Your AI Is Only as Good as Its Data

Richard Byrd: Alright, welcome to the show everybody. Today we have a special guest. We have Dr. Sukumar from OSG. Welcome to the show, Dr. Sukumar.

Dr. Sukumar: Thank you very much. Thank you for having me here.

Richard Byrd: Yeah, well, I'm happy that you came onto the show. On the Above the Clouds podcast, our first question always refers to birds. So if OSG were a bird, Dr. Sukumar, what kind of bird would it be?

Dr. Sukumar: That's a great question. I'd love to be a combination bird, if there is such a thing, and I'd combine a peacock and an eagle. I'm not sure if you've ever had anyone be a combination, but let me explain why I think it makes sense. As we talk a little later, I'll tell you more about what we do. But if you think about the peacock with its feathers and hundred dots—for us, those dots reflect humans and what people want. And what we do is all about thinking about what consumers and customers want. Then we help companies soar like the eagle. The eagle has a lot of foresight. It thinks about the future. It thinks about where to go next. So if I think about who we are as a company, we really are a combination of those two things. We're very customer-centric, just like the dots on a peacock. And we like to help our customers be confident, be firm, have that foresight, and be able to soar into that growth, into that future, just like the eagle does.

Richard Byrd: I like it. That's a really good one. Nobody said peacock yet, and those are really cool birds. Our office backs up to this wooded space along Buffalo Bayou, and there are a lot of wild peacocks back there—you can hear them. They sound like dinosaurs, they sound crazy back there. One was walking across the street the other day, and I don't think people realize how big those birds are. They're pretty big.

Dr. Sukumar: They're pretty big, and they're beautiful when the weather is one of those where they actually pull out their feathers and display them in their entirety. I think those colors are just a reflection of how we as people are so very different, and the need to understand us is so critical.

Richard Byrd: Yeah, that's great. Tell us a little bit about OSG for the people who don't know about you guys.

Dr. Sukumar: Absolutely. So look, there are three letters here: O, S, and G. As a legal name, we go by Optimal Strategic Group, but the three letters mean something. The first letter, O, is all about identifying opportunities. The second, S, is all about designing strategies. And the third, G, is about delivering growth. When you talk about what we do as a company, we live in a world today with a lot of data. A lot of that data is historical—it's about the past. And what we're looking to do is combine that data from the past with what consumers, customers, patients, people want in the future. So as to help identify new opportunities, new innovations, new products, new solutions for our customers. And with that human being in mind—and with all that data, and so much automation and digital infrastructure—we help design the right strategy to ultimately drive engagement. And customer engagement is so critical to drive growth. That letter G is also about human engagement, patient engagement. When we do work in the healthcare environment, we know we need more patient engagement. We need people engaged in their care, engaged in what they're trying to manage to have a better life. And as a result of that, they grow, and they come out of the challenges they face. So the three letters mean a lot. For us, they become the guiding principles of what we do as a company.

Richard Byrd: Wow, that's excellent. I think you touched on a couple of things there, because a lot of our clients in the marketing world, or in sales consulting, are always looking for growth. But that engagement is also super important, especially in the life sciences where you spend a lot of your time.

Dr. Sukumar: Absolutely. The only way you can drive engagement is by getting to know what your customers want, getting to know what matters to them. So we offer certain infrastructures and technologies that are unique about understanding what our customers' expectations are, what people's needs are, and how you begin to address them. Today's world—we live in artificial intelligence and speed and automation, which is great in terms of leveraging all that historical data. But if you're trying to understand what matters to your customers, your humans, you need a unique technology that begins to learn about that into the future.

Richard Byrd: Yeah, I agree with you for sure. That technology you're talking about—tell us a little bit about that.

