Pod Insights Summation with Auren Hoffman Morning Consult CEO Michael Ramlet - AI, consumer sentiment and reinventing survey research About this episode Michael Ramlet is the co-founder and CEO of Morning Consult, a global decision intelligence company with more than $100 million in annual revenue. Morning Consult conducts 30,000 daily interviews across 40+ countries and serves more than half of the Fortune 500. In this episode of World of DaaS, Michael and Auren discuss: The evolution from telephone to online polling Building representative polling samples in the modern era Real-time data collection and economic indicators AI ...
Transcript Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.
0:02 Auren Hoffman? Hello fellow data nerds. Welcome to World of DAS. I'm your host, Orin Hoffman, CEO of Incubate and GP of Flex Capital. Discover more episodes, get weekly data as a service news, original content, articles on data and more at worldofdas.com. That's world of das, D-A-A-S, dot com. Hello fellow data nerds. My guest today is Michael Ramlet. Michael is the co-founder and CEO of Morning Consult, a global survey, research and decision intelligence company. He started the company in 2014. It has grown to over 100 million in revenue and Morning Consult now conducts over 30,000 interviews every day across 40 countries. Michael, welcome to World of DAS.
0:46 Speaker 01 Thank you for having me.
0:47 Auren Hoffman? I'm very excited. Now, why did traditional polling just break down?
0:51 Speaker 01 I think you have to think about polling more broadly in a historical context. So it's gone through mode shifts in the past. You go back to the Roman census and I guess all the way to today's census, they're still knocking on doors and doing live in-person interviews. The next big mode shift was to mail-based surveys and then you get a mode shift to telephone. And then you've now got this most recent shift online and essentially from desktop to mobile. And so I think you think about each one of those components and parts of it are different. Just think about from the labor cost standpoint, the cost of a survey, if you're doing live telephone or in person is really the cost of the person's time conducting the interview. That's very different in a mobile and online world when there's not a live human being conducting it. And so I think what it takes to be good
1:36 Speaker 01 at any one of those modes is very different. And most of the legacy players started in a live telephone or a live in-person world.
1:45 Auren Hoffman? Now, when I was in high school, and this is hundreds of years ago, I used to fill out these surveys and then they would send me 10 bucks for filling out the survey and I loved it. I was just constantly filling out surveys and I just tried to make as much money as possible doing it. But also it was fun to fill out the surveys and stuff. And that was in the mail, physical mail I would send it back.
2:04 Speaker 01 Yeah, think about it this way. It's a real problem of the time value of money if you're doing it to make a living. And we ask a lot of psychographics about the motivations of taking surveys. And I think the two things that stand out are, there's a lot of folks that don't necessarily have a job or a role in their family in which they make a lot of decisions. So I think they find it empowering to share their opinion. And I think, secondly, not every job is a highly engaging intellectual exercise. And so I think other folks find it intellectually stimulating. And I think those tend to be more so the case, especially among older respondents or among respondents who are of higher incomes.
2:35 Auren Hoffman? Well, people like answering questions about themselves.
2:38 Speaker 01 Yeah.
2:38 Auren Hoffman? They love that. I mean, because you felt like those personality tests online are super popular, like, what kind of cat are you or something? So people love that.
2:47 Speaker 01 But it speaks to how hard it is to go find a representative sample. And that's been the big failure of the legacy players is that they just didn't take the time to understand what it takes to be successful from a supply chain standpoint. It's harder to find a representative sample than randomly calling the phone book. Lot easier to run a polling operation where you just randomly dial the phone book. Much harder to go find representative samples, not just in the US, but certainly abroad, especially in developing markets. And I think that's a big challenge that the industry, legacy players, just didn't care to focus on. They were going through the private equity turnstiles. So the idea of investing in what was a cost center for the industry, I think, held it back. Everything that happened in ad tech over the last 20, 25 years just didn't happen in the survey market research industry.
3:27 Auren Hoffman? And often the N is pretty small. So once you start subdividing the N, even if the N is 1,200 people, well, once you start saying, okay, well, this age group and this gender and this state and this, all of a sudden, the N could be five people or something.
3:42 Speaker 01 That's the problem with live telephone in person. It is extremely expensive. I think one of the things that was our first big unlock inflection moment was when we realized that people were not differentiating by how difficult it was to get that survey respondent. There are more women in the Western Census region between 18 and 29 willing to take a survey and make it between 5000 K. Then there are over 65 men without a high school degree living in the Midwest. You should pay more for those men and less for those women. And that was the first big unlock for us. Essentially, at our core, we're like a commodities trading hedge fund. We're better at acquiring a higher quality interview. But because you can drive down the unit costs through those bidding algorithms, now all of a sudden you can do these much larger sample sizes. So to your point before, the big frustration point was, once you got to a second demographic or a sub-geography,
4:26 Speaker 01 you had very small sample sizes. And that was always the Achilles heel of these large telephone operations. You just couldn't get enough respondents. The other part is you can't show visuals, you can't do other dynamic tests in an audible telephone call, and it takes longer to actually read the interview than it would be to read the screen.
4:45 Auren Hoffman? There's also a spot poll or spot questionnaire where you want to figure something out, but then sometimes you want to ask the same group of people over time, similar questions and see how those people are evolving as well. And there it's really important. You have as little churn as possible in that poll, I presume.
5:05 Speaker 01 Yeah. And if you think about it from our perspective, some of this, we got lucky. So I think we had good ideas and we thought we were going to work really hard, but right time, right place. We got to start the business in a cloud-based computing world when everyone else had on-premise servers in the industry. I think we looked at it and said, we wanted to go every day because our whole mission was we wanted real-time intelligence on what people think around the world. And from our standpoint, that meant you had to constantly be asking them what they think. And so to your point, a set standard questionnaire that goes every day. So it's 30,000 interviews, no matter what, we're going to go and collect the data. But what became the lucky part of that was because we had guaranteed demand, you could have peaks and troughs for other surveys that are going on. So the supply chain really loved us because we could guarantee an extremely large volume.
