In this episode of Key Takeaways, Rakesh Patel of Allstate shares how he transitioned from engineering to market research and how that analytical foundation shapes his work today. He discusses the evolving role of brand marketing, the rise of AI tools, and the importance of mentoring the next generation of researchers.
In this episode of Key Takeaways, Chuck Murphy and Maggie Bright sit down with Rakesh Patel, Senior Manager of Market Research Insights & Intelligence at Allstate, to discuss his unexpected career shift from electrical engineering to consumer insights. Rakesh reflects on how an analytical mindset has shaped his approach to both qualitative and quantitative research, how the industry has evolved over the past two decades, and why brand marketing remains essential. The conversation covers balancing short-term performance with long-term brand building, adapting to faster timelines, and exploring the potential of AI tools. Along the way, Rakesh shares lessons on curiosity, adaptability, and mentoring the next generation of researchers.
Key Takeaways:
Quotes:
Chuck Murphy 0:00
Hi everyone. Welcome to key takeaways. I am Chuck Murphy, and today I am joined by Rakesh, Patel of Allstate, and Maggie bright, who is always here for these. And we are excited to talk to Rakesh about kind of a variety of topics his career, a little bit of AI, little bit of kind of brand versus performance marketing, whole plethora of topics. So thanks for joining us today, and we'll dive right in.
Today. I am joined by Maggie bright and Rakesh. Patel, Maggie had to run and do a quick errand, so she's gonna be here five minutes, but Rakesh and I are gonna dive in and get started, and Maggie can jump in when she is ready. So Rakesh is someone that I have known and worked with for a long time, but I am going to allow you to give your intro and tell us a little bit about yourself and how you got into market research?
Rakesh Patel 1:03
Sure. Well, first of all, thanks for having me, Chuck. I really appreciate being on your podcast here. So, yeah, a little bit about me. I'm a marketing researcher with the Allstate Insurance Company. I've been with Allstate for about 19 years now. As Chuck said, he and I have been working together for a long time, almost, almost the entire 19 years, maybe a little bit less than that, before that. So I was originally trained as an electrical engineer in in college. Did did electrical engineering work for a couple years, went back to get my get my MBA, and then after my MBA, started with the, I guess I would call it the original at&t. It was a long distance company. They competed against MCI and Sprint, you might remember from back in the day, and then they got bought and sold and dissolved, and anyway, so I started in marketing there, and I kind of fell into marketing research, kind of by accident. I was doing, like, you know, a lot of marketing and advertising work, working with agencies on developing messages for long distance calling and whatnot. And I went to a few focus groups, and I was like, Oh, this is interesting. You know, you can talk to consumers and you can get answers from them, and it seemed a lot more in line with, like, like, my engineering kind of background, especially the quantitative work. And so I applied for a job within at&t. This was back in the mid 90s, and, you know, got into marketing research that way, spent a few years there, and around 2000 I joined the Marriott Corporation in Washington, DC, did marketing research with them, working on their full service brands. So like JW, Marriott, Marriott Renaissance, the Ritz Carlton, which was really interesting work. The Hotel Lodging industry is very fascinating. And then my, my wife, got a job in Chicago, so we moved to Chicago, and that's when I landed the job with Allstate.
Chuck Murphy 3:33
So the the that's funny, I actually don't, I don't think I know this part, but how did you the original electrical engineering background to marketing? How was that jump intentional? Or was that something where? Because that's unusual, right, even for like, at&t, how did, how did that first jump into marketing happen from electrical engineer? Yeah,
Speaker 1 3:54
so it was, I got my MBA, and I it's interesting, because I was told you this some point we did my, in my, in my career as an electrical engineer that, like, oh, you know, it was kind of almost like, kind of like the in thing to do to go get an MBA, because, like, your career as an engineer is going to plateau at some point, you know. And like, I was designing circuit boards for big mainframe kind of computers and whatnot. And so, like, you really got to get them, you know, the management experience in order to really, you know, excel in your career. And so, like me, and like, I don't know, hundreds of 1000s of others, people, you know, did the same thing. Like, it's like, oh my god, we got to get our MBA, you know. So, so honestly, it was like a lot of just like following the crowd, and then after I got my MBA, I was like, Okay, well, I mean, I guess I can't really do engineering anymore.