Dr. Sukumar: Yeah, absolutely. The technology is called AceMap—A-C-E-M-A-P. The letters MAP reflect multi-attribute preference, and the first three letters, ACE, reflect adaptive self-explication. The idea is that you can feed this algorithm a large number of things that are potential solutions to what customers might want in the future. And through this adaptive algorithm, we collect data individually—one individual at a time, one customer at a time—to understand what they would prefer. And what would that hierarchy of preference be, across the many things you might bring to market? So AceMap is a very powerful algorithm that helps us understand what drives people's choices, what drives their future behavior, and how you leverage that to look at the market as a collection of these individuals for whom you're trying to address the things that matter to them—and how you bring these new opportunities to the market.

Richard Byrd: Man, that is such an important thing to do. I think people make a lot of assumptions about what their customers want. We see it all the time—people get product feedback from the salespeople, and customers say this and they say that, and ask for this and want you to develop that further. But they're not always willing to pay you for it. That's where—we've done some conjoint analysis in the past with clients, and it really helps to put that “what are you willing to pay for it” overlay on top of it. Because if you don't, people will ask for all kinds of things. But it costs money to develop that stuff.

Dr. Sukumar: Absolutely. And Richard, you know this—you mentioned the word “conjoint,” which is a combination of two words: “considered jointly,” and the idea that we as individuals, when we make choices, consider many things together. And we make trade-offs. You just described one: I want all of these things, but I can't pay for them, so I've got to balance what I get against what I'm willing to pay. AceMap is a trade-off analysis technique that's kind of on steroids, because it lets us look at a space consisting of many different potential value drivers or benefits you can bring to market. And you trade it off against whether people are willing to pay for it. And by the way—we talk about price and money as “willing to pay,” but in a world where we're poor for time and bombarded with new technologies, that trade-off isn't only about money. It's about time, it's about effort, it's about the learning you have to put into something new, and whether I'm willing to spend that. So there's a powerful need for a technology that can do this with many more of these dimensions, can do it faster, can do it individually—not a collective algorithm, but one that can actually reflect the differences between Richard and Sukumar, or Cindi. You need it to be very individual, capturing what each of us is expecting, what matters to each of us in the future.

Richard Byrd: Yeah, that is so true. We work with a lot of B2B clients, and there's always a buying committee. Many times they're at odds with each other on what they want to achieve or what their expectations are for a service or product. We really see it a ton with conjoint. It helped us get through that with them, because what a driller is interested in is maybe very different from what a geologist is interested in, versus an asset manager, or whoever we were talking to. And it was really helpful to see that—by company, by geography—and you realize there's no one value proposition that works for everybody in a buying committee. Being able to educate the salespeople going in to talk to them is so valuable, because you could really be barking up the wrong tree. If you go in and tell a driller the same thing that made the geologist happy, you might get choked to death.

Dr. Sukumar: Absolutely. That makes perfect sense. And Richard, you gave a great example, because if you think about drilling companies, you just described a few things. One, there's a buying committee. Two, there aren't thousands or millions of them the way consumers are. So you need a technology that works for a small, finite number of individuals, trying to understand what's driving their behaviors. But more importantly, you described something—there's potentially a physical product, but also software that drives that product, plus support and services that may be important. And if you took a geologist, maybe what they really care about is the data and analytics that come out of it, because they're trying to forecast, and they care about the quality of the data and whether it can be taken across platforms. So as you think about where those unmet needs are, what matters to each of these stakeholders, I really need to understand each person on that purchasing committee individually and separately. To recognize that if I just took an average, I might be in a position where I'm not satisfying anybody. As someone said, the traditional conjoint models put together the guy who wants hot tea and the guy who wants cold tea, and come back and say the market wants lukewarm tea. Now the market spits out lukewarm tea—when either, if you're down South, you really want cold tea, or if you're up North, you want really hot tea. So we truly need to understand what matters to each of those individuals, so we don't make the mistake of the law of averages and put together something nobody really cares about.

Richard Byrd: Yeah, the law of averages works great for textbooks, but not in the real world.

Dr. Sukumar: Absolutely, absolutely.

Richard Byrd: Well, tell me a little bit more about your team. How big is your team over at OSG?

