AI Is Changing Research. But Can We Still Trust the Data?

Check out the full episode below! Enjoy The Exchange? Don't forget to tune in live Friday at 12 pm EST on the Greenbook LinkedIn and Youtube Channel!

AI funding just topped $400 billion. A research giant went private at a huge premium. And fake polling data fooled the media. What do these have in common? In episode 144 of The Exchange, Lenny Murphy and Sarah Snudden dig into who and what we can actually trust in an AI-saturated research landscape, and what it takes to stay ahead of it.

Thanks to our producer, Karley Dartouzos. 

Use code EXCHANGE to get a 15% discount on your general admission IIEX tickets!

IIEX West 

IIEX.AI 

Stay Ahead of the Curve! Subscribe to The Exchange Newsletter on LinkedIn Today! 

Transcript

Lenny Murphy: And we're live. Happy Friday, everybody. And today, we knew Karen wasn't going to be here, but we were going to hold it in reserve on who our surprise guest is, one of our favorites, returning once again, Sarah Snutton. Sarah, welcome.

Sarah Snudden: Thank you. Tournament of Champions, admission accepted.

Lenny Murphy: That's right. Oh, seriously, welcome back. And for those who don't know you, once you. Quick intro so they can level set.

Sarah Snudden: Yeah. I'm an insights enthusiast. Mostly I've spent my time on the client side, but a bit of supplier side in between. And oh gosh, I've worked innovating products and growing brands at Clorox, seventh generation, Keurig, Dr. Pepper, and JDE Peats. And now they're merging, which as that was going down, I'm a bit of a free-range character right now, hoping to find my next place to land. But I feel like the market's picking up, and we're going to cover some of that in our next topic. So, life is good. And I'm currently with you from Wisconsin, my home state, doing the end of summer hurrah finale. And we'll go out through Minneapolis tomorrow and not tomorrow, but soon. It's all ending really soon. The back-to-school reality is really real right now.

Lenny Murphy: Yeah, but right through with you. And for those who know, we have to tell the story, right? Since you are our mutual hometown. That's like, well, yeah, home for a while for me, but that was the point. There's some point in the early 80s that Sarah's from Eau Claire. I lived in Eau Claire. We were both probably hanging out at the skating rink on a Friday night. London Hill Skateland. Yeah. Yep. But we did not know each other. But it's just a small world. So.

Sarah Snudden: I think we were at opposite ends of that snack bar. I would slow roll my raspberry slushy because not the best skater, but I fought darn hard to get to the skate lane. That was where the action was.

Lenny Murphy: It was the only place where there was action in middle school in O'Claire, Wisconsin in their early years. So I actually, I don't know that I. Even, I spent my time at the library in a skating rink, and that was the library still doing really well.

Sarah Snudden: That's a great library. It is a great library.

Lenny Murphy: The Oakla, Wisconsin had a great library, had a great library in the 80s, has a great library now. Anyway, we digress, but let's dive in because there's a lot to cover. Before we do that, a quick plug, Karley, if you will bring up Grit, the new GRIT survey, the latest wave, the business innovation wave. This is the one with the GRIT 50. It is out. So please participate in GRIT. Let us help you. Help us help you by participating. There's a bunch of, just as we know well here, the pace of change is unrelenting. There was a time where, you know, it was fairly static in the industry. We are not. At that place right now. I know we just did Grit just a few months ago. Things are different. They've already changed. This is how we measure it. There's additional questions we're asking this wave around MCP adoption and some of the stuff that we're going to talk about today. So please take a few minutes and help us help you participate in Grit.

Sarah Snudden: Lenny, as a lifelong member of the marketing family, granted, insights is kind of the red-headed stepchild sometimes in the marketing world. I wonder if Grit might need a mascot. Could we get a cameo appearance from Gritty? Then, you know, maybe, maybe, in an off-season kind of way. He's got also kind of the red hair, orange hair vibe.

Lenny Murphy: So, you know. Maybe it was kind of like Bucky's, right? Can we do Bucky? Uh, you know, uh, they probably get sued for beaver, but we could do something, yeah.

Sarah Snudden: They've been on a little suing streak for anyone who uses anything close to a beaver, so I would, I would, but I feel like gritty might be available, great, yeah, okay, they do so well with grit, so yeah, maybe, maybe the uh. Would you wear a gritty costume at IIEX if uh if I mean if it if uh wasn't gonna get sued I I would I would I'm gonna bring it he brings it see this is it.

Lenny Murphy: That's why I know you've always focused on innovation, being one of your key areas. Guys, we're seeing it in real time.

Sarah Snudden: Yeah, there's that part of innovation where no idea is a bad idea, and then there's the real world.

Lenny Murphy: That's right. That's right. That's right. We're going to throw it in against the wall stage right now. Right.

Sarah Snudden: So attention, getting attention and awareness, though. I think you could do that.