Chuck Murphy 4:50
It was kind of like a career advancement strategy. That was the advice is to get the MBA, and then once you get the MBA, you're kind of like, Oh, my sales or marketing or. Yeah, yeah, exactly, exactly. So when you It's funny, I had this client, this is, this is at least a decade ago, but from this tech company, and we're doing focus groups, and he was, like, a fairly young guy, and he's like, he's like, Hey, can I go in the room with them? And I was like, most of the time that would be kind of a no. Most, you know what I mean, or my job would be no. But he's, he was, he's, like, a pretty he's a good communicator, and, you know, a very comfortable person to talk with. And I was like, Sure, sure, you know. And, and he's, he's like, I really like talking to customers. And he went into the room, and, you know, there was already a moderator in there too. And you know, the moderator looked at me like I was a little crazy at first. A little crazy. At first. I was like, just talk directly to the and actually, it was really great, because he, you know, he understood some of the, you know, whenever you're doing focus groups, the moderator, you know, gets up to speed quickly. They're good at it. But, you know, the client said people tend to have more depth of knowledge about the question, what the question tree might be, but he actually fell in love with it and never went back. He's been in market research ever since. Loves to do that. We'll often let the moderator Do you know the first hour, and then he there. If there's threads he wants to continue pulling on, he actually prefers to just walk into the room. And he's actually really good at it, you know, I've told him, like you know, I've told him, like, you could actually be a moderator. You could do this. It's really if she would. Some people, I think they see it and it's interesting to them. And, you know, for many people, it's not. But I was the same way, like when I first saw it, I was like, Oh, what a fascinating field. Like, you just get to try to figure these problems out directly, yeah. And
Speaker 1 6:38
I think that's exactly it. I think it was the problem solving aspect of it, right? And that's that I think, like, carries over from the engineering mentality of like, it's all about solving problems in engineering. And this, to me, like it was just That's a perfect application of that, and in a business setting, yeah,
Chuck Murphy 6:55
it's funny, people will sometimes ask me, they're like, you know, what do you like about your job? And I was like, I, I said, I, you know, I always say, like, you know, the thing I like about my job is the same thing that people who do crossword puzzles, like, right? Like, some people totally like that and that I am, like, that type of person, but I also totally get that many people don't right. Many people are like, Why would you sit there and just try to guess what word goes in there? But I actually kind of like the, the problem solving aspect of it, the it's always interesting to me, even though I've seen many of these play out, it's still fascinating to me sometimes that you're still surprised. Every once in a while, like, you know what I mean, even if you get pretty good at it, you're like, oh, actually, no, I think this is different. Like, this is a different dynamic here.
Speaker 1 7:36
Yeah, I had a younger of a younger person on my team, he's, like, in his mid 20s and pro like, we made him program his, made him. We asked him to program his first survey survey ever on one, one of our DIY survey platforms. He'd been doing exclusively qualitative research up to that point. And, you know, we're a little hesitant, and I already, like, you know, is he gonna like it? Is he not gonna like it, or whatever? And he came out of it just absolutely thrilled. He was like, it's like putting a puzzle together. Like, yeah? Like, actually, I never really thought of it as, you're right. It's exactly like putting a puzzle together. Like, gotta get the skip logic right, the display logic and all this. And, yeah, what's the satisfaction at the end
Chuck Murphy 8:21
of it? So do you feel like, I mean, it's good that you bring up, like, dia platforms, because do you feel like, after 19 years, does it feel like the same job to you? Or do you feel like it's changed so much because there is so much new technology, or does it still kind of feel the same?
Speaker 1 8:38
I mean, so like, you know, you guys know better than, way better than I do that. Like, the discipline itself hasn't changed much, right? Like, you know, a dispute choice studies still a dispute choice study, and Max diff and so forth. But yeah, I would like the how we do it, like, or the mechanics of how we do it has certainly changed. Or, and, I mean, like, you know, focus groups are now mostly online instead of at a facility. And, you know, there's, there's changes like that, I guess. But I would say that the one thing that has changed the most within, within my job function, is, is the expectations of our partners, our internal partners, and we work with practically every organization within all state, and that's a few dozen, at least. And you know, timelines have changed pretty significantly, like their their patience for, you know, waiting for six weeks, you know, to get a study done is just not there anymore. And so a lot of times, you know, it's just a handful of questions that they want answered. And, you know, they need, they need an answer within, you know, a week at the most. And so, so that, that the evolution of DIY within our organization, you know, is mostly the result of that. It's. Like, okay, let's find some platforms where we can, you know, churn out three to five to eight questions, get the answers to them within a few days, or, you know, a week or so, and and turn it around. So that, that I would that I would say, is the biggest change, like just, just getting the tools that meet the expectations from a timing standpoint for our clients?