Dr. Sukumar: Yeah. We work across industries, as I mentioned. Healthcare is a big space—both pharmaceuticals and medical device companies—but we also do work in consumer packaged goods and with retail customers. So there's a diverse set of people required on our team. We need people who have knowledge of the domain. We need technologists who know how to handle data, because while we collect and bring data, we also look at historical data and need a data strategy for how to bring it all together to solve the business problems and address the problems around growth. So we have people who are really good at data strategy, at looking at data across different warehouses and technology platforms. And we need people who are good in the use of AI—particularly agentic AI—when you're trying to predict, but predict the data that's more relevant, more current, as against just historical data. We need a very diverse set of individuals, including project and program managers. So we're a pretty decent-sized company, with about 75-plus individuals who understand different domains, different businesses, different business challenges, but who are also continuously learning about new technologies. We also work a lot with partners who bring us talent—think of it as just-in-time talent. As our customers have needs in marketing or marketing technologies, these partners bring us talent we can leverage as we go. The key thing, Richard, is we also work across geographies. It's not just the US. We've done work in over 60 countries, believe it or not. And one of the powerful things about AceMap is that because you're looking at how people make trade-offs—as against asking a question on a scale from one to five—you remove cultural biases and really understand what motivates people to make the choices they make. If you ask questions on a one-to-five scale, you know there are individuals who culturally want to be very nice and will give you all fives. And there are others who are very strict in the grading scale and may only start at a four. How do I compare these individuals? How do I bring all of that data, which gives you signals that are more noise? With AceMap, we get to those cultural differences and bring a deeper understanding of what matters to individuals across different geographies, cultures, and languages.

Richard Byrd: Wow, that is such a good point, because the way people score things can be really different. I got in trouble with people in my company because I would never give a meeting—we rate our meetings after we finish—I never gave anything a 10. And they were like, “You are too hard of a grader. The highest you've ever given us is an eight.” And I'm like, whoa, nothing's perfect, that's just the way I think about things. But other people are like 10—I know they talked me into it. I'm soft now. I give them 10s if it's good.

Dr. Sukumar: The more critical question, Richard, is: you gave them an eight—was there a diagnosis around what would have made it a 10? What was the gap, and how do I diagnose that? Lots of companies do customer experience measurement, as you know. You get off a plane, or you finish staying in a hotel, and you get a one-question survey in an email or text: “Rate us on a scale from one to five: how likely are you to recommend?” or “How satisfied are you?” But step back. Sometimes that data is a lot of noise, because it never told you—if I gave you a four, why? Did I give you a four because of X, Y, or Z? It's very hard to get people to tell you what they were expecting when they came into the service—the airline, the hotel—and why their experience fell a little short. That's very critical. You make a good point, but the real question becomes: can I get more diagnosis? Can I actually fix the gap between your eight and 10? And what would it take to do that?

Richard Byrd: That is the perfect follow-up question. I've always felt that way about net promoter score. Okay, you rated me one to 10 on how likely you are to recommend me—you gave me an eight. I don't know what to do with that data. We have a bunch of eights, and there was one mad customer who gave us a four. I don't know what we're supposed to do with that, other than track it over time. Something a little more diagnostic would be a lot more useful.

Dr. Sukumar: Absolutely. And that's a classic example of when individuals who gave it an eight fall into “passives,” because only nines and 10s are advocates, right? So your scores start to wobble quite a bit over time, depending on who's responding and who's not. I think that score's been misused—misunderstood or misused, I'd use both words quite extensively. If you go back and read Fred Reichheld's book when he wrote that, his whole thing was that what you really need to do is go deeper and understand what your customers are expecting. You've got to segment your customers to understand what they were expecting and where the gaps are. That score was meant to be more of a high-level indicator, not something we'd use as a currency. The reality is it should largely be one that tells you that you need to understand better what your customers are expecting.