Lenny Murphy: Yeah. Yep. Yep. All right, we got a lot to cover. Let's dive in and uh see where we get to. Um, so uh, as always, we follow the money, so the funding MA, the big story this week was sent, uh, you know, going from public to private. Um, you know, so look. They were at a billion-dollar valuation at one point when they were a public, multi-billion dollar valuation. They didn't sell it. And I think a lot of folks kind of look at that and go, not my perspective, at least. I want to get yours here. During a period of transformation, public companies, it's really hard to transform, right? When there are systemic issues that were affecting the industry and they were affecting SIMP because they were such a significant supplier. Going private was the way to relieve the pressure so they could reposition themselves for success. And I have 100% certainty that they will do that. It was interesting that they kept leaning into trusted human intelligence as in their positioning through this, which I thought was just really indicative of where. We're moving from sample to data and human intelligence and doing that at scale. So that's really what was more of a focus on getting away from just the survey as the driver, but being a system to drive human intelligence. What was your take, especially more from the buyer's side?

Sarah Snudden: My take on the words is more that they focused on the true benefit versus the functional execution, because trusted human intelligence is in fact the benefit of what you would do a survey for. But I also think it gives them room to maneuver and also distinguish themselves as like the source that connects back to the people. I think the positioning was smart, especially given all the turbulence and waves of change. I also have lived through one, I've lived through multiple mergers where you're at a company and another one's got to come in and your lanes have to come together. At Keurig, there was a moment where, you know, there'd been fast growth and a lot of changes, and JAB Holdings took the company private. And I actually experienced that as a really good thing. It was a chance to kind of the pressure of Wall Street coming at you from all sides. Um, yeah, it was always a lot. And there was a phase where I felt like Keurig was one of those child stars that you're just waiting to go to rehab. Um, you know, she's big, she's dynamic, she sometimes overcovers little Lindsay Lohan.

Lenny Murphy: Is that the yeah, yeah, I mean.

Sarah Snudden: But then you kind of get to do the nanny 911, behind the scenes or do a little rehabbing or get the pipes, you know, cleaned out or whatever it is, And I experienced that as a real positive in the phase where it happened. And I wish the same for Sint, for sure. They're such an important part of our industry.

Lenny Murphy: Yes, agreed. And I love those analogies, Sarah. Yeah, so it is a strong market signal about what's being valued from a pure financial standpoint, right? They did get a premium, uh, 42.5% on its share price, um, which is great. That share price was low, um, compared to where it had been, but it was enough for obviously for investors to say, hey, this is still a good asset, and I'm aware. Of a few things that are not necessarily public of other companies that are, I thought, were potentially getting poised for sale because their private equity owners were probably not thrilled. But instead, they're doubling down, which I think is so this, along with that stuff that's still kind of behind the scenes, I just happen to be privy to. The signal is being sent out to access human data. Is worth investing in. And you know, and to your point, now it's around the mechanistic components of repositioning that for growth. So I think it's a very bullish signal, even though I'm sure that there's some investors who are not thrilled with that. But it's a strong signal of confidence in these assets. Being repositioned for the market. So, oh, it's my take. Then you go to the other side.

Sarah Snudden: Struggles are real and consistent. So, Sint isn't unique, but it's very big. And there's a few of the very big ones that they're kind of foundational cornerstones to a lot of our industries. So I'm hopeful. I think it's smart. Positioning and the rest remains to be seen.

Lenny Murphy: Yeah. Yeah. It's all about execution now. So we'll see. Now, this pitch book, AI VC funding has already lapped all of 2025 as we're heading into the last quarter or so of 2026. H1 2026 AI venture funding, 407 billion across 3,500 deals versus only 264.1 billion for all of 2025. Follow the money, right? I mean, there's your double down, money. There's the double down. But it will, because what do they need? They need data. Yeah. All of these things are going to require it. Maybe not all coming from panel companies, et cetera, et cetera, but increasingly that human-validated, calibrated, correct, non-AI generated recursive data. That's the deal. So I see that as one, anybody who is still on the fence on AI, guys. That fence is crumbling. I don't know if there's any fence left. You know what I mean? You know, the money flowing from a technology standard. Yeah, that's right. That's right. It's like the ship sailed a long time ago. Follow the money and think about what it is, you know, that is the new reality. Where do we fit? So, but what's your take? Right?

