, you know, table stake stuff, but it's out here on the fringes that you know, AI can help uncover some stuff as far as ideas to explore. Now, whether that's a large enough market to go after, it'll give you some idea of that. Give me an example. Okay, so let's say I want to have a new chewing gum that's energy chewing gum, right? I'm just making that up. Energy chewing gum, right? And and so I can go out there and do research on I want to know all about the chewing gum market, who's chewing gum, and then I also want to know about the energy market, right? So I'm kind of synthesizing those two together. AI will do that more easily than I will to find the commonalities and the corporate points of that idea, right? And it'll tell me, okay, well, look, you know, chewing gum is good, you know, it might appeal to kids. You know, we don't really want that, you know. So it might appeal over here to, you know, an adult who already chews gum uh to give the extra booster in the day where they you know don't really want to drink a drink or something, but chewing a piece of gum four afternoon is a good thing for them, but they don't want to be kept up all night. So it's kind of like fine line of the benefits and synthesizing two disparate ideas, right? So that that would be one example that just pops in my head immediately.
SPEAKER_01Cool. If you can only do one, what's more valuable? Would it be a focus group, a survey, or say, analyzing conversations with customers from the product and sales teams? If you could only do one of these, would it be conversations, focus group for service?
SPEAKER_00It's going to be based on the objectives of what you're trying to achieve. Are you trying to solve around what? What are you going to do with the answers?
SPEAKER_01Let's let's for the sake of this, let's just say it's uh our positioning, we feel it's wrong, and we're trying to improve our positioning.
SPEAKER_00Okay. So here's the thing about that. If I deal with my existing customers and salespeople, the salespeople alone, if you're relying on them to convey the truth of that conversation, they're going to make themselves look good. Okay, no matter what. And and the customers love us and they love me in particular. Now, is that the truth? I would begin to question that. But then I would go out and say, okay, if how about the people who are not doing business with us right now, or maybe people who have done business with us and left, are they good sources of information on position? Now, I could actually answer a lot of that without even talking to anybody because AI is good enough to understand positioning and it's good enough to understand most industries, right? So if I want to know how my positioning currently sits among the industry and what are those perceptions of it, AI will give me what it has and what it's trained on, and again, the sources of what it has, right? It's not just making stuff up, it's according to this journal in this place. Here's what they say about this, right? Now, I would start off by talking to AI and see how it thinks positioning is currently, right? And I might ask it, what are some opportunities that you see in openings for my product line or service line in that area that's not, you know, that has a good opportunity. And it might reveal some of those, but then I have to go out and explore or examine those opportunities, right? Let's say I take a low-hanging fruit and it says, Wow, you're missing a whole world market out here with 18 to 24-year-olds. If you just shift this color in your whatever, right? And I'm like, okay, maybe that's true, but I now I need to go talk to those people. And those people who are never been my customers, I want to be able to get their actual feedback to concepts without swaying them one way or another. So I have to present that concept along with like three or four others. So they're not just living in a vague, hey, what do you think of this? Yeah. So what do you which one of these do you like and why? Right. So to answer the question, what is the best survey? No, I would say not. Okay, because I don't know which questions ask yet, right? I would say that the source of talking to my own salespeople is probably going to be biased, is unless I have really direct input from them on stuff like transcripts, right?
SPEAKER_01Well, that's what I'm talking about. You're listening to calls.
SPEAKER_00All right. So if I have transcripts, I can put those transcripts in and bypass the salesperson, right? But typically, you know, are those outgoing sales calls or are those incoming complaints? I have to understand why they are having this conversation.
unknownYeah.
SPEAKER_00You know, if it's a sales call and that's recorded, then I can sort of see what the salesperson's positioning a company has and whether the company perceives it, the customer, the target guest, whether they perceive that same positioning or where are they pushing back? So if that's the case, that's cheap beta. It's already existing in call center or wherever that's safe.
SPEAKER_01We'll have to pay for it. Yeah.
SPEAKER_00Right, right. I'm gonna go towards options first so I don't have to pay for. Okay. And then, you know, as I whittle it down, there might come a time where I say, ah, I really need to talk to people to see what they think about this before I make a decision or a recommendation.
SPEAKER_01So for the initial research, you would go and analyze the calls and maybe use AI to synthesize it, right?
SPEAKER_00Yeah, yeah, AI can definitely synthesize.
SPEAKER_01And then you're gonna be probably left with more questions, more fine-tuned questions after that, in which case you could potentially do a survey.