Richard Byrd: Yeah, that's right. Because if they give you an eight, but their expectation is “yeah, eight's great, I like them, that's the kind of company I want to work with”—then your marketing people are going, “Hey, we have all these passive people, nobody loves us.” You can really react to that data in the wrong way.

Dr. Sukumar: Absolutely, yep.

Richard Byrd: Walk us through your career, because it's interesting where you are. I think when you were probably a little boy, you didn't think you were going to grow up and get into this voice-of-customer and data world. How'd you wind up here?

Dr. Sukumar: Not at all. I've got to tell you, when I was a little kid, I grew up in a big city called Mumbai in India. At that time, my parents told me I either had to become an engineer or a medical doctor. And of course—what happened is, yes, I became a mechanical engineer, got my degree as a young undergraduate. Then I came to America. I was very keen on thinking about products and how to develop them, and I came to get a degree in marketing—or what really started as operations research, and then finally ended up as a PhD in marketing and analytics. This was many years ago now. I started my career as an academic. I was a young assistant professor at the age of 25, and believe it or not, Richard, exactly where you are—in Houston, at the University of Houston, and then at Rice. So I started as a marketing professor, teaching in the undergraduate and graduate schools there. I spent more time in academia, until one fine day I decided to leave to help bring some of the things we teach and develop in academia to the corporate world—to help shape how people think about customers and consumers, whether B2B or B2C. That's how I transitioned from being an academic for 15 years to being a practitioner. I founded OSG as a company that started by helping understand what physicians wanted, what patients wanted in new treatments, and how to bring those to market.

Richard Byrd: What a ride. That's so interesting. How do you make that transition from academia to the commercial world?

Dr. Sukumar: Oh gosh, that's an interesting question. When I was in academia, I loved it. I loved teaching. I loved being able to shape young minds while teaching in Houston, Texas. Many of the oil executives were folks in my executive MBA class, and I had a ball. I decided at one point that I could continue teaching if I started working directly with corporations, helping shape things. It's change management sometimes, because many folks in senior positions are driven by their gut, which is not a bad thing. But quite often these days, decisions made from the gut, without evidence or strong data, lead to a lot of risk in how you drive growth. So it became one of those where, if I could teach as well as bring solutions that were data-driven, I could have the best of both worlds. I could create change management inside large corporations while demonstrating strong customer insights and evidence that would get executives to make the right decisions and de-risk making a bad one. I think that's what got me excited as I transitioned out of academia into starting a business of my own.

Richard Byrd: Yeah, you touched on so many things there I couldn't agree more with. That ability to use data to make better decisions that are less risky is critical to everything. When you go with your gut, it's a 50-50 shot. Everybody thinks their gut's a little better than everybody else's, but statistically it doesn't bear out. People who seem like they're right all the time make just as many wrong decisions. Placing the right bets in business is so important. And then you went out to start your own company—and that's when you're making all kinds of bets, right? On a daily basis, you're making bets, and you really want to make the ones that are going to pay off, and base it off data.

Dr. Sukumar: Absolutely. Yeah.

Richard Byrd: The only problem with being a small-business person is you don't always have the data.

Dr. Sukumar: True, true. But there have been times when we've actually collected it—I'd say I drank my own Kool-Aid. I'd go out to our customers and ask what they're looking for when it comes to things like customer insights and analytics. And like the example you gave, we often have multiple decision-makers who ultimately decide not only who to choose as a vendor partner, but also how to use the information and insights we bring. So it's very critical to understand all the stakeholders. As part of that change management, Richard, it's very critical to understand what's in it for each of them. What are the expectations? How are they going to use it? The use of the insights, analytics, and technology we bring is different for someone in marketing than for someone in sales. It becomes very important to ensure they trust and believe, and are willing to leverage that data and insight to make the right choices.