Sarah Snudden: Yeah, I mean, like, connecting the dots. Connecting the dots on the, you know, there's the survey part, like asking people what they think. But I think that I'm really excited to see how behavioral data is operationalized. And a lot of times behavioral data might point to a little different answer than what people say. And that's the classic say-do gap. And I think being able to understand that from a much more behaviorally integrated angle, is for me really interesting. And there's a few thoughts I have on some of the topics we'll cover next that will kind of circle back to that reality and how you connect and close that side gap that I think is interesting. I also had a really great chat with Smriti Gupta at Quirks. And it's interesting, the people who are using AI in a real ground, a strong ground game. Way. Like her point to me was: if you're using AI in that way, you don't necessarily need a gajillion-dollar valuation because if you're using the AI in a very strategic, focused way, it's not like you're paying a lot of humans. You've got a few humans who are, leading it in direction. And I thought that was such a nice counterbalance because I do feel like, you know, having lived through the dot-com bubble and in a way, the SAS wave had a lot of you know that same like all in, everybody's got to sass it up. Um, I feel like remembering that ground game that you don't have to be a gajillion. There's kind of that boy kind of my money's bigger than your money thing that I think you can get caught up in. And there's a part of me that feels like really smart innovation is also going to come from the people who are just like, like, have a strong vision and work a strong ground game, and maybe don't have the most funding, but are thinking really well about how they execute or integrate or do a really focused play. So, I want to kind of shout out those players as well, because I think that that world is like a whole new set of potential. And you don't have to get kind of bogged down and be like, oh, my funding's not as big as somebody's, but it doesn't matter if you're playing smart.

Lenny Murphy: Yeah. Agreed, although I have to push back a little bit on that, solely from the kind of the competitive framework standpoint of these companies that have they achieve that level of funding, they definitely have that there are many perils that come with that, you know, overcapitalized, but being able to just. Feed on the street, you know, you know, marketing budgets, et cetera, et cetera. There are advantages. It's the rabbit and the hare, or sorry, the hare and the turtle. Yeah. Type of model.

Sarah Snudden: So I guess, you know, I would sort of say it's like it raises all boats, but like, not everyone has to be a big boat, is maybe my Wisconsin cabin lakeside analogy. You know, and so it's, it's more of a like, you know. Everyone can play, but get in the water.

Lenny Murphy: So it's not the size of the boat, but the captain, yeah, something like that. Yeah, yeah.

Sarah Snudden: The taller you are, the harder you fall. I don't know. Probably shouldn't go to Odysseus, honey.

Lenny Murphy: We probably don't. Probably not. So, all right. This Vibe IQ raised 22.5 million for its apparel decision layer. And that's what. Caught my eye on this. And this was just the same idea, you know, of decision intelligence as infrastructure, consumer-facing, but focused on clothing and apparel. I just thought that was pretty cool. What did you think?

Sarah Snudden: I mean, that's one of the things. One of the hardest things to do sometimes if you're shopping is to visualize the possible. A lot of people have more of a concrete, sequential take on things. You and I probably are not those people, but I feel like that ability to see apparel or see the goods more easily, visualizing your style is kind of one of those things that's hard to quantify. So the more you can kind of bring it to life, the better, I think.

Lenny Murphy: Yeah, I do have a hard time relating to that.

Sarah Snudden: Might some people have big ideas, and some people like to build from the ground up.

Lenny Murphy: That's true, that's true for all the people, but that's true. Uh, the but that you know, as you look at that step back, it's 20 worth five million, and as a data asset, I mean, increasingly, that's what I kind of think of that. It is. Building a really interesting kind of flywheel model for hey, consumers, we're going to do this cool thing for you, but what they really want is the data and how that compounds in value, etc., etc. And they're going to see more and more things like that that really are fundamentally data places.

Sarah Snudden: So that's true. And I also think, like, if you're getting at the intelligence more in context, I think about Orchard, one of your great winners, that I feel like, that, what you do on the internet with your eyes half closed at night when you're scrolling, scrolling, and you're like, all of a sudden, you buy something and you're like, my waking brain is like, what was that about? You know, the rational, you know, like getting at sort of that emotional layer and building that data out in a way that helps make it more predictable. That's really interesting.

Lenny Murphy: Yeah. Yeah. All right. Let's talk about data quality, fraud, and trust. You're in Wisconsin, right? So, guys, if you hadn't heard, there's this whole thing that came out, a company called Median Strategies that was publishing all of these polls. The Wisconsin gubernatorial primary, also the California primary. And those things picked, they got picked up and they were being used, especially in Wisconsin. Then, you know, oh, yeah, this one candidate ahead, you know. 20 points or whatever. And it was all bullshit. It was made up. It was a social experiment, as they would like to say, or as they said. And man, I just saw that. And my, my heart just sunk, right? It's like there's enough problems with polling to begin with, of bullshit biased polls, but for somebody to just totally, absolutely make it up. But propagate it out there to see the impact of BS polls in affecting real-world performance. Oh, boy, that was what scared me. That was slippery slope stuff.

Sarah Snudden: So, I don't know. Says the guy who had Cambridge Analytica headlining, although I learned a lot from it, to be fair. There's a lot of slippery slopes for sure. And I will say, driving around Wisconsin and you know, I've got a 15-year-old son who's on YouTube, and we watch stuff together, and like all the ads on YouTube being, you know, different candidates. And like, the minute you turn on a radio station, you've got you know, two sides, and and they're also like each casting the like this one's gonna be all the big data centers over the state, and the other side's like, No, this one. And that's what you pick up if you're on the ground here. And I talk to my family and friends about it, and they're like, You be grateful you don't have this all the time because it's it's it to live in it is it's just constant noise politically. So, I have you know, being from Wisconsin, you've got people on all sides of the red to blue spectrum that you know and care about, and it's it's um. It's interesting to try to put all the pieces together. But Wisconsin's definitely one to watch because I feel like it's been so on the line, and how those pieces come together say a lot about the bigger picture. I think it's a microcosm I care about not just personally, but intellectually.