Chuck Murphy 10:26
Yeah, I think that's, that's the biggest thing. That's the biggest thing that seems different to me, too, is is getting used to the timing. And there are, it's funny for the client or the, you know, the the stakeholder, there are a lot of benefits, but for like, people on our side, it does make it feel like more of a grind, because it's just so fast, like you're just into the next one, right? I feel like, when I was, like a younger analyst, you know, we oftentimes had a week and a half to four to five weeks sometimes, like, you think about like an old Marriott project, right? When you I've done work with them in the past, you've got to find these business travelers or whatever, and you would spend weeks finding the right people to talk to and ask the right questions. And that period as an analyst was a little bit of a relief, right? You're not thinking too hard. You're still executing, but you're thinking about the analysis. You're planning it. You're, you're kind of getting your head around, like the way you want to think about this and talk about it. And especially as you're seeing data command, you're hearing about it, that that period was a little bit of a recharge your batteries period. And now it's, it's the timelines, I, uh, especially for that, that level of the the business that is right in the analytics is it's fatiguing, I think, to not have those recharge batteries, and it's something we're constantly trying to figure out how to navigate around. Hi Maggie, so, yeah, so that's it. That is a it's a difficult problem to solve, and I don't think there's an easy way to Yeah,
Speaker 1 12:09
yeah, no. I mean, I mean, in some ways, we're having fun with it too, right? It's like we were able to turn some things around very quickly. Our clients are, are generally very happy, or even sometimes quite surprised by that. So there's a surprise. And built the light kind of thing going on there in terms of, like, our ability to turn things around. I mean, of course, the flip side of that is that, like, now we've set expectations that, like, we can do a lot of that work within just a few days. And, you know, we're so we're, it's constantly, you know, we're, we're always explaining to our internal partners the pros and cons of doing the DIY approach versus, you know, something that should take, you know, four or five weeks to do, yeah,
Chuck Murphy 12:53
and that is, it's funny, because that actually leads into one of the first questions I wanted to talk to you about very well, because there's, there's been this change, not just in research, but in marketing. Right of things can be done very quickly. Questions can be answered very quickly, but also campaigns can be rolled out very quickly. Feedback on campaigns can be rolled out extremely quickly. And you know, this it all, I think, obviously started with social and performance marketing and this digital feedback. But I mentioned this to you earlier, but there was, you know, Scott Galloway, who, you know is so good at his job, but also just makes me cringe sometimes that, you know, he likes narratives, but oftentimes doesn't understand the details well enough to understand why the narrative is wrong. But he gave this presentation last year saying the error of brand is over. And, you know, he basically went through just this focus on what I would call, like, more tactical campaigns, right? Because, and tactical campaigns in the performance marketing era have this tangible benefit of like, Hey, I drew, you know, X percentage of people to this, you know, buy now button in the last 24 hours by saying x, right? And so that feedback is a little bit like a drug. It's like, Oh, wow. We see we spent this amount of money, and we see this our way in the last 24 hours. Ergo, let's do it again. And it's in we see companies I do a lot of research with, with tech companies. And in tech companies, the way they think about marketing is very much this way, and it's they don't understand what I would call, like, you know, the older style kind of brand era. But when I watched this guy, Scott, Scott Galloway thing, when I first watched it, it really did make me CRS, like, I can't believe he's saying this. Because he's a business school professor, professor. And I think one of the things I say all the time tech companies is like, it's actually quite easy to put an ROI on Brandon and to prove that it works. So you can't ignore it. It may. It's totally fine to have these performance campaigns, but you also need to have a brand if you if you want to long term. Hand out. So I wanted to get your reaction to that, to see if, because it did occur to me that Allstate is actually an excellent brand marketer and continues to go down that path very aggressively, as do you know. So I just, I thought you might, yeah,
Speaker 1 15:13
no, I completely agree. I mean, at least for our category, and you know, maybe, maybe Scott's right about other categories, but our brand is very, very important to us. I mean, we invest, you know, over a billion dollars in marketing, and you know, some years close to 2 billion, as do our competitors. You know, progressive state farm and Geico, our main competitors, we all spend tons of money, you know, trying to stand out in the enormous amount of noise on TV, especially in insurance advertising. So it's extremely important to us. I mean, you know, we obviously do a ton of brand tracking internally, trying to understand, like, how our brand distinguishes itself from from the competition. And, you know, we have, we have a legacy and a history, right? I mean, we've been around for, oh gosh, maybe 90 years, maybe or more, and we've, we've developed a reputation, along with State Farm, I would say, as being a very high quality insurance provider. And, you know, that's a tough thing to do in a very commoditized space, right? I mean, there's 1000s of companies out there, but we spend a lot of money trying to tell customers that we we are better than them, like some companies might not cover, you know, some of these, these types of claims or accidents, but we have a history, a proven track record of doing that. And so, yeah, it's, it is super important to us. We, you know, we do qualitative groups all the time where people, a lot of people will say, you know, gosh, if I could afford Allstate, or if I get, you know, have Allstate for, you know, my insurance, I would absolutely get it. But for, you know, for a variety of reasons, maybe it's price, maybe it's just they have, you know, driving history or whatnot, they can't get it. But it is, in at least in the insurance category, a very aspirational kind of brand. So, so yeah, it's super important. The other thing I would say is that, like, you know, one of the reasons we invest a lot in brand is that, like, we also have a lot of different brands within the all safe family. So there's all state that, like, you know, we have a company called National general, we have another company called direct auto, which offer different types of insurance for different types of drivers or different types of homeowners and whatnot. And so, you know, we're, we're right now in this, in this phase where we're trying to figure out, like, what's the best way to market those sub brands. And, you know, how much do we want them to lean into, you know, the high quality, premium, you know, kind of thing that Allstate offers versus, you know, sort of being a little bit more mainstream.