SPEAKER_00Yeah, and I also want to look at you know, size or opportunity. If it's repetitive and comes up all the time, serendipitously, you know, then you know, then it's something to look at, right? If it's just one person's idea and no one's really resonating that or echoing it, then then I don't want to pursue that. Yeah. In many cases. But then again, research is one of those weird things, you know, we can only deal with what we know now. And if we'd asked that question, you remember when uh smartphones came around? Well, your first thought was, I've got a camera, I've got a phone, I've got a uh an iPod, I've got all these things, I've got my GPS. Why do I need all that you know in a phone? What? So Steve Jobs is famous for never doing any research at all. He just refused to do it. And Apple still has that kind of way of going about themselves, right? But if we'd asked the question, then they may have not made the smartphone. You know, if you ask questions about wow, you know, how would you feel about getting into a metallic tube with these you know jet engines you know strapped onto wings and people you know, throw peanuts at you, crammed into a seat?
SPEAKER_01Yeah, no.
SPEAKER_00So, you know, a lot of times we don't have, we can't even access what we're asking.
SPEAKER_01Yeah, that's Apple's a very interesting example on that. I definitely have a strong feeling on this, but I'm gonna ask you, as marketers, are we becoming overly reliant on analytics dashboards to the point where we're not spending enough time actually understanding and talking to customers?
SPEAKER_00That's true in customer experience more than in you know marketing directors, right? In customer experience world, I mean, they're measuring every interaction, every touch point is being measured, right? Some way or another, either through a voice of customer or through operations metrics, right? So, you know, what they're doing there is they're trying to figure out what's broken. Is there something wrong? Is it systemic and what's the cause? Right. So we can correct it, you know, don't just put a band-aid on and don't ignore it. So dashboards for customer experience people are important, like having an alarm that goes off. It's kind of like, yeah, I've got sensors in the world, and and if one of them tings a little bit, yeah, I've got a problem, right? Well, then I have to go examine that problem and dig into it a little more. So as a first stage kind of alarm system, I think it's good to have dashboards, especially the ones that will notify who needs to be notified down the line. Now I'm used to dealing with large retail chains like Walmart, where they have districts, divisions, regions, you know, everything you can imagine, and directors of those, even down to departments, right? So we can make that message go to the right person, direct it to whoever needs to be notified to fix it is notified, right? And now that that's within the world of CX, which is kind of in marketing and and also not because customer experience, you know, it's supposed to be systemic to the whole company. Sure. But in large part, it lives in operations, not marketing. Marketing is what creates the expectations we have around a company, right? And those expectations have to be delivered by operations, and CX I see as fitting in between those two to make sure we're not making stuff we can't sell and make sure we're not selling stuff we can't make. CX kind of was created it with that in mind, but it became such a point of let's measure everything in the world, and and if you don't buy our million dollar subscription to our platform, you know, you're not doing CX right. Yeah. And I think that that also was a fallacy because now you're you're basically training people how to make dashboards instead of understand what the data is saying. Now, the data may give you an alarm, you know, there's an alarm going off, and so you upstream the message or whatever, but what is the cause? You know, getting into that level of repair or correction, you know, takes a little digging deeper. So we ask these questions of metrics like it's scale of one to ten or scale of zero to ten, right? I'm gonna ask you to give it a rating, but I'm only doing that to anchor you. I really want to know why you gave it that rating. And that's an open-ended comment. AI can now translate those and analyze that in the moment at scale. And I remember not too long ago, not more than six years ago, or maybe seven, that that there was a whole bunch of tool makers out there creating tools just to read those comments at scale so that people could make decisions faster.
SPEAKER_01Faster, yeah.
SPEAKER_00Right. Now AI does that, you know, pretty much instantly. Yeah. But again, it all depends on data being fed in, right? So, you know, you want to have a specialized tool looking at that, not something that's gonna hallucinate and add to whatever is coming from the field.
SPEAKER_01What would you say is a waste of money right now in research?
SPEAKER_00Oh, gosh, that's that's a good one. Well, I hate to say it, but you know, sometimes these incentives get out of control and the uh and the recruiting costs get out of control. So trying to recruit in the really hardest-to-reach people for traditional research is a waste of time. You know, getting surgeons to show up at a focus group facility, it's a waste of time. I've had to do that many, many, many times. It's a lot easier to go to them wherever they are.
SPEAKER_01And do a one-on-one.
SPEAKER_00Correct. Do one-on-ones or dyads or triads. So traditional study design, trying to apply it to the real world doesn't work always. You got to be creative and how you get information from the qualified people. And paying them a lot of money is not always the right way to go, right? Sometimes they'll do it because they're interested in the topic. I'd rather talk to someone on that than just because I'm giving them 500 bucks. So that I think is an area of work that we have come to, as researchers, you know, get lazy and we're like, ah, just pay them more money, you know, you'll get them. I I gotta get those last 10 people to interview, right? Throw more money at them, you know, find them on the golf course. I don't care.