Lenny Murphy: Yeah, get all that. I think that when I saw this, set aside the political aspects of of that uh was the the increasing awareness of the uh the media infoscape fragmentation uh and the and the lack of sophistication in just data literacy unfortunately um uh and how things just spread virally that are just I got an email yesterday for, because I get all types of stuff for PR and everything of a poll, political poll, and we won't get into all the topics. But all I do is look at it and say, God, this is just crap. I know these results are absolutely crap. So there's no validity to them. It's totally biased, blah, blah, blah. But how many millions of people got this email? Yeah. Right. And how many others they're targeting media outlets, right? And how many are going to pick this up and propagate it and put it out there? Oh, when just like this, I think it was based on a real poll, as bad as the poll was, but solely to drive the marketing outcome. Which, and you're right. That's why I came to Analytica to speak at IEX, you know, back in 2016. It was an incredibly successful marketing plan. It just was. So you may not like the results, but it worked. And that's the fine line, I guess, here that is still trying to unpack. Like in the world of marketing, you know, yeah, you're, you're, you, you do things because you want people to sell more stuff. When it comes to political polling, even though it is fundamentally a marketing issue, it feels ickier. But to know something was totally made up, it's like it wasn't even biased, it was just made up.

Sarah Snudden: I actually am getting out of you here because I actually think the reason it happened was just a different form of Cambridge Analytica. And I see it less about marketing success and more about wake-up calls to all of us to get more dialed into how things get shaped in the broader world or how influence occurs. My take might be that it was less about. Maybe again. This hot take, and I'm not the expert here, but there were enough people seeing that rise that were thinking bigger picture about what happens next when it's possibly more of the classical Democrat versus Republican. And you've got the DSR kind of in the other slot, and how Wisconsin is so on the razor's edge that maybe they were trying to get that vote. Count up more from the people who are more the vast middle-ish part. You know, I don't know. I, it's hard to know who I don't know the people behind it. But when I look at the impact and when I talk to people on the ground that, you know, saw that and got out to vote, maybe in a primary more, although Wisconsin does get out to vote increasingly because of that razor's edge. I don't know the answer, but to me, I see a wake-up call, and I'm not sure how. To know how to, you know, sometimes you wake up from a dream and you're like, I don't know what it means and how to fix it, but it's kind of paying attention to me, I guess, if nothing else.

Lenny Murphy: Yeah. Yeah. Well, we'll see. I mean, of course, now we're, we're heading into November and it's these, these whole issues are just going to continue. Um, and a broader conversation about, about shaping, about narrative shaping in the info scape with data. I think that's probably the thing to pay attention to here overall. The Insights Association did come down hard and heavy on this, as they should. Yeah, it'd be interesting. But that dovetails right into, and you and I both are our next topic, AYTM. You worked with AYTM. I worked with AYTM as an advisor. We know they take data quality really seriously. And I thought pretty ballsy that they published their data quality benchmark report, which really was kind of opening the kimono and saying, look, here's what happens on our panel. So I really liked that level of transparency putting it out there.

Sarah Snudden: What'd you think? My take on this was like, this is very on brand for AYTM. I feel like Lev Mazen has been one of the truly authentic drummers of the drum like to pay attention to the data quality. And he and I have had some moments where like, Lev, man, as the client, I want you to be the one caring about the data so that I can like, to me, like, data quality at that point was the monster under the bed. And if I could, like, sleep soundly and know that I was always going to pick that route. Um, but not that we disagreed, but just more, more sort of where I wanted the buck to stop wasn't with me. I think increasingly, we all have to pay more attention. But I say, really, kudos for AYTM for really always being. Consistently about that and starting at a, you know, with roots at a time when no people like me really honestly was like, I don't really want to hear this.

Lenny Murphy: Like, I want to be able to speak that's right, right, right. Right.

Sarah Snudden: Like, you tell me it's good, and I'll believe it's good. And I think, you know, that's one of the nice things about AYTM is they've always had that built in. And as I've gone along and experienced different things. Suppliers and agencies and consultancies. There are those moments, or that I've been increasingly able to do myself and have the experiences where I'm just like, wait, I'm not trying to do any elaborate red-herring fraud detection. But if you tell me you drink coffee a few times a week or more, that you make it at home and that you have a curie, if I give you a well-constructed question with photos of what K-cups could look like, and you can't. Find a kick up. You know, like those moments are the individual wake-up calls where I'm like, God, Lev was right. I didn't, but I didn't really deal with this.