5:50 Speaker 01 And in exchange with the bidding algorithms, we were essentially able to achieve most favored nation status. So we drove down the unit cost while actually increasing the volume of surveys that the supply chain could expect. And those things allowed us to get smarter and smarter around what it costs to get what population in what country. And so it became this reinforcing network effect.
6:11 Auren Hoffman? The Bureau of Labor Statistics traditionally called people up or called employers up or called HR departments up.
6:18 Speaker 01 Still do, to be clear.
6:19 Auren Hoffman? I know they still do that. But then there's also real data that they see. I think they get data from ADP, they get some other data in some ways. That's almost like a passive survey. How do you think about that? Because obviously you can get data of like, are you still using Netflix, but you could also get the credit card data as well, which would prove if they're actually getting billed by Netflix or if they're turning from Netflix to Disney Plus or something like that.
6:42 Speaker 01 Yeah, I think it's a combination of both worlds. You're thinking about it before you do it. So the tighter that timeframe, if it's an impulse purchase at the convenience store counter, the transaction data is probably going to be extremely accurate because it happens at the frequency. The even better idea is just the tier to which some of those CPG companies don't have that point of sale data point and have it processed instantaneously. And so I think when we look at it, a good example would be the work we do at the Chicago Fed. For their Carts model, they're pulling in the large data processor, credit card transaction data, and then they're marrying it with the survey-based outlook data on consumer and retail spending. So I think from our standpoint, we really try to benchmark against what would be those real-world outcomes. So number one thing we do is we benchmark, for instance, against equity markets.
7:27 Speaker 01 On a platform called Meet and Century, they'll do all the backtesting for all of our brand indicators. You might be thinking about taking that vacation weeks before you actually make the transaction on Airbnb or another booking platform. And so subsequently, that's a two to three week alpha that you might be able to leverage out of that database and then to be able to benchmark it against the underlying SEC reportable financials. I think we've seen that as the sweet spot of behavioral data married with that real-time sentiment representative survey data.
7:55 Auren Hoffman? In some ways, if you think of data companies, they're often doing one of two things. It seems like you're doing both. One is they're trying desperately to predict the past and really trying to get an accurate position of what happened in the past. And then there's other companies that are trying to predict the future, and they're both actually incredibly difficult to do. So you can have, hey, what do you feel about this? But you could also, hey, are you going to make a purchase in the future or something as well?
8:20 Speaker 01 Well, and that's where it comes back to the frequency part of it. You look at the Michigan Consumer Stim index, it's great that it goes back to the 40s. It's deeply problematic that it's at best 600, 700, 100 people. It takes three weeks to process the data. So by the time they're reporting the data publicly, that was our big unlock at the beginning of the pandemic was we had gone to the Fed and said, hey, We're doing instead of 600 a month, we're doing 6,000 a day in the US. And oh, by the way, we translated 24,000 reviews in those 43 international markets. It's programmatically available on API, 1 a.m. East Coast time. And that was a big demonstration of credibility and capability. What we originally approached the central banks about was are there other indicators? Like inflation at that point, you really hadn't seen any meaningful inflation. So there's a lot of discussion about should the market basket change?
9:05 Speaker 01 And so we actually started working with the central banks around what are new potential indicators? Given the gray economy and the gig economy writ large. And what then happened at the beginning of the pandemic, the feds came back to say, hey, remember that daily data you're talking to us about? Can we get that? And I think to the surprise of all our large banking clients who use the brand tracking capabilities, it was like, wait, you have daily data on consumer sentiment in the US. I think for us, up until that point, you just didn't have the, oh, my gosh, I needed this moment to make a decision right now. And I think that has sped up since the pandemic.
9:38 Auren Hoffman? Yeah, we're involved with so many data companies and a large portion of those sell data to the Fed.
9:43 Speaker 01 Yeah.
9:43 Auren Hoffman? So the Fed is a very big buyer of different types of, so if they find an interesting data set, they're willing to buy it and see if that can be predictive or help them understand a nuanced thing.
9:55 Speaker 01 I think the approach of going on now casting is that right middle ground. The challenge of predicting the future, like so many insert favorite quote. But I think that now casting ability of here's the accurate barometer of higher frequency data at this moment in time. And I think that's what you're seeing, or at least I'm seeing from CEOs, CFOs. They're pulling guidance because the ability to give that confident guidance nine, 12 months out. But they still have to get on the earnings call and talk about what happened in the last quarter and what do they expect in the next quarter? And so that's now casting capability, which to your point, the Fed's been a pioneer in. I think that's where we're seeing some of the most sophisticated activity.
10:29 Auren Hoffman? And if you remember, there was this study of Harvard students that JFK was actually in that study and a lot of famous people. It was a 60 plus year study where they actually followed these students and they would follow with surveys of them and they saw how they changed over time. But of course, it was this very narrow, super. Elite, mostly super wealthy, all men back then study there and they have all these findings from this, but again, so narrow. If you could reimagine in something like that in today's world, what is that? Do we get their medical records? Do we get their bank statements? Do we get their Fitbit and Apple Watch readings on a daily basis?
11:13 Auren Hoffman? How could we reimagine something like that today?
11:16 Speaker 01 I mean, we think a lot about how challenging it is to get a representative sample of a population to get the data off of. And so I think we felt really lucky. We also played with a little bit of house money. We bootstrapped the first five years, so there wasn't much parental supervision. And I think in hindsight, that was to our benefit because the idea of committing to going to do that many interviews on a daily basis, really committed to a high fixed cost. But the reality was if we could sell it and stay ahead of that curve, then we could continue to expand the data set. And so I think from our standpoint, we were excited because the thing we knew you couldn't do is go back in time and collect the data. And the representiveness of what you could do with surveys was the really valuable.