Chuck Murphy 18:16
Yeah, yeah. And that introduces a quick, tricky issues as well, with, like, the multi brand strategy and how you kind of, how you kind of position and how much the spend is on each one. Yeah. And by the way, you know what? It just kind of made me laugh. But that Scott, when that Scott Galloway quote occurred to me, I went two weekends ago to a Colorado football game. I had never, I live in Los Angeles, and I had never been to see Colorado and and I noticed that the, you know, the net behind the field goal. I was like, How long has that been there, that all stage? And I was like, trying. I was like, I remember that net in and when I was a kid, I was like, You know what? They have been doing this for a long, long time, yeah, you know that just over time that the benefit of that compounds, right? It's just, it's just subconscious in me at this point that I know what that means and and that is such a good visual representation, or of the of the brand, right, the hand thing, and catch the football like it's a beautiful little thing that is amazing. And that's when that Galloway quote occurred to me, that which I have a lot of respect for things like that, where people kind of, because I do think that that's how brands are established, right? Just like you get this repetitive, right? That is not, you don't even have to, like, sit there and think about it. I mean, obviously I'm weird, and I sit there and think about it, but most people don't, but they're still seeing that like with every kick that goes through, and it compounds over time, right? And it's just kind of built in there.
Speaker 1 19:47
Yeah, that's been a very, very successful program for us, our partnership with the NCAA. I would say, gosh, maybe at least 10 years, maybe even longer. I. And we've been doing that. And, yeah, it's become so ubiquitous that, like practically every major college football game, you know, you're going to see that those nets, other outdoor advertising events at the stadiums and whatnot. So yeah, that's been a great, a great brand partnership for
Chuck Murphy 20:21
us. Have you seen in any and this doesn't necessarily have to be related to your current tile state, or it could be outside too. But have you seen people push back on that spend in different areas or question the logic of that spend?
Speaker 1 20:39
I would say initially, yes. I mean, like, you know, it those kinds of things, as you know, are very hard to measure, right? Like, you, you buy, you know, these properties, or these sponsorships and, and then, you know, well, how many, how many people bought insurance as a result of of that, right? It's, yeah, it's hard. So it's like a very, you know, at least initially it was a very touchy feely, like, you know, you could only sort of, like, describe the benefits qualitatively. But we have put some programs, a lot of programs into place in terms of, you know, just the awareness among college football fans and sports fans in general about the different properties and whatnot. And we've definitely seen a huge uptick in, in, you know, our presence with, with those consumers. So, so, yeah, I mean, there's always, there's always going to be sort of a more sort of qualitative aspect to all of it. But I think, I think we feel good, and we feel very invested in that, and I think it'll go on for a long time into the future.
Chuck Murphy 21:47
Yeah, yeah. It's funny, because, you know that in that Scott galley talk, he talked about brand marketing as a tax on the company, right? Which kind of killed me, because I was like, you know, the other side of that equation is, like, the price sensitivity. You could model price sensitivity pretty easily, right? And you could see, I could put a shoe in front of you next to the same shoe with an air jordan logo on it, and I know you'll pay more for the Air Jordan logo. It's been proven over and over again, so it's like, there it's not a tax in the sense that it's an investment to me, and it's, you know, we're at the right level of investment is could be hard, because it does take compounding over time to you can't exactly point back to where that willingness to pay came from. It could have been from five years ago. It could have been from last year. So what that is, the biggest difference between performance marketing is there's still some of an art to it, although we have, we have tried to come up with very creative ways to measure that. Maggie's done a lot. I was just
Maggie Bright 22:39
gonna say we did a study years ago where we we looked at sideline like, what's the value of a brand on a sideline? And one of the things that we learned is that you can't always count, you know, put a direct ROI with it being there, but there's a huge deficit to the brand if you take it away from being there. So once it's been there, what if you remove it? Not only are you, you know, you sort of, you lose the value that you've built up over time, but someone else immediately swoops in and they get, they get instant value from from taking over that space that you left. So I think there's a not only is there sort of like you're building the brand, but you also are defending against other people taking that space and building up the brand. And I think, you know, I think sometimes we underestimate the value of heuristics too, and sort of like, hey, you've, you've spent years creating equities that go along with your logo or your brand, and now people can use that every time they see it. It's a shortcut to, hey, I know this is high quality. I know this is trust. I know this is value. And I think that if brands don't take the time to build up that sort of heuristic shortcut to what they mean, then it's a lot harder to build longevity for the brand. And I've been thinking about that a lot, as we did some groups the other day where it was we couldn't figure out why this one symbol was working so well, and it basically the symbol was just a heuristic for all these emotions that are impossible to articulate. And the figure we were talking about at that point was like, I was a mother figure in an ad, and it was like a mother figure can mean so much to so many people, but it's like this really nice, positive shortcut to like, empathy and trust and great attributes. So I think I would strongly disagree that the era of branding is dead. I think we just don't see it as it's just not as it's not as easy to see exactly where the benefits come in, but they're there always.