SPEAKER_01Yeah. Yeah.
SPEAKER_00That that I think is is an area that needs some help. Secondly, is just this whole idea of competitive intelligence, right? And that's a whole industry. And I have friends in that industry, and I think that a lot of that is a general thought is available on the web because you're dealing with what's going to be publicly available anyhow. Uh competitive.
SPEAKER_01You still have to synthesize, you still have to find it and synthesize it.
SPEAKER_00Yeah, but a competitive intelligence professional can't pretend they're a customer and mystery shopping. It's unethical, right? So the AI will deliver a lot of stuff that you might try to get by mystery shopping, like pricing, or like what's their main pitch? You know, what's their benefit draw they're doing now? Buy one, get one free offer, or whatever it is. AI does a pretty good job of answering those questions if it's publicly available information.
SPEAKER_01Right.
SPEAKER_00You don't need to spend money on that.
unknownYeah.
SPEAKER_00In the past, just think of all the different subscriptions you had to have to different you know industry journals and so forth to be able to get information. For instance, I don't know, you say I had to subscribe to Arbitron and Nielsen just to get the radio to be able to make media buys. Uh and nowadays, you know, a lot of this stuff is more is available in without subscriptions.
SPEAKER_01Yeah, that's really good insight. I would not have thought of those two things, but competitive insight and going out of your way to pay people to be a part of a focus group or something. Yep. Okay. All right. So we are drowning in customer data, feedback, etc., website analytics, even social listening tools I've used. Why do so many companies still struggle to understand their customers?
SPEAKER_00First of all, a company has a lot of different kinds of customers, right? Uh typically, unless you're hugely, you know, all I do are root canals. That's your if that's your thing, you you know, it doesn't matter who your customer is, just knowing that that you're just a root canal person, right? But if you're like a you know a dentist in your in your market area, right? I I want to know at that level who's in my market. And I've got my regular customers, so I want to make sure they're happy. So, you know, just some sort of informal, hey, how do we do today? You know, kind of thing, as well as how do I get new people? All right, those two marketing initiatives, keeping who I've got and getting more like them. So I might have to do a little bit of research, and and that's a good example for someone with very little budget, right? But how can I get more people in my market who are just like my customers? And I might float out some promotional concepts among my existing customers to use them to get more customers. So testing that idea, you could do it one of several ways. You could either just come up with it and try it. That's not a bad way to go. Because you're actually doing something as you're learning. And the other way is to examine it, you know, through the lens of an AI and say, okay, this is my market area. Who are they? Who are these people? Who is the main dentist? For instance, you can go in there right now and just say, give me the top three dentists in X zip code. And it'll tell you if these people are most visible because they have websites or they have more traffic or they claim to do X, right? And that's all based on their own claims and their own communication of who they are, as well as maybe some reviews. And it'll read those reviews too. So the dentists, you know, whether or not they're taking time to even examine existing data is a question. You know, because I can go out there, even though you're saying we're all swimming in data, but knowing that you can go and say, who are the top three dentists in this zip code, or maybe the four adjacent zip codes, right? What are they doing that people like? Why are they the top? Is it because they're cheap or is it because they're good? You know, why are they going there? So you can interpret a lot of the comments and see if people are happy with them or not. You can interpret some of the things that they're putting out on the web and then make some sort of hypotheses from that, right? And as a dentist, you're not dealing with multi-million dollar budgets for marketing. You're, you know, you're dealing with a small ship that can turn on a dime. So you can actually just go try those things, right? But you don't know what to try until you see what others are doing. You know, you're known as a high-end dentist that deals with you know cosmetic dentistry, right? Whatever. And all of a sudden you're offering a two-for-one on, you know, wisdom teeth.
SPEAKER_01Yeah. Right. Yeah.
SPEAKER_00You know, you've got to find your place, I think, is what you can do with that. And you're only going to do it through examining these data and asking the right questions. That's the biggest problem with most marketers is not asking the right questions.
SPEAKER_01Yep. After four decades of work studying human behavior, what is the most important thing that you've learned about people?
SPEAKER_00I think most people want to give you their real opinion, their real feeling about stuff. Many of them are doing it in an unfiltered way. If you don't let them tell you what they want to tell you, because your questionnaire or line of questioning is so narrow that it doesn't allow them to answer that. They're not happy about that. You know, we're not looking around a focus group or whatever it is, right? I can tell someone's come there really ready to share whatever they want to share. And my line of questioning might have nothing to do with what they want to share, but they've got to get that out, right? And they're not really ready to listen to anything until they get that out. So I guess it comes down to making people feel valued and that their opinions are important, right? And so then setting up your design, your research design to let them blow steam, let them do whatever they need to do.