Lenny Murphy: Yeah. Yeah. I think increasingly, and it, well, it goes back to even our previous story. It is probably overly simplistic, but increasingly I see the spectrum of our industry, and really, probably kind of the world, is there's data and there's sense making. Those are the anchor points. Everything in between is a process, right? And that's where AI comes in. It's shrinking the process. But the currency across the board is trust. Trust that the data is good, that's informing the process. So the sense making and trust that the person trying to make sense of it is not full of crap. So the And I think that's to your point, that was always probably kind of under the, almost under the radar, right? It was, it was subtext. It's becoming pure, straight up context now. So it is very front and center of trust, discernment, and critical thinking. I mean, all that type of stuff. It is just increasing in importance and in relevance from a decision standpoint of who you're going to engage with. What partner, what software, what data set, you know, whatever is that trust. It is a defensible moat.

Sarah Snudden: I would also build that point a little bit because we are often the data trust decent making, you know, in the insights market research industry. And I think this is a point we'll circle back to in some of the later points on our agenda today. But I think within the sense making, there's times when you need guidance or a little nudge in a consumer direction, like to be truly consumer-centric. And then there's validation or polling. I would polling, I would almost kind of validation, calibration.

Lenny Murphy: You just need the data. Yep. Yep.

Sarah Snudden: Yeah. And or validation in the sense, like, sometimes, like, if I just need to decide between like you know, attributes, a shape, a color, a claim. I think that there's, you know, guy, you can use, for instance, synthetic in my mind, really well to kind of give the nudge in the right direction, the product or concept guidance that you need.

Lenny Murphy: Should the widget be blue or green, right?

Sarah Snudden: Real-world evaluation. To me, like, you know, that's like where maybe Nielsen Basie's or, you know, would have played. That's kind of a different world. So, when I think about how we weave synthetic things through things, there's a line there that I'm still trying to make sense of within the sense making.

Lenny Murphy: Well, that was a good segue. So, I see what you did there, Sarah. So Qualtrics this week released their synthetic data FAQ. And I really hate them. It was not, we could do anything with synthetic. It was like, nope, nope, there's some real limits. And we've, I, well, I haven't thinked about this. I'm sure that you probably, what you just said, you have been too. Is the uh for again with that spectrum. It's like there's less consequential decisions and more consequential decisions. More consequential decisions, you need to talk to humans, you need to make sure it's real, it's valid, right? The truth, the data, it's valid. You have to go that extra mile and make sure that it's real and true. You know, less consequential decisions, you know, yeah, synthetic absolutely has a role, early stage ideation, you know. Early stage concept testing, those types of things. And they were really clear about that. So I liked, I really liked that they said, do not use this for should we make a widget decision. You can decide whether to make it red or green, but don't use it to decide whether you need to make one at all or what to charge for it.

Sarah Snudden: So I would say. For me on that one, um, I thought, and I had the luck of getting to sit with Jordan Harper at um TMRE last year and have a really good chat about what they were doing on synthetic. Um, I think synthetic's interesting, right? Because you can make it with structured data like survey responses, scaled responses, you know. Or you can do it with unstructured data, like a lot of them are more personal, like that's more fluid, but also has that more emotional richness and context, you know, that comes from having the real large language model kind of oomph behind it. Qualtrics leaned really heavily on the ingredients they had, which was structured data. And I was honestly really like, I kind of had to sit there with him and he was kind to sit with me and kind of answer my questions. But for me, I, boy, if you really know your space well, if you're like living a brand experience and you know your consumers pretty deeply because that ideally is your your job, doing synthetic unstructured data for me feels kind of um almost like again that say do gap danger of like you know because you wouldn't say to someone how much how much do you love me and and lenny's like you know Sarah today on a scale of one to ten right and so there's a part of me when I think about synthetic that I always try to to before I think too deeply say like what are we synthesizing what are the ingredients that we're working with and in part, I learned, you know, after the Cambridge Analytica thing, like, really understanding the richness of unstructured data is a different world. So I thought that what Qualtrics did was probably very smart positioning because they're less, to me, at Read less as like transparency and more about making what they had built make sense to people and not get used incorrectly.

Lenny Murphy: Yep. Yep, yep, absolutely. Yeah, there is a little bit of an element of kind of a little bit of a CYA there. And rightfully so, right? I mean, educate people don't use the product incorrectly. If you're injured doing this wrong, that's on you, right?

Sarah Snudden: Yeah. And I also feel like it illustrates the point of if you're the research buyer in a world where you've got so many things coming at you, even thinking about what are the questions I should be asking about like, yeah, sure, it seems really magical having data really fast, but where does it come from Like, I often use the analogy, even with the unstructured data where you really truly have people pouring in, you know, their feelings and thoughts and telling their stories, which may or may not be fully connected to how they do what they do in the world. But I think about like the sourdough starter, a lot of us got excited about, you know, making sourdough in the pandemic or at some other time, and you forget to feed it, and you come back, and things look funky. You've got to, there's a level of maintenance that you have to do to keep those models truly fresh and on top of it. And, it's easy to get kind of wooed by the glory of the hot bread with the butter on top and the magic of that. But the reality is kind of messy behind the scenes, and you have to keep feeding the starter.