Chuck Murphy 24:30
We know, I think it was like David acre, one of those brand guys that wrote books, had this, his whole chapter on the you know, for many, many years, when I was younger, you know, Michelin had the baby in the tire ads, right? That this thing, they did it for something like 20 years, like, did it for a long, long time, and at some point that, I think it was in the mid 90s, they, um, they said, you know, we've done this, we've established it, we're going to stop. And I think less than a year later, another tire company basically copied the exact same campaign. They throw a baby in the tire and you. Could see the sales numbers like it was, it was at a very effective like brand imagery, and there, as soon as they stopped it, it did die, unfortunately. Like is, there is those things have to be maintained to a certain level, which is why, you know, companies like Nike and Disney are still throwing, you know, this, those super effective, awesome campaigns out there, but I love that David acre store. I know they teach that in like, business school now, I think it's David acre, but yeah, they show that. They show the sales just flip, and the safety perceptions, which is obviously very important tires.
Maggie Bright 25:34
I think consumers still like the entertainment value of a branding campaign too, right? And I know that that's hard to put an ROI on as well, but it's really hard. It's so much harder to break through now than it ever has been in terms of media. So if you don't have something that, if you're just pushing a call to action or a price point, or, you know, something that's very commoditized, I just don't think it's compelling. I think, you know, it obviously can work in the right moments, but if you're not telling some sort of a story about who and what your brand is, then I don't think you have you just don't have enough legs to stick around for a long time. Now, I know there are examples that would contradict what I just said, but I do. I think consumers want to want they expect more from brands, and part of that is a story, whether it's an origin story or a reason to be around or, you know, kind of your trajectory story. I'm not, you know, it doesn't matter always what the plot of the story is, but they want to be entertained and they want a story. And that comes back to the branding of a company or an idea, by
Chuck Murphy 26:35
the way, can I? Can I switch from a topic that's as old as time to a topic that is really very new. Like, I know we, we were, we were talking a couple months ago, but I love to get your thoughts on AI, kind of, in general, and maybe we'll start there, actually, because I know we had talked a little bit about synthetic data as a particular example. But have you, guy, are you interested? Are you playing with any AI tools? Have you? Have you found, is there anything on your radar that you're kind of like, Hey, this is gonna, this is gonna definitely change the way I work.
Speaker 1 27:10
Oh, definitely. I mean, we're super interested in it. And we have played around with with, you know, things like chat, GPT and and llama and Gemini, and trying, trying to find, like, certain applications for it. So, so we've been doing a few little experiments in house, just for fun the a couple weeks ago, so we have this bike sharing program here in Chicago, like all a lot of big cities do, it's called divvy it's run by the lift Corporation, but it's owned by the City of Chicago. And, you know, just like most bike sharing programs, you can just, you know, scan a code or use an app and, you know, get a bike to ride around the city for a few minutes or even a couple hours. And this company, they post for everyone the ride level data onto their website. So every month, you know, they give you an Excel file like, you know, like the million rides that were taken in Chicago on the divvy system. And like, from what station to this, you know, from the beginning station to the end station, you know, what was the longitude and latitude of the beginning of the end, end of the ride? How long was the ride, and things like that. And so, just for fun, we're like, okay, let's, let's have chat. GPT, just do an analysis of of this file. And, I mean, it was just mind blowing in terms of, you know, how accurate it was, how good it was. It sort of parsing insights from this. It was just raw data. It was like I didn't even really have to tell, tell, chat GBT, you know, like, what to really do with the file. I just gave it a little bit of a prompt in terms of, like, I mean, this is what the different columns represent, and the in the file and whatnot. And, you know, it plotted out line charts. It's like, okay, this is, this is the demand for bikes during the day during the 24 hours of the day. You see this big peak at five o'clock in the afternoon, or, like, you know, what's what? How many casual riders, or versus members of divvy take bikes during certain days of the week? And there were, like, about a dozen of these kinds of questions that we threw at it, and we checked it all by hand, and it was, it was spot on, like it was incredible just how accurate everything was. Now, I know, you know, there have been instances where, like, you know, the large language models do make mistakes, but, but this one was, was totally spot on about a week ago, just for the fun of it. We. Tried out some of the large language models on a on a site called hugging face. So people who are in the machine learning space will will know, will know all about hugging face. It's named after the hugging face emoji. And basically it's like a repository for all these machine learning models that that everybody can access. And you know, you basically use the API to access these models. And we were interested in playing around with some of the natural language processing models, and so we threw a bunch of qualitative transcripts that a few of the models in hugging face, just to score sentiment, right? Like, it's like, we have a whole, like, 300 open end comments from the survey. Let's, let's have these models score the sentiment, and it went through. And then, you know, like, one of the models was like, you know, scoring each one positive, neutral or negative. Positive, neutral or negative. And, you know, went through all 300 comments and scored each one as positive, neutral or negative. And even that was, like, incredibly, incredibly accurate, I would say, like, you know, based on our sort of eyeballing of the comments and how the models evaluated them. So, so, yeah, we're doing little experiments here and there with it, and we can see what the applications will be. I wouldn't say that we've actually started using them on a on any kind of scale basis, but we can see that it will be part of our future. Yeah,
Chuck Murphy 31:31