11:53 Auren Hoffman? But with some data, you can. There are some data sets. You actually can go back in time and get payroll data. So there are these data sets which are somewhat substitutes for surveys.
12:04 Speaker 01 But you get really limited demographics off of it. I think that's what we were always excited about. I think when we look at our data set, what makes it different is the breadth of the data. If we think about every row is a person, every column is a question and you cut every column on another column. And so I think from our standpoint, where we look at other numerical data sets, and I think what made it challenging from an engineering standpoint, was just how the breadth of the database is a little bit different than some of the really large numerical data sets.
12:34 Auren Hoffman? If you think of Siraj Chetty, economist at Harvard, he's got access to the IRS data, which is this massive study on over 100 million Americans, their kids, their grandkids, how are they doing? How do people change? How much do they give to charity? How much do they move around? I mean, it's this incredible study that maybe four or five researchers in the world get access to, maybe unfairly. Could you imagine if everyone could get access to that, again, in a somehow anonymous way, it could really blow out the world's knowledge.
13:07 Speaker 01 Yeah, but it's still confined to what they ask the questions on.
13:10 Auren Hoffman? Sure.
13:10 Speaker 01 I think that's always when we look at is this desire to ask anything we possibly could want in the future to try and get even more breadth. So for instance, one of my greatest regrets is I wish we attract the housing market sooner. It just would have been even more compelling to have a larger longitudinal database to it. But I think obviously you have to make gives and takes when you have the capital to make those investments. But I think from our standpoint, the exciting thing is we could keep adding to it. Obviously, the investment in some of these large longitudinal studies at a governmental level has dissipated, or they're locked into a mode. I think that was the challenge for Michigan is just they had always done it telephonically, and they just didn't make the jumps. They made the jump a year ago, and all of a sudden, the trends all break. And it's a lot more
13:54 Speaker 01 challenging for them to find a representative sample online than what they thought was sort of representative sample on telephone. And you see this then in terms of market movements. You had some pretty wild swings in the first three months of this year with that monthly data set. I think that's the innovator's element at core. You're going to have some mode effects. That's what keeps me up at night is, okay, is there a mode effect change that takes place?
14:15 Auren Hoffman? Having representative samples always hurts. Even the S and P 500, those 500 companies are very different than the 500 companies that were there 50 years ago in the same index. And so they have to slowly add, they have to slowly reweight. It's very nuanced, very hard. And of course, everyone hates it. Everyone thinks it's stupid. Everyone argues with it.
14:35 Speaker 01 But everyone wants to be in it for higher trading.
14:38 Auren Hoffman? Yeah, of course, everyone wants to be in it, of course. I think these things are really hard to get any type of sample that people will believe because it's always a religious battle when you're talking about a sample.
14:48 Speaker 01 Yeah, I think if you had a synopsis of what went wrong in the survey market industry, because it's a way bigger industry than most people realize. Andreessen wrote a note a couple weeks ago that pegs at 140 billion. The number I tend to look at, at least the quantity of size, around 80, 90 billion. So there's plenty of money being spent in category. It is extremely fragmented. So I think that's one of the things that oftentimes surprises me.
15:08 Auren Hoffman? A survey could just be one expert, too, sometimes.
15:11 Speaker 01 A hundred percent. And also, there wasn't economies of scale around the tech stack. And so I think if you looked at what was the problem in the industry, it was too slow in terms of the frequency and processing of the data. It wasn't getting the large enough sample sizes to do that second, third level common to real look. And then the third was just, it was too expensive. It just was prohibitively expensive to go and collect that data. And I think that's what's exciting now about where the industry is a lot of those things, I think, have been unlocked. You still have the legacy pieces of it. The big question will be for survey research, what is the next evolution to look like in terms of not just traditionally survey data in the context of political polling, but in the context of economic data, there's a lot of pressure, especially in even developed economies at this point, to find representative samples.
15:56 Speaker 01 And then frankly, even more so the case in market research at corporate capacity.
16:00 Auren Hoffman? The market itself, in some ways, is a massive survey where people actually put their dollars. The market about the Apple stock price is actually a very efficient market that's out there. And then, of course, now we have these prediction markets where people are trying to opine on tons of different things. And those tend to be really interesting and quite predictive as well. How do you think about the markets in general?
16:25 Speaker 01 I think at our core, we want to be a leading indicator of any market. So regardless of whether it's betting markets on a particular event, or if it's equity markets, or if it's currency markets, because obviously tracking a lot of the underlying economic sentiment pieces globally plays into the currency FX play. But I think from our standpoint-
16:42 Auren Hoffman? Sure. Not necessarily from your standpoint, but how do you think about, just in general, these prediction markets? What do you like about these markets?
16:48 Speaker 01 I think making markets creates transparency, and I think that's a good thing. So I don't think there's any downside. I think the question is, what's informing that market activity? What you saw even political prediction markets is that surveys were being used throughout that process.
17:01 Auren Hoffman? Of course.
17:02 Speaker 01 The whale on that larger Harris Trump bet was conducting his own private surveys. So I think the degree to which it's going to inform markets, I think our view is that the more market making that's going on, the more opportunities there are to leverage the data. So I think we're pretty pro of the development of it.
17:17 Auren Hoffman? Some of these markets, a lot of Americans can't bet on them, et cetera. So there's less liquidity than there could be in these. And maybe it's less accurate because of that, or I don't know. How do you think about that?
17:28 Speaker 01 I think it's the breadth of things that you'll want to track go beyond what the markets cover. So they tend to be very event-driven markets.
17:36 Auren Hoffman? Yeah, well, they tend to be things you can easily adjudicate. So it has to be, will the Fed increase 0.25% or not by this date?