Lenny Murphy: You are the queen of and metaphors, Sarah. You just really are. I just gotta say, I love who else would equate sourdough bread to data quality and usage. And it made perfect sense. I like it. It's a metaphor. What can I say? Oh, all right, audience. This is why I love when Sarah steps in. A lot of fun. Fun. All right, well, that's probably a good segue. Again, the new product launches, quite a few that we're interested in this week. What did you think about the Burke and Verve? Their approach, very synthetic. So it's almost in the same neighborhood, but opposite of Qualtrics. Their approach has been different in building that, but yet highly governed and specific.

Sarah Snudden: Anyway, what did you think? Yeah, look, I think the Burke team has a lot of earnestness, like they roll up their sleeves, and I really like that. So it's almost like when you know a political candidate, you may not agree with every view that they have, but you know they're doing the homework. And so I really like it for that.

Lenny Murphy: Yeah. Yes, 100%. I've had the privilege to get to know the Burke team over the last couple of months in a way that I had never had before. And they are first rate. To your point, right? They're just super smart, earnest, focused, you know, sincere people who want to do really good, solid work for their clients, as many other companies are. I'm not throwing anybody else in the boat. I've just had a chance to get to know them specifically. It's just an example of that.

Sarah Snudden: There's a lot who can take on slippery slopes and I would trust them to build a good set of stairs.

Lenny Murphy: Absolutely. 100%. 100%. So this uniform what jumped out at me was that this was around small language models or domain specific language models and kind of building the systems to do that in a different way. They were taking some research functions and incorporating that in a different way that didn't have much to do with at least traditional research, the way we think about it.

Sarah Snudden: What was your take? Yeah, you know, as someone, again, kind of thanks to IIEX for always kind of keeping me like right at the you know, on the cusp of a lot of waves. I got really into the initial version of large language models back when they were a lot more manual and it was Dakota, or maybe you called them Dakota as an example of when you'd go in and you take all the unstructured data and you would kind of pull it in and you would really look for those emotional bits and you know. To me, I read this thinking about the building at Keurig, where we pulled in all the data from all the land, like all the product reviews, all the new user satisfaction scores, all the CRM data, all the survey open ends, all the qual interviews. When you pull that in, you do get a set of classifiers. At the time, we had about 350 that indicated something really special for our world. So I think creating your own kind of, you know, world within the large language models, there's to me, having built it in the old kind of more pencil and paper and less AI magic way, really, it makes a lot of sense. So I read it from that angle, thinking about like, yes, you absolutely wanna know how the word rich, when it's in a coffee context, what that means to consumers and how you deploy it. If people are, you know, needing something, I, you know, being able to kind of like I loved the idea of the two by two grid where you've got kind of volume of contents versus emotional intensity, and where you find those business levers. And so, I feel like that small language model to me Read, like, oh, this is really like the heart of the matter, you know, yeah, yeah, and I think we're gonna see more and more of that.

Lenny Murphy: The, the, the, uh, the big LLMs, frontier LLMs, they're all at parity now. I actually think intelligence is now an absolute commodity. They're all going to keep getting better and better at capability. Yeah. But and for process, for function, fine for tasks, but for differentiation, for understanding, for depth of understanding, for, you that's, I think that is going to be small language models, the domain-specific language models, increasingly, that's where that stack is going to be. That's where the real action is. Yeah. Yeah.

Sarah Snudden: And I mean, I kind of, my metaphor for this one is like, as a, as a good Scandinavian, thanks, mom, you know, origin. I think about it like boiling the ocean versus having your farmed fish, you know, like you want to keep your circle of salmon in good health and know what's going on in that part of the ocean. And boiling the whole ocean every time is probably not the best option.

Lenny Murphy: I'm glad to use fishing instead of like Viking reaving as the example there.

Sarah Snudden: I didn't want to get going too far, Lenny. You know, there's going to be a fish fry in it.

Lenny Murphy: Yeah. Yes, absolutely. So Walmart Data Ventures, Sintilla. Full disclosure, you they are, I've done some work with the Walmart team, Sintilla team. So I want to get your take, especially as a client. Did you ever use Sintilla or Walmart Data Ventures on the client side?

Sarah Snudden: You know, there were. When we were in the last five years, I've been kind of running an innovation offense, you know, kind of threading from e-com back into brick and mortar more. So that meant that I was kind of out of that loop for a bit. But I think this move is a total no-brainer because if you've got, you know, a great capability, roll it out more broadly. And it makes total sense that they would get it really dialed in the bigger. And then kind of extended out. But yeah, and I think one of the challenges of any of these loyalty card data is you can get really excited about the results, but they can also be, it's like the equivalent in my more recent past is like you can look at your direct consumer data and conclude like, oh God, they're not repeating. And then, you know, you got to go over to Amazon, though, which is a behavior a lot of people, you know, shift to. And when you put those pieces together, you get the fuller view of, oh, they are repeating. It's, you know, you can't get to tempest in the teapot about either, no matter how big your teapot is. But the bigger the view you can get. That's also why I'm excited about, again, any data source that can really. Integrate those pieces of different behaviors in different places.

Lenny Murphy: So, yeah, yeah. Total number. That's an, I guess we did actually say that Walmart has lost Nutella in the Sam's Club. And part of the family. Right, right. They're not sisters or cousins, I guess. But yeah, they have that entire integrated system, they have so much data and the way they've been building that to uh. Talk about a flywheel. I mean, that's a flywheel. So, Walmart's got the flywheel.

Sarah Snudden: So, Walmart was also a thing in flywheel before a lot of people were.

Lenny Murphy: Yes, yep, absolutely. All right. Let me cost because you and I can go on and on. Two, let's run through these things real, real quick. Interesting reading. The MCP march continues. You know, Cape Lena. Now MCP, Getty Images, MCP. Point being, my, my takeaway on that is the disruption to the buying process fundamentally. So the old way of engaging is hell buying anything, but certainly now, you know, services, technology, et cetera, et cetera, it is about where's the control point? Because you don't really have many anymore. So instead, it's going to be about distribution, and the MCP is the distribution taken. Are you? This is one of those topics where we can be over our skis because everybody's saying, Oh, yeah, MCP, MCP, but clients aren't really adopting it yet. And I'm sorry, my dog's as high. What's your take on it? The client organizations of how, what, what, what, what, demand, what need is this to say, yeah, I don't want to have to send you an RFP or go through this or log in. I want to just do it in Slack or whatever, whatever the hell platform that you kind of live in every day.

Sarah Snudden: There's honestly so much churn in the water in this space right now that I don't, I don't have a firm take on this one, honestly. I, I, I'm a little bit like, wait and see. I, you know, it's, a lot is changing really fast. And if you can be convenient and get the job done accurately, effectively, and well, there's going to be people who are excited to do it that way.

Lenny Murphy: Yeah. Yeah. Well, there we go. Well said. The IBM part of OpenAI for their consulting practice. I mean, I'm sure you've been seeing the past few months of every, you know.

Sarah Snudden: The the forward deployed engineers consulting you know like well of course uh so this didn't yeah look IBM has really historically on deep brand equity in a strong ground game like and so the part I kind of liked this this one because to me ai is a little bit like ethereal and you know cloudy and and I think fusing two things together like that. And look, IBM was really thinking about it. I remember like Watson and thinking about those models back.

Lenny Murphy: You IBM hasn't been doing nothing.

Sarah Snudden: They've had a strong force. And I think I like those two pieces put together in that way. It makes me intrigued.

Lenny Murphy: Yeah, I agree. And IBM, we worked with IBM and tried to build Beroglyph with their consulting component. Uh, I mean, they're, they're super smart. Uh, and that consulting component, actually, I think they're probably better at well, not better. I like that they determine, hey, that ship sailed. We kind of missed that one a little bit. Okay, that's fine. Well, I've seen clients figure out how to do this.

Sarah Snudden: Yeah. And I think about it as the metaphor of like the healthy tension where you've got like a product marketing organization or people who are, you know, trying to grow. A brand business like IBM did. And then you've got an ad agency. I think of the way those tensions pull off like how you dream big and how you take people along for the ride.

Lenny Murphy: And I like it from that angle. Yeah. All right. Let's get an interesting Read. Did you Read this Forrester AI disruption model?

Sarah Snudden: Did you get a chance? You know, I mean, Forrester does quadrant maps better than most.

Lenny Murphy: So, but did you see, though, what jumped out at me was infrastructure, data, data, and AI, and security were the only categories that they considered not disruptive or prone to disruption. And God, for this. Forrester went through and said which categories and business are prone for disruption. They kind of scored them and did a quadrant map. But it was reassuring to go back to our previous conversation that, like, oh, data, the need for data, that's actually not disruptive. That's fundamental. That's going to continue.

Sarah Snudden: Yeah, I look, I also think there's the reality within data where some of it needs to be disrupted. There's numbers and columns that, you know, could be more nimble into turning into insights. So, I think part of the good of a quadrant map is you can always have a good argument about the what's and the whys, but probably good on them for, again, staying true to their equity in their four boxes. 

Lenny Murphy: Yeah. That's right. That's right. But yes, true to that. But I was glad to see somebody tackling this, you know, and they make as much sense as anybody of like, all right, what's because I've been thinking about that a lot with all the new entrants coming into the space that are all, well, one, they're clustered around specific issues and it's just really crowded. But the other is, but where's the defensible moat on some of these? And I'm not sure where that is because of the ongoing disruption. They're disruptive players who actually probably have a pretty limited shelf life before they're disrupted in general. So just what an odd time.

Sarah Snudden: It is. It is. There's a lot of splashing in the water, but yeah, a lot of sizzle is the worst of the steak or the worst if you're going full.