I think there's no question, right, that it's coming. And it's you definitely could see, you know, that the Salesforce guy, Mark mania flows, is talking about the future as agents, and you can see, you could see where that's going. And I've already, you know, I've, we've played around with a little bit similar to you, where it's still just kind of experimental, and we're not really putting any, you know, any sensitive client data in anything. But, yeah, that's, yeah, it is. It is a little and that part makes me very nervous. Yeah, it makes us nervous too. Yeah, these things tend to have a memory, and they seem to leak information even when told not to, and they're clearly trying to learn. But it's kind of crazy, if you see, there's so many demos now, and there's so many things out there, and this idea of like a synthetic moderator, interviewing a synthetic participant, and it's a wild, wild world, and, you know, the synthetic respondents is something that makes me very nervous long term, but I could see short term why this is going to be used, and people are going to like this idea, you know, kind of like we were talking about earlier, like that this, the speed of it is pretty astounding that you could go out and talk to these synthetic people overnight and get a report that will be, you know, using the 8020 rule. It'll be directionally accurate, right? But it's a little bit like the performance marketing brand thing, in my mind, that thing, in my mind, that where the seductiveness of the timing is gonna keep allowing people to push off the big questions, and at the end of the day, there's no way a synthetic audience could see something new coming, right? You know, it's, it is or respond in a novel way to something novel, right? They're just going to repeat. So it's a very tricky little thing to navigate. I think, long term, like, I
Speaker 1 33:36
agree completely. I mean, I think that honestly, is our biggest what you just said, Chuck, is there is our biggest concern about the whole synthetic respondents thing. And, like, I think we're at a point where it's like, we probably believe that synthetic respondents will get us to the right answer in air quotes. Yeah, on. You know, a lot of, I mean, these models are being trained on, you know, you know, terabytes of data and, you know, I think we're confident that they can spit out the right answer for pretty much anything that we throw at them. What we're worried about is, kind of, what you just described, is like for new ideas, or for new concepts, like, where is the ability to sort of think out of the box, or to, you know, come up with an insight that, you know, we've all been in focus groups where, you know, somebody said, Well, I wouldn't, you know, I wouldn't go for that idea, but if you changed it to do this, this and This, then I totally go for it. Or, you know, like, something that we we'd never thought of before. So, like, it's kind of that regression to the mean kind of thing. It's like, you know, or all of our synthetic respondents just going to regress to whatever you know, their demographic would normally say in a particular situation. And so, like, even on a quantitative basis, that's right, okay, like a bunch of 18 to 24 year olds, you know, would say x, and the synthetic 18 to 24 year olds are going to be probably like the rest of the 18 to 24 year old so it's like, where is the breakthrough, other than, like you said, cost and timing. Yeah,
Chuck Murphy 35:22
so it's all going to be cost of timing. And it's, it is interesting, because you years, oh, sorry, wait, Maggie, you're on mute.
Maggie Bright 35:28
I'm having, I'm having a day. I said, I can see this world in which everyone's using the exact same synthetic data set, right? And so everyone's getting the exact same answers. And so every there's no sort of product innovation or idea innovation, and it's just like, this sea of sameness. And, you know, it's funny,
Chuck Murphy 35:46
there's, there's already, like, I think I, you know, I it's got to be less than a year since the first time I saw someone say you should use synthetic data. So it's still very new. But it actually is funny to watch the pitches and the thinking about it evolved very quickly. And like, I've already seen people be like, hey, our synthetic data set was we individually trained on, you know, 4000 different people, or what, you know, and you're like, it's gonna be, this is really gonna be wild to navigate because, because it's, you know, it is probably very good in a lot of situations, right? And it's going to be fast and cheap. Obviously, it's, I mean, it seems like the cost of this is going to get down to near zero. It's more it's going to be in a rounding error for big companies to do this stuff. So why shouldn't you do it for everything? But it reminds me a little bit of like when the tech companies figured out that they could just AB test, you know, campaigns, you get this dynamic where it's like, yeah, you're you're incrementally improving this idea. But it's very different than a smart marketer sitting in a back room listening to people talk and noodling ideas and then having a breakthrough, right? Like, and we even see that with the online qual like, you don't get the same kind of rate of breakthrough moments. You don't quite have the same engagement from the marketers as you do when they're there in person. And so then they don't have the kind of new ideas you get a little like, there's this dynamic that happens a little bit when we it's like, we emphasize convenience a little too much, when some of the better ideas do come from, from a little bit of tension, or a little bit of, you know what I mean, really sitting and and trying to solve it. Yeah, I don't know. I don't know. It's the other thing it reminded me of is when, when, when, you know, over the last 15, you know, 15 or 20 years ago, you didn't have very many companies that had these massive databases of their customers, right? Like some of them did. If they did have them, it was very difficult. Then all of a sudden, everyone in the world could, you could grab, you know, 1000 customers in a couple days, and ask them a question. So then it becomes very seductive to just always use the customer database for everything. But it's like, hey, look, how are you acquired new customers if we don't go the hard way sometimes, and like, if it I feel like that's going to be the thing with synthetic data a little bit too. It's going to be so seductive to just do this real quick, do it real cheap with these people, that the you know,
Maggie Bright 38:13
I often think too that as there's so many things that you have to think about when you're commissioning a study and fielding a study and all of that. And I think there's, there is sometimes a blind trust of what the sample is and what's in it. And so a lot of times, people aren't really examining, like, well, you know, do I even have synthetic sample in here? What's the percentage of it versus actual respondents? And I, to me, there's a big educational curve that has to come along with it too to explain exactly what that means. And, you know, Is there good synthetic data and bad synthetic data, just like there's good, you know, panels and not so good panels. And I think that that conversation hasn't, at least, I haven't heard a lot of that conversation at a larger scale, with the people who are making the decision to buy a lot of the sample. And so I think that's something that's that's dramatically missing from this conversation is like, what's actually in this data set and how can you control it, or when is the appropriate time to use it versus not? And, you know, I think the the quality of sample is something that is going to get increasingly hard to monitor and control. And so it's this adds another layer of complexity to that that maybe we aren't talking enough about as as researchers. So I'll be interested to see how sort of that conversation evolves. And I don't know Rakesh, if you have had a lot of people coming to you and you know, wanting to have that conversation, or if it's even being talked about at all.