17:45 Speaker 01 Yeah, will there be X amount of rainfall?
17:47 Auren Hoffman? Yeah, yeah. Then it's easy. If there's a nuance, it's hard to bet on it.
17:50 Speaker 01 Totally. But that, I think, is why I don't view them as replacing. I think it's a way to monetize that type of knowledge and to really hone in. And I think there's just such a broader use case of the data. Understanding your customer base isn't going to be derived out of prediction markets or understanding. A good example might be what is the underlying interest among a emerging population in a particular product? One of my favorite, when did we have the early data? And I wish we had our own hedge fund was you could see the effect of the Bad Bunny Post Malone sponsorships on CROX sales six weeks ahead of reported earnings. Just this massive pop in the daily tracking around interest from Gen Z and then from
18:33 Speaker 01 Gen Z's parents. And you could see it in the data. And you could specific when did the Bad Bunny sponsorship drop and when did the Post Malone sponsorship drop? And you're like, man, I wish we could trade on that type of information. So.
18:44 Auren Hoffman? Is CROX a public company? I don't even know.
18:47 Speaker 01 It hasn't been something. We track so much on this. We get so much non-public information. I don't trade, but we have a bunch of buy-side clients.
18:53 Auren Hoffman? You personally don't trade because of the information or?
18:56 Speaker 01 Yeah, we have over a quarter of the global 2000 are clients. And so I think we're constantly exposed to non-public information.
19:02 Auren Hoffman? Got it. So you just find it safer not to trade at all?
19:04 Speaker 01 I think companies should take the same position, but you know, yeah. So I think from our standpoint, we really feel like prediction markets are additive. I think in the same way that we're really excited about being able to benchmark against equity benchmarks, against currency benchmarks, those things create opportunities for real-world comparison and improve the process. The political point is really hard. And I think this is not fully understood. It's not about funding representative sample of the US because there's not compulsory voting. Not everyone votes. And the issue is every election cycle, whether it's a midterm or presidential, has an entirely different participation composition. And so you have to both predict who's likely going to show up, and you then have to go find a representative sample of that prediction. And so I think that's why it's the ultimate test.
19:49 Auren Hoffman? It's like a double problem.
19:51 Speaker 01 It's an incredibly challenging proposition. I think if you're running a business, you're probably not looking to win.
19:58 Auren Hoffman? And also, there are some things that everyone in the country has an opinion on. But basically, after the president, you go down just one more level. A lot of people don't even know who their governor is.
20:07 Speaker 01 Right.
20:07 Auren Hoffman? It's really hard to hold on these lower information types of things.
20:12 Speaker 01 Yeah, I think you've got to think about what is the activity? If you're running a political campaign, you're running all the way through election debt. So you're looking at a lot of the momentum shifts and the tactics, whether it's working or it's not. I think if you're looking at it from a business standpoint, you're not trying to win the product category at 48, 47. Hopefully, you're trying to win it by a much larger margin. And I think what folks misunderstand is that the data gives you a context of where it is today. It doesn't mean that's guaranteed where it will be tomorrow. Obviously, the shorter term prediction, the more accuracy. But the issue is that you can move public opinion. I think that's the thing that differentiates really great executives or really great political leaders. It just gives you a sense of how much you're going to have to educate or persuade and move a larger population. And I think that's the difference in we need leaders and leaders with conviction is,
20:57 Speaker 01 hey, I understand where the data is. This is the right decision to go in that direction. But I recognize in the short term, the onus is on me to provide that education or hold steady for those numbers to materialize over the intermediate long term.
21:10 Auren Hoffman? World of DAS isn't just a podcast. World of DAS membership is a private invite-only community for founders and executives of important data and AI businesses. It's a place for off-the-record, high-trust conversations focused on the biggest questions data leaders are facing. The community is about peer learning, curated intros, and access to private events. Everyone in the group is building or leading something meaningful in data. If that sounds like you, apply to join at worldofdas.com. That's world of D A A S dot com.
21:50 Auren Hoffman? If you look at the last presidential race, Trump-Harris, it did seem like a lot of the Trump ads were about persuading people. They were finding niche audiences, black men or whatever niche they could find. And then even maybe more niche than that, and then trying to persuade them to move. And if they could just get X amount moving, that would lead. And of course, they were only trying to persuade people in the seven or eight states that mattered and stuff where it seemed like the Harris was much more broader, just trying to maybe get people out to vote, or they were advertising in states that they didn't. I know you're not a political person, but how do you break down their strategies?
22:30 Speaker 01 Yeah, I think from our standpoint, obviously make the full API available. So you can take advantage of the six plus million interviews we do in the US. And you can see the dramatic events like Biden's debate or stepping out of debate. You obviously see anything within the margins, within the margin. That's what is incredible about how tight the most recent cycles have been. I think what stands out to me is the way you described it, just the political aspect of spending money in this space is even more aggressive on an even tighter timeline than what large corporations do. They're essentially standing up a full-scale multinational company overnight in three supplies. They're moving a billion dollars. And yes, there are appendages and all the different pieces, but the reality is you're
23:11 Speaker 01 essentially scaffolding on the fly at, outside of OpenAI, very rare levels of that level of growth in terms of the scale of an operation. What we're surprised by, and I think this also surprised many folks in the political spaces, for how advanced the ad tech side of it is, the polling and research side of it, these are predominantly a small group of still a lot of just predominantly men in the DC area that have been doing it for 20, 30 years.
23:37 Auren Hoffman? Yeah, a lot of them are super old.
23:39 Speaker 01 They own their own call center. They'll send you a physical binder. You walk into any major campaign office and there's a wall of them.
23:46 Auren Hoffman? I know a lot of them, and these guys, a bunch of them went to Harvard in the 70s, and they're super smart. They were the first people to think about data in the 70s and stuff like that. They're still the guys who are polling. It's just crazy that it hasn't really evolved.