SPEAKER_01Let them talk about whatever's on their mind.
SPEAKER_00Yeah. So let them get that off their chest early. And so that way you can go into the conversations you you need to have about whatever your ideas are, you know, to go to market, right?
SPEAKER_01I don't think your Coke machine has enough CO2.
SPEAKER_00Yeah, whatever. Yeah, exactly. That damn thing doesn't even give me the Coke after I put in the dollar or fifty or whatever, right? Yeah, some you know, whatever that is, they're not going to pay attention until they say what they want to say.
SPEAKER_01Right.
SPEAKER_00Okay. And that's almost universal.
SPEAKER_01Yeah. Your parting wisdom for uh any company, B2B, B2C, CBG, whatever, that is looking to do its first round of research. What is your parting uh wisdom to them?
SPEAKER_00Know what you're going to do with the research before you get it. All right. So don't do research just for the sake of doing research and don't ask questions you can't do anything about. Okay. That's it. I mean, yeah, there's so many times I see proposals come down. We need to do this research. And I'm going, why is what what what are you going to do with this?
SPEAKER_01Yeah.
SPEAKER_00What are you going to do with it? What can you possibly do with what are answers we give you? And are you willing to do any of those things? Yeah. If you're asking to do this research, because a lot of research is political internally. So this one department wants to do this thing, or this one person has an idea. Everyone else is like, nah, that's gonna work. How about my idea? Right. So that's where research comes into play, usually to kind of solve those internal disagreements, right? And decide on a direction. So don't do research that you already know the answer to just to prove it to others. Okay, right. I can go out and hire actors, and I've had that asked of me before from ad agencies. Go hire some bunch of actors, tell us how great the ad is.
SPEAKER_01Don't do research on something that you know is true. Yeah. What's if no one else at the company knows it's true?
SPEAKER_00Then there's ways to get that across. And you know, there's other ways than spending a whole bunch of money on a focus group or whatever. Right.
SPEAKER_01You know, for instance, I could just I've been faced with this dilemma personally.
SPEAKER_00Yeah, yeah, exactly. You know, try to bring together competitive analysis. There's so many creative ways to do with AI. I can go out right now and do a digital twin on anyone who has a digital footprint, right? So if I want to, I could create digital twins of my competitors if they have enough digital footprint and I can have them tell the story. You know, there's lots of ways to make that happen to convince your team, right? And if the team doesn't like you for whatever reason, you know, it's just not gonna work anyhow. You know, I mean, you're kind of doomed from the start. Even if the research says everything in the world's great and dandy, if whoever's up above says, no, I don't care what he thinks so much, yeah. Uh it's it's for waste.
SPEAKER_01Sure. That makes sense. Well, Mark, thanks so much for being on. I've learned a lot of things, and uh I think uh listeners are gonna learn uh a thing or two as well. How can people get in touch with you?
SPEAKER_00I would say that my website's a good place to start, threadsmr.com. It stands for threads marketing research. And there's not a lot on the site. You know, the best way is use the site, learn a little bit about what I do and what my company does, but then yeah, reach out, yeah, email, phone, LinkedIn, uh, or come to any one of the events that I run, you know, across the region.
SPEAKER_01Including the AI collective.
SPEAKER_00Yeah, those those are you know, I like in-person stuff personally, it's just gets me out of out of my my place. But you know, start the conversation, don't be afraid to ask questions. There's no such thing as the wrong question, yeah. You know, and so asking questions and being curious is where it starts. And then, you know, when you approach a professional researcher, you're like, okay, either companies already have an idea of what they need to research, right? And that they they have an RFB in mind with objectives and all that, or they're just like, I have this feeling that this thing may be a good opportunity. How do we figure it out? Yeah, so I'd rather deal with someone at that stage where they don't have a prescribed approach to doing the research as a commodity, right? Yeah, bid on this. This is what we're doing. You're gonna bid on recruiting 10 people that fit this criteria for five focus groups, blah blah blah, right? I see that all the time. And I'm like, well, is that the best way to do it? I don't know. You know, until I know the objectives and I know the context of what you're going to do with this information. You know, it's sort of silly to just answer a RFP for commodity style production.
SPEAKER_01Agreed. Well, I Mark, I'm sure I will see you at AI collective meetings going forward. And thanks again for being on. My pleasure.
unknownAll right.
SPEAKER_01Good way to Justin.