Speaker 1 39:39
No, we, yeah, we, we, we're interested in it as a research team, right? Like we're just, just curious about the whole thing and how it all works. So, you know, we might, we might try a couple of little experiments in 2025 with it. But no, no, we haven't had any internal partners, you know, kind. To us and ask about it, or, you know, or I'm sure a lot of them aren't even aware that it's really a thing. So, so we're not being pushed in that direction, for sure. Now, having said that, like, I do think there's going to be a whole as to your point, Maggie, like, there's almost like an internal sales process that we have to go through, right? Like, like, we have a hard enough time convincing our internal partners when we use real respondents, right? Sometimes, like, this is what the research said. So, like, even just the education on, you know, synthetic data and synthetic respondents. There's a huge, there's a huge education that's going to have to happen within our company for that. So, so, yeah, it's a huge for us. It's a huge wait and see. It's like, you know, and we're approaching it with a healthy degree of skepticism, but we do want to try it and just sort of see what, you know, what it's all about. I mean, we've been, and we've been pitched several times, right there. A lot of companies, you know, like you guys, have been pitched as well. I'm sure have come to us with, you know, synthetic respondent, kinds of solutions. And, you know, some of them sound a little bit more promising than others. I think all of them still, like none of them, have given us a satisfactory answer on the, you know, the questions we've been like playing around with here in terms of, like, innovation and new ideas and whatnot. So, so we'll see how it evolves. But, yeah, I mean, I would say too, that a lot of these companies, to your point, Maggie, like, they, they tell us that they've signed a lot, I mean, they throw out a lot of, you know, big names out there, like a lot of Fortune 500 kinds of companies. So yeah, there are definitely a lot of companies getting into this sooner rather than later. And I think more our attitude has been like, let's, let's, let's let others sort of kick the tires on this first and then, and then we'll, we'll try it out.
Chuck Murphy 42:08
Take the apple approach, let the market defend itself. I do think, by the way that, you know, they always say that they've signed those companies, I think a lot of people are just playing with it, like, right? You know what? I mean, it's still in the like, let's, let's try this. Let's see it. Because the demos, I mean, it is, it's cool. It's really cool. Like, I tell you, what sold me on it is that, you know, this is from, like, a year, last summer, or whatever, but that, that fake Eminem song, like, where I was like, Wow, this actually sounds like an Eminem song. Like, I was like, okay, there is, there's something here. And, you know, in pricing research, if you think about higher global days, that's that essentially is using, in a much smaller version, synthetic people, too. And so for things like that, where there's a gazillion combinations, it does, you could see how it could help, which is interesting. I know we're, I know we've, we've taken a lot of your time. I wanted to ask you one more question here before we let you go, but let's go back, back over your career. But do you have like, a, like, a favorite study, or a study that surprised you, or a story you tell, like young researchers when you're starting, like a success thing in your career that you like?
Speaker 1 43:18
So, so we so we do have a lot of young people on our team, like, about a year and a half ago, or maybe even close to two years now. We merge our UX research and marketing research team teams, and as a result, we have a lot of people on our team who are, like, in their mid 20s, just, you know, a few years out of college, getting started with their careers. And I'll say, like, it's exciting to me to see how many of them are excited about being in research now, and also, like, excited about being at all state like, I mean, you know when I, you know when I was their age, you know, I didn't think about, you know, working at an insurance company, for sure, but a lot of them are really excited to be with the company and with all with, and, you know, being in marketing research at all state. So, like, so I don't often refer to old studies or, like, you know, that kind of thing for For Inspiration. But, you know, I a lot of it is just like having fun with the new stuff, like the like, the chat GPT and and having fun with them, with it right, like, and I think I enjoy staying ahead of the trying to stay ahead of the curve and on a lot of these developments, just for their sake, right? And just to, like, help them along. And you know, when we did that little divvy experiment that I was telling you about, the ride sharing data, and just we did it live, you know, on on, on chat, GBT, it's like, Okay, let's see if it can do this. And let's see if it. Do that and like, and I could hear the excitement of their voices like, as they were throwing out ideas for like, let's try, let's try to make it. Do this. So that that to me. I mean, that's motivating for me, of course, but I think it's also really motivating for them to see like, the possibilities. And I think that's sort of the key for me in this job today is just keeping them motivated. And it's just a really exciting time to be in this field right now.