Richard Byrd: That is such a point, because you mentioned sales and marketing. But one area I'd love to see technology like AceMap used more widely is in the R&D section of a company. A lot of our clients—I don't know if this is true in the world you work in, especially the pharmaceutical world, where it's so expensive to develop any kind of drug—but a lot of our clients are very engineering-led companies. They love to develop technology, and they've built roadmaps in their minds of what they're trying to achieve. And so many times they haven't asked the customer recently, “Do you really want this? And are you going to reward us and pay a premium if we develop it?” They make things—maybe they engineer 10 features on a new offering, and it takes five years, $20 million, whatever, to build it all out. Then they turn around and find out the customer only wanted two of those features, and they weren't willing to pay as much as we hoped to capture. It becomes a marketing-and-sales problem at that point, instead of a product problem. But those things can be solved so much earlier and save companies so much money and trouble if they knew what they were developing right off the bat.

Dr. Sukumar: You're so right. You're absolutely right. And it's across industries. Today we work in the medical technology industry, and it's got people just like you mentioned—engineers. Medical technology has become so complex that there's so much data being collected, and so many different machines. Developing that kind of technology and software requires really smart engineers. But like you said, these engineers already have a presupposition of what they think the right product should be—the target product profile, the optimal product, the minimum viable product. And quite often, when they go to market and try to assess what customers really want, they're surprised. Sometimes what they thought was an attractive solution—only certain elements are appreciated, and certain others are not cared for. It's a good exercise, a good tension to have, but as a result, it helps you prioritize what the development roadmap should be. So when you think about why products fail, that's one of the areas to keep in mind: when you develop something the customer really doesn't care about or won't pay for, that's definitely one of the reasons why. You didn't have the insights, or you didn't create the differentiation that really mattered to those customers. So it's very critical to think about it that way.

Richard Byrd: Yeah. One of my good friends gave me a term I use all the time now. He said, “Is this a regular problem, or is this a bleeding-from-the-neck problem?” Because if it's a bleeding-from-the-neck problem, customers are going to pay for it, and they'll want it fixed right away. If it's just a regular problem, everybody's got problems, and they'll maybe fix it one day. But if you're not solving the bleeding-from-the-neck problem—and then talking about it in your sales and marketing—you're not going to see the growth you're probably looking for. It's a nice-to-have. “We'll look at it next quarter,” which means never.

Dr. Sukumar: Absolutely. Yeah. And depending on which customer you call, you get a lot of anecdotal evidence. If you call on the customer that's complaining all the time, you get a lot of reasons for why they don't want you, and you feel like solving all of those problems is important. But the reality is it's anecdotal, it's one-off. And is that sizable enough for a company to invest in and design, develop, and launch a new solution? That's the question one has to ask.

Richard Byrd: Yeah, 100%. And data is one of those interesting things—all data is a snapshot at one point in time, right? If you're only talking to a handful of people, and you talk to that one person, and he's having a bad day, and he tells you all the things he hates about everything you do, then you could be getting skewed data. How do you guys get around that with AceMap?

Dr. Sukumar: Well, we clearly start by trying to understand all the possible problems. There was a methodology called “jobs, outcomes, and constraints.” The idea is that you gather all those anecdotal issues that the engineers think are an issue, and you begin to pull that in. So we might do workshops internally with the engineers, to understand better what all the challenges are that they feel their customers face. And then we go out to the customers and observe them. We observe: what is their journey? What jobs are they trying to do? What outcomes are they seeking? What constraints do they face with existing solutions? Are they in a situation where certain things aren't working? Why not? Do they have a workaround? With all of that qualitative understanding, we transition it into a more definitive description of a benefit that might come from solving that challenge—being able to address that job they've got to do that they cannot do with the existing solutions. At that point, we generate a good, solid list—sometimes as many as 50 or 75 various needs and benefit statements—which we can then take to a larger sample. It could be as little as 20 or 25. Depending on markets, if you're dealing with a medical device, you might get lab technicians in the hundreds. Or a pharmaceutical medication, you could get physicians in the hundreds. So depending on the area, we need only as little as 25, but we can go as high as you want, because this whole technology works one individual at a time. We're then able