24:01 Speaker 01 At that point, it's more about you're this trusted strategist than you are a data-driven operation. I think that's where the differentiation really splits between larger corporate applications or financial service applications versus political applications. It gives you a read on that. I think that'll change at some point here. It evolves, I think, when we first started, because we started the company out of DC and a lot of us had different backgrounds. And it's just a joke, the old school pollsters couldn't program anything other than their nest. And I think the degree to which-
24:29 Auren Hoffman? By the way, I don't think that's true. I think a lot of them, actually, they were good programmers for their day. They actually certainly could write good sequel, but they still came out of that 70s. The reason why they were successful, they were the first people to adopt computers. If you read the histories of all these pollsters who started in the 70s, they're amazing.
24:46 Speaker 01 Yeah, Ken ends up running the overnight, like, mayoral races. Totally. But the reality is, it's been the same way for political polling for two or three decades. One of the reasons why we never went into this space, the other is it's just not a particularly good business model. To follow, I think, from our standpoint, what we wanted to do was to benchmark daily data sets against the toughest challenges. And one of which is, how do you track politics? If you look at the modern political polling space today, I think we take a lot of credit for, we forced everyone to have to move to daily. That was never something that was an expectation. All your old school media partnerships, once a month, at best, 600, 800 responses. If you look at it in the last five years, and I think this is starting to change too, it's not just about US politics, and it's not just about US politics in the context
25:31 Speaker 01 of US politics. So for instance, those 24,000 interviews we do a day in 42 international markets, you can see the impact of American politics around the globe. You can see, for instance, geopolitical risk implications. China, in the lead up to the tariff announcements, beginning with the North American tariff announcements, China passes the US and global net favorability for the first time ever in our tracking, presumably for the first time post-World War II. And you see the US dynamic playing out internationally. You obviously had about 75% of the world's population-
26:03 Auren Hoffman? Yeah, which is super worrisome.
26:06 Speaker 01 Right, 100%. In terms of the expectation, I think that general America is like, oh, America has this really legitimate goodwill in foreign markets. That's not the case. The BRICS, obviously, most recently, with their announcement on visas.
26:16 Auren Hoffman? When we think about, let's say, consumer sentiment, how is the tech industry doing in consumer? You hear about things where the consumers are becoming more anti-tech, but then you also hear the companies that are most admired in the world are companies like Apple and Amazon and Google. So where is that going?
26:33 Speaker 01 It's like Congress. People hate Congress. They like their congressperson. Even in this era of unprecedented disruption, still 90 some percent of them get reelected, but Congress approval rate never low. I think the interesting dynamic on this is actually developing.
26:47 Auren Hoffman? And by the way, I don't know that they like their congressman. They just don't like the other person running against their congressman. I don't know if they have any true love for their congressperson.
26:55 Speaker 01 Yeah, I just think the general thing is I hate Congress, but yet the average congressperson gets reelected extraordinarily high clip. I think what's interesting to me is this distinction between established economies and developing economies. And you see a 30-point percentage gap between developing economies and developed countries, over 50% plus net favorability towards AI or trust in technology and AI. And in established economies, you've got 20% and declining.
27:24 Auren Hoffman? So you're saying developing, it's even more pro-AI?
27:28 Speaker 01 Higher level of trust, higher level of.
27:31 Auren Hoffman? Oh, I wouldn't have thought that. I would have thought they would be, in some ways, they might be the first guys on the chopping block who get hit by it.
27:37 Speaker 01 Possibly, but in terms of consumer.
27:40 Auren Hoffman? I think AI is going to hurt India and the Philippines much faster than it'll hurt Germany and America or something.
27:47 Speaker 01 Obviously, it's not always the case that the public thinks something that's necessarily reflected in the numbers.
27:51 Auren Hoffman? Correct.
27:52 Speaker 01 I think from our standpoint, probably the thing that stands out the most in terms of favorability or outlook and general feel towards AI, the second most favorable outlook among the G20 is among China. The second worst outlook is among the US. As you see this very significant.
28:09 Auren Hoffman? Who's the most favorable in the G20?
28:11 Speaker 01 Nigeria.
28:12 Auren Hoffman? Oh, interesting. I would have not thought that.
28:15 Speaker 01 Yeah, it goes Nigeria, China, Argentina, the least favorable, South Africa, US, Canada, UK, and fourth least favorable.
28:22 Auren Hoffman? Interesting. What's your psychological profile of why the US is 19th out of 20 in favorability toward AI? And what's your psychological profile of China, where they're two out of 20?
28:34 Speaker 01 Presumably, some of it's also information dissemination. I don't know what, obviously, in Chinese front of those markets.
28:40 Auren Hoffman? Yeah, AI is a scary boogeyman.
28:43 Speaker 01 Yeah, obviously, too. I think this degree to which a lot of the largest, most developed economies, they have the most to lose in terms of the average economic participation. And so I think that probably has a factor into the fear gauge aspect of it.
28:56 Auren Hoffman? As you're seeing other interesting consumer trends, what other trends would you think are interesting that are happening or things that people were excited about and now they're less excited or vice versa?
29:06 Speaker 01 I think the number one thing is the degree to which consumer sentiment has been fragmented and just the underlying health of the consumer. So in the past, you had those national metrics around here is the US consumer sentiment. And the issue is that it is fractured dramatically around income ones. You see some reporting on that, but it's even more dramatic if you're in the QSR space right now, or you're in casual dining, or you're in fast food. The difference between going to the higher end of that spectrum and down, middle income consumers look much more like lower income consumers. I think you see it by geography, the housing market. That's what I was saying before. My bigger regrets, I wish we had our housing data going back further because you can see that in certain cities, the housing
29:47 Speaker 01 dynamic, I feel a lot wealthier if I'm in a stronger housing market, if I own a home, than necessarily what I may be doing in a market that's in a more significant contraction. So I think it's just the fragmentation of it.
30:00 Auren Hoffman? How does housing play? Because America is a weird country where it's very high percentage homeownership. It's, let's say, the reverse of Germany, which is very low homeownership, even also relatively wealthy country that's out there. I don't know the facts, but maybe US is one of the highest in the world in terms of homeownership.
30:18 Speaker 01 Take a second step onto that dynamic. The US also uses fixed interest rate mortgages. That's not the case in UK.
30:25 Auren Hoffman? That's right. It's very pro consumer to just lock that 30 year mortgage in at that a few percent.
30:31 Speaker 01 And so I think the market's going on, I guess, 18 months of recessionary calls, the bears. If you locked in a Two percent mortgage rate, you're not changing homes because you're not going to get a two percent rate today.
30:43 Auren Hoffman? You're sitting pretty.
30:45 Speaker 01 But you feel pretty good about the rise in home value and that gives you some economic boost or sentiment boost and you're going out and spending.
30:54 Auren Hoffman? Yeah. By the way, I don't even understand if you have a variable mortgage rate, imagine every month, if your rent was changing every month, you would just go crazy. And especially if it doubled over a six month period.
31:05 Speaker 01 It's going to be shorter duration too. So it's an even larger swing.
31:10 Auren Hoffman? Yeah, I don't know how you deal with that. So I understand why so few people own homes because they probably have to put way more in cash because getting a loan is pretty scary if it's going to change all the time. Whereas in the U.S., you only have to put 20% down or something. It's a lot more easy to afford something.
31:26 Speaker 01 So the most sophisticated use case that I've seen here this year is the degree to which you're seeing now those CFOs be able to tease out customer bases. And we're seeing it from sell-side research or we're seeing it from buy-side. It's not like, how is the housing market or home improvement market? The CEO of Home Depot broke this out on their earnings call with our data. And it was, what's specifically the sentiment of the Home Depot customer base? And how is it similar or different to the competitive set? So I think it's leveling the regional or sub-population dynamics.
31:58 Auren Hoffman? And Home Depot might be more concentrated in the southeast of America and it might be more male dominant or, you know, whatever it might be.
32:05 Speaker 01 Yeah, all the different components of it. But the idea is that essentially in the past, you had to go off these national metrics. I think the hardest thing, this is what I'm excited about with the AI applications, is just, it's telling people to ask the question. Because I think one of the hardest things, the technical limitations of what was possible for, meant that you had a closed-minded understanding of what you could ask for. It's like, ah, they told me I couldn't ask for Atlanta-based data because it didn't exist 10, 15 years ago. I think that's the aspect of it that gets problematic, I think, for older executives in particular. And I think it's just the degree to which you have to stay open-minded. I mean, you know, this is across, obviously, the world of DAS. The assets that are coming online are so dramatically expansive compared to what they were even 10, 15 years ago.
32:47 Auren Hoffman? When I talk to a 20 year old, it seems like 100% of them are using some sort of AI tool on a daily basis, they're a daily active user. When I talk to a 50 year old, it's very rare. And when I talk to a 70 year old, it's super rare that they're using any of these tools. Maybe a couple have AI in the background or something like a Duolingo. They're not actually using them as much. Am I right? Is it that bifurcated?
33:12 Speaker 01 How is it recognized that they're using them? Because they might be using Google search.
33:16 Auren Hoffman? That's right. But I'm saying they're not actively thinking they're benefiting from AI, whereas the 20 year old is actively going to perplexity or chat GPT or anthropic or Gemini or something. And they're actively engaging and asking questions.
33:29 Speaker 01 February 2023, we asked US adults. So this doesn't have the age breakout, but the figure of well, just 2% of US adults in February 2023. So they use chat GPT several times a day. As of April of 2025, that number was 14%. So 600% increase in terms of multiple daily use cases in the population. So my guess is that it would follow a similar but accelerated.
33:52 Auren Hoffman? It's very concentrated in the younger people.
33:54 Speaker 01 I would look towards that of our data. So what was the adoption curve for Tik Tok? Or what was the adoption curve for Instagram? Tik Tok's great because I can see it in the full 10 year data set, the event of it. And then what are those tipping point moments where all of a sudden it goes from it's a Gen Z thing to a millennial thing to a boomer?
34:11 Auren Hoffman? Yeah, I just came from a family gathering and I was surprised at how little the adults there were using. I wasn't surprised that the 20 year olds were using it because if you're in college, you have to use it. It's necessary, even in high school, it's necessary to be using these tools. You would be lapped very quickly if you're not using those tools. But I was surprised really that everybody was AI curious. These are smart people, my family, but their usage was quite low.
34:39 Speaker 01 This is something we think is a huge opportunity, especially with older executives or established executives is they might be afraid to send the analyst an email asking for data point. But in the quasi anonymity of AI, you can ask wherever you want. There's no reputational threat.
34:56 Auren Hoffman? Totally. Ask the randomest thing.
34:59 Speaker 01 I think we look at the AI application for our data as democratizing access. Before it was very much analytical functions or its insights, the analytics. They might have the raw API, they might use the SaaS platform. But with the AI tool, I think it opens up the ability for more and more executives to ask questions they might be afraid of asking another team.
35:19 Auren Hoffman? How are you personally, not how is your company, but how are you, Michael Ramlett, using AI personally?
35:24 Speaker 01 I think on a daily basis, the ability to use AI to come up with really compelling things for kids. So I have a six-year-old, a four-year-old and an 18-month-old. So you can come up with obviously all kinds of different bedtime story.
35:35 Auren Hoffman? Stories and stuff like that. And you're actually doing it. I keep hearing about people doing it, but you're actually getting cool stories.
35:42 Speaker 01 That's one aspect of it. The way I still use it the most is for business questions. I have an obsession with all of our code names, our 20th century industrial revolution names, literally the book spot. It's the Titans, the House of Morgan type naming conventions. And so instead of having to think about how to explain why we went with this code name, it's an Operation Bessemer. So asking AI to make up a summary of the Bessemer Press and how it transformed steel and the civilization of America. It's not really my day-to-day job.
36:10 Auren Hoffman? It's kind of fun.
36:11 Speaker 01 Yeah, you come across something and it's just a little bit more compelling to go back through it. So I think on a personal basis, more the history.
36:18 Auren Hoffman? Where in your personally using it, where do you see that official frontier where there's a limitation that you see, but you're confident in the next six or 12 months, you're going to be able to use it for that?
36:29 Speaker 01 I mean, I think about it on a professional side for me personally is just right now, for instance, we have the capability of asking the jobs to be done analysis before I go into any pitch or any client meeting. And it will pull up all of the publicly available information, but then it'll cross it with all of our underlying data access.
36:46 Auren Hoffman? Internal.
36:46 Speaker 01 And so I think what does that mean in terms of the, we're trying to talk about in the context of, don't think about where it's going to be in six, 12, 18 months. You don't know that definitive answer. Think about what would be the best possible end destination and just assume at some point between now and then there's an inflection moment that makes that possible. And so I think when we look at it, a lot of those things require some multi-step pieces to it. The reality is agents should be able to at some point in the not too distant future. The earnings prep pack that we should be able to send to you should be entirely automated from start to finish. And it should look and feel differently. It's not like an automation of just replicating the same report. Every time eight, 10 weeks before your public earnings call, you should be getting a generated earnings prep pack from the platform.
37:29 Speaker 01 And I think those things, all of it gets tighter and it's less of the multi-step piece to it, or less of the human in the loop review component to it. I think that's where I see the biggest change for us, but I also think that's true of most businesses. How many people are in that support function, putting together that earnings prep pack? That all of a sudden now, whether it's operational data, whether it's poke opinion data we're collecting or financial data, all of those things can be organized and synthesized so much faster. The most recent case that saved me a ton of time, we're building out this outreach campaign to equity research analysts. We had all the inbounders sort of thinking, okay, this is generating it. The ability to write essentially a prompt that can pull who the equity research analysts are for any given security.
38:13 Speaker 01 You can go find it, you can dig through different finance platforms, but to get it within 15 seconds is magical.
38:20 Auren Hoffman? Is it a feature or a bug? If you go back 11 years, I think you founded the company in DC. Now you're, I believe you're in Midwest. Is there a feature bug that you weren't in the Bay area, the San Francisco Bay area?
38:32 Speaker 01 Well, I don't want to jump the gun because you guys traditionally ask that, what's the conventional wisdom advice?
38:37 Auren Hoffman? Okay, yeah, well, let's jump the gun, let's do it.
38:39 Speaker 01 Didn't want to make the mold. The conventional wisdom that's wrong is you have to raise from blue chip VC. That's the core conventional wisdom. And I think that it was probably right during a period of time. Obviously, access to capital has changed and evolved, but I mean it less in the sense of it's not the signaling factor of it. Because I think you can now go raise money from a lot of places, more about the alignment aspect of it. And we got really lucky. We thought we were going to raise a round from a fund who one of their partners got indicted the day that the academic cheating scandal broke. It was like the best thing that ever happened to us. We ended up going and raising from three really great, non-traditional blue chip VCs. One was Lupus Systems from James Murdoch. The other was Advanced Venture Partners from the US family.
39:20 Speaker 01 And the third was the fund out of Susquehanna that it's not in one time limited space. And I think our advantage was that we had all the meetings. We bootstrapped for the first five years of the business, but we were out of DC. So that already made us a little weird, especially pandemic. Even though we had the who's who of New York and Sound Road, I think from our standpoint, the biggest thing was this flexibility of longer duration investing. We were not a great three to five year bet because we had to go build the data asset.
39:47 Auren Hoffman? But most ventures is a 10 year plus bet on something usually.
39:52 Speaker 01 Yeah, I just know that in terms of from the where we're based type dynamic, it didn't matter at that point. There was that, at least pre-b瀨 pandemic, much more stronger feeling of need to be in the Bay and those pieces. I think DC was unique for us. I was in DC. My national expansion strategy was to move to Chicago within 15 minutes of my mother in law and 30 minutes to O'Hare. I can make a lot more east and west coast flights. I have a co-founder in New York and a co-founder still in DC. Our headquarters in DC, but this DC DNA was compelling for our field because you had this familiarity with economic data, with political data. You had all of these underlying knowledge bases around survey research. So it was a natural hub and still is the headquarters and probably about 40, 45% of our staff are there. But I think that dynamic, the conventional wisdom being you start a high growth technology
40:36 Speaker 01 startup in the Bay area, you lose money there.
40:38 Auren Hoffman? Yeah. And we also started bootstrapped and then raised money, which is also different. Some people either raise money right away or they'll start bootstrapped and not raise money in the future.
40:48 Speaker 01 Yeah. The thing was that we got to the point where we could collect the data in the US. And then we went through, we thought we were going to do our series A and we found everyone's giving a discount to could you collect this data internationally? Because at that point, you think about Nielsen, it's predominantly US data set. Yes, there are some international pieces to it, but there wasn't a lot of international data sets in the survey research space like that. And so I think one of the things we realized was we ended up benefiting from bootstrapping because I think you had to learn the hard thing about hard things. And I feel like now it's like bootstrapping, you learn the difference between AR and cash really quickly. And I think the degree to which we benefited from is it gave us time to grow up. I was lucky I had an incredible co-founder that had a data science or a research background in Counter-Op and then I had another technical co-founder, more of the computer science background
41:33 Speaker 01 in Alex Doolin. And for those two guys and myself, it gave us time to get your sea legs and grow up a little bit. And I think what the value of the capital did was allow us to say, okay, we're going to start tracking a number of these international markets, even if the client base today is predominantly US centric, it creates the opportunity to sell multinational companies, global assets, and it created the pathways for international expansion down the road because we knew you'd have to collect the data to then be able to go to market with it. Maybe the YC factor is still real. I mean, obviously, valuation pieces are there. Defense tech would be another one. Obviously, DC space has a number of unique attributes to it. DC, I would tell you, is the best.
42:11 Auren Hoffman? Not really. If you think of all the best defense tech companies, I mean, very few of them are based in DC.
42:16 Speaker 01 But they have large DC bases now.
42:18 Auren Hoffman? Of course. Yeah, eventually they have to have their sales team there.
42:21 Speaker 01 It's a great angel investing community because a lot of folks run lobbying and communication. I think for us, we really benefited from the angel stage when we bootstrapped.
42:29 Auren Hoffman? Yep. All right. Since I asked you the last question we ask all of our guests, we ask you now the second to last question we ask all of our guests, which is, what is a conspiracy theory that you believe?
42:39 Speaker 01 Oh, so I grew up in Madison, Wisconsin, a Packers, Milwaukee Brewers, Milwaukee Bucks fan.
42:44 Auren Hoffman? I was just in Wisconsin last week. I love Wisconsin. It's my favorite state, especially in the summer.
42:50 Speaker 01 That's a Packers shareholder certificate up there.
42:52 Auren Hoffman? Ah, okay. Yeah, I have a few friends that are shareholders.
42:56 Speaker 01 It's an invitation to the best shareholder meeting. You get drinking tickets to go to the annual meeting at Lambo.
43:02 Auren Hoffman? Amazing.
43:02 Speaker 01 But my conspiracy theory that I do believe, I think is acutely the case in the NBA, is that I think that there is a scripted nature to it. There's certainly a bias towards non-major markets. You look at the finals examples around officiating or the way that players move across the league. I'm all in on a number of the NBA conspiracies.
43:22 Auren Hoffman? Oh wait, sorry. I don't even know these NBA conspiracies. Give me an example of what's going on there.
43:26 Speaker 01 Oh, so it goes back to you being in the 85 draft. Was it orchestrated for the Knicks? There's a lot of conspiracies around the Jordan retirement in Chicago. They covered in The Last Dance. Pretty nauseam. The 2002 Western Conference Finals between the Kings and Lakers, just some very bizarre officiating, unfortunately for Sacramento.
43:47 Auren Hoffman? Oh, guys, so you think the NBA is, they took the scales for the big market teams like the Lakers, the Celtics.
43:54 Speaker 01 I think it's a miracle. Giannis still in one.
43:57 Auren Hoffman? As a lifelong Knicks fan, I don't feel like anyone took the scales for the Knicks, but maybe, maybe you're right.
44:01 Speaker 01 Not by the lack of trying, maybe.
44:04 Auren Hoffman? Yeah, yeah, exactly. They've tried, but the owner keeps messing it up. Okay, yeah, that's fair.
44:09 Speaker 01 I feel like there's a good WWE vibe to some of the underlying elements of NBA.
44:15 Auren Hoffman? Yeah, it does seem like somehow the Lakers and the Celtics always have an amazing team.
44:20 Speaker 01 Yeah, you get a crazy trade out of like Dallas.
44:23 Auren Hoffman? Yeah, that's true. Yeah. How did they get that trade?
44:25 Speaker 01 Surprise. Is everyone in the league?
44:27 Auren Hoffman? I'm with you.
44:28 Speaker 01 I'm not saying I agree with everyone.
44:29 Auren Hoffman? Well, because when you're an owner of a team, it's actually not necessarily in your best interest to have your team do well. When you're an owner in the NBA or owner in the NFL, you're actually really a 1/28th or whatever owner in the NBA. You really care that the NBA does super well. You don't care about your specific team like the Dallas Mavericks.
44:53 Speaker 01 There's a lot of incentives to think that there could be some real merit. You get some referee gambling and sagas.
44:59 Auren Hoffman? Oh, it makes sense. You just want the NBA to get the bigger contract, the bigger TV contract.
45:05 Speaker 01 I like all the games. I think that there's some real merit.
45:08 Auren Hoffman? By the way, I hate the games. They're terrible. I hate watching the games. I don't watch them anymore. They're completely boring. I think they're losing fans. They're just, it's not interesting. I'd love them to completely change the game. So, yeah, that's my whole shtick too. All right. This has been amazing. I love this conversation, the whole conversation we've had. Thank you, Michael Ramlet, for joining us on World of the Desk. By the way, I follow you at Michael Ramlet on X. Definitely encourage our listeners to engage you there.
45:35 Speaker 01 Yeah, you can access all the data at morningconsult.ai. So everything we talked about today, it's all right there.
45:41 Auren Hoffman? Oh, yeah. Awesome. Okay, yeah. Do that too. People could just DM you and ask your AI questions as well. So this has been amazing.
45:48 Speaker 01 Thanks for having me.
45:49 Auren Hoffman? Thanks again for coming on.
45:53 Auren Hoffman? Thanks for listening. If you haven't already done so, please subscribe to World of DAS wherever you consume your podcasts, YouTube, Apple, Spotify, and more. And please help us get discovered by leaving a review. And check out worldofdas.com. That's worldofdas, D-A-A-S, dot com. And of course, connect with me on Twitter, at Auren. That's A-U-R-E-N. Would absolutely love to hear from you.
Previous Atlan CEO Prukalpa Sankar - AI agent workflows and hiring AI native talent Next Dwarkesh Patel - AI Timelines, Productivity & Global Stagnation Page updated November 1, 2026 · transcript and summary are machine-generated