Lenny Murphy: Yeah. Did you do that, the AI competence judgment cap? Did you go through that exercise? I did not. Okay. Well, here's the gist. It was pretty cool. The, it was a model put together that on, um, basically, does your use of AI make you stupid, or does it make you smart? That was the essence of the essence. And there's a lot of just the basic kind of outsourcing your thinking that makes you stupid.

Sarah Snudden: You don't use it, you lose it.

Lenny Murphy: I'm not the higher my odometer miles go, so yes, yes, but the collaborative thought partner, you know, red teaming, stress testing, synthesizing, pushing back and forth, iterative process that is actually what they found. And this is through Brigham Young University that both the AI benefits. But the human benefits as well. And that's what I've been telling myself: like, no, I'm getting smarter using these tools. I'm not getting dumber. My kids disagree. They look and say, Dad, you're, you're, yeah, that's not right. But hey, Brigham University says, no, depending on how you use these tools is how it really does become a a net positive and augmentation. And I just thought that was just really interesting.

Sarah Snudden: Yeah, I always try to maintain the practice of saying the Zen breath makes it sound like I've got some really deep religious philosophy behind it, but like the look before you leap or like the assess before you run in. And I think, my take on this one was like, makes sense, you know, we'll keep an eye on it. I think, in a world, and it's interesting having a teenager because I feel like they're tuned into watching for AI in a different way, just because it's so like they're getting all their tuners tuned, you know, they're kind of their brains are still forming, and it's interesting. I've had a few times where someone's like, you know, that's an AI picture, or that's a whatever. And I think all of us have to kind of develop our sort of, you know, spidey sense.

Lenny Murphy: That's men. Yeah.

Sarah Snudden: Your surface validity, as we'd call it, you know, in the market research trade, of just, but remembering to kind of always keep that on in a way.

Lenny Murphy: Yeah. Yeah. I agree heartily. The last thing, just because we mentioned that there, that, uh, you know, DeepSeek, the open source model, they're version 4. I'd say it's important to keep recognizing that OpenAI and Anthropic and Google, and you know, they suck all the oxygen out of the room. They are not the only players, and the open source models attain the same level of capability as the Frontier models within days. And the economics of that increasingly are going to be important. So we all may have our personal favorites from a marketing standpoint or whatever that we adopted. And I'm lazy. I don't want to switch to stuff from our user experience perspective. Right. But from a scale standpoint in the future, I think the open source models will increasingly make a big impact. And so we need to be reminded of that because they do. I mean, they're cheaper, they're faster, and they're as good, if not, you know, if not better. And we just need to recognize that. Yeah, games have a lot of players.

Sarah Snudden: And I think in our industry, it's especially, you know, again, the real power of insight often comes from connecting different pieces. So the more you, going back to our Cintilla example, like the more you can. Pull in all the purchases from all the land and all the cloud and all the brick and mortar, like all having more gives you the fullest picture. And open source does that in a different way than the frontier, you know, branded bordered models. So it'll be interesting to watch for sure.

Lenny Murphy: It will. Yep. Odd times. So we have officially, this is now officially our longest. Oh, wow. Yeah. That's all right. Cause we it's good stuff. I think. Hopefully, listeners, you'll let us know. You take the ride. Yeah. That's all right. Don't get to Wisconsin Redheads together. Anything else that we want to cover?

Sarah Snudden: I think it's good. I'm excited for IAEX West when it comes around.

Lenny Murphy: So that's right.

Sarah Snudden: And yeah.

Lenny Murphy: Yep, absolutely. Absolutely. Well, Sarah, thank you for stepping in. It's such a joy having you. Hopefully, our listeners enjoyed it as well. We'll be back next week, Karen and I both again. But Sarah's on, she's on speed dial. So.

Sarah Snudden: Hit me on LinkedIn if you need to find me.

Lenny Murphy: No, this is great. Thanks so much for the opportunity.

Sarah Snudden: And thanks, Karen. I hope your vacation is going well.

Lenny Murphy: Yes. Have a great weekend. Enjoy Wisconsin time, the Renfest, and back to school time as well.

Sarah Snudden: Yep. It's coming. All right.

Lenny Murphy: Bye. Bye, everybody. Take care.

Links from the episode:

Cint goes private after steep valuation drop 

AI Report: $407 Billion Raised as Megadeals Dominate 

VibeIQ Raises $22.5 Million for the Apparel Decision Layer 

Wisconsin Polling Firm Admits Its Results Were Fabricated 

aytm Publishes First Quarterly Data Quality Benchmark Report 

Qualtrics: Synthetic data for market research: Your questions, answered 

Burke and Verve Team for Synthetic Customer Solution 

Uniphore Launches Marketing AI 

Walmart Brings Scintilla to Sam's Club in 2027 

Caplena Adds MCP Integration 

Getty Images Launches an MCP Server 

IBM Partners With OpenAI to Build a Dedicated Consulting Practice 

Forrester Introduces an AI Disruption Model 

The AI Competence/Judgment Gap 

DeepSeek Ships V4 Pro, Ending Its Preview Window