Chuck Murphy 45:31
It is an exciting time. It's fun to watch, because there's no question it's going to, it's going to help in a lot of ways, right? It's, it's going to help, like, marketers, for sure, right? Like, there's going to be a lot more. We're going to get marketing a lot more heavily over the next decade. As it's hard to imagine that we could get marketed to more, because I feel like it's gone up so last decade, but I you could definitely see how it's going to make our jobs a lot more efficient, too. And to think things like charting right, like we spent so many hours, yes, right, and pulling data, checking data, and you can immediately see how they're not far away from shaving a lot of time off those processes, which, you know, those things are hard. They could be hard on people, right? Like they could be, they could be, it could be tiring to, you know, I know, like when I had, like, a big report to you as an analyst, you spend a whole week like, lining PowerPoint and wording it correctly, and it's going to get a lot more efficient, which is exciting. If it
Maggie Bright 46:26
becomes more efficient to the point that we are allowed to sort of unlock our brains to be more creative, that's the ultimate win, right? Like, take away all of my menial tasks and let me, you know, spend time being creative and thinking about innovations and promise, right? That's right, like, that would be fantastic. I would love that. I would do with every part of my house. I would be like, take it over. Ai, just let me spend time being creative. I
Chuck Murphy 46:49
mean, I know we already use the focus group analogy. Is like, you know, it's like, oh, you could stay at home and watch focus groups over zoom. But the seductiveness of that promise is that you could check out a little bit mentally. That's, that's going to be the that's going to be the tension, right? Is, are you going to let this chat GPC write your deck?
Maggie Bright 47:07
I think too, if you, if you read a whole bunch of AI written things, they do start to sound the same. So like you sort of you, you lose this authenticity and uniqueness that I think is very valuable in what I hope is very valuable in what we do have you
Chuck Murphy 47:19
guys, Have either of you seen that Ben Affleck thing from, I think it was like a week or two ago when they asked him in the interview what AI is going to do for screenwriters. Have
Speaker 1 47:29
you seen this? I heard about it, but I did. I did not actually, it's really,
Chuck Murphy 47:33
it's actually a really great little it's kind of funny, because Ben Affleck, I mean, on one hand, he's an actor, but he is really, really smart. And, you know, he he gets a lot of credit for correctly predicting streaming 20 years ago. He has this speech when he's being interviewed, when he basically explains what, what turns out to be Netflix, very accurately, and but he gets this really interesting speech where he's like, look, it's gonna, you know, in Hollywood, you have like, these, the high level writers that are doing story and the big stuff, and then you have a lot of low level writers that are filling in gaps, providing details, providing color, you know, pulling threads, whatever. And he's like, he gives a really interesting analogy. He's like, it's gonna be so good at that. And he's like, but he's like, we're 30 years away from from Ai being able to tell a story and cut something from a script that's necessary and edit. And I know is, he's like, it'll just, he's like, a computer is so far away from understanding how people will react to something. And I do think that there's going to be that. That's the worry with some of this. Is, like, I could quickly imagine 120 slide deck with, like, all this detail, but the stakeholders, like, what are you giving me? Like, it's going to be too easy to produce so much. But the hard part is still honing in on like, hey, look, this is the problem. The question we need to answer right here. Like, and do you have the answer? What if you don't have the answer? That's actually good. Let's, let's sit with this and let some of that tension build that forces us to really think deeply about it.
Speaker 1 48:59
Yeah, I think that's exactly why we've been sort of like, let's not jump into, you know, the synthetic respondents and all that kind of stuff to we have a lot of, you know, obviously, I said we had a lot of young people on our team, but we also have a lot of season not a lot, at least a handful of seasoned researchers on our team who are, you know, that's their bread and butter. Is telling a story and, and, you know, not doing it in 100 slides, but you know, in 15 to 20 slides and, and, you know, really making sure that each insight is impactful as it can be. And you're right. I don't think, you know, I worry about that too, Chuck, like, it's just gonna spit out a lot of charts, but, like, not much of a story. And so we're still like, like I said, we have a healthy skepticism toward all this.
Chuck Murphy 49:52
Yeah, yeah. All right. Well, I know we've taken a lot of your time, Murphy, I really appreciate you joining us and being so patient with with all. Questions and topics.
Speaker 1 50:01
Yeah, this was a lot of fun. Thank you for having me. Yeah, it really
Chuck Murphy 50:04
was fun. It was great to connect with you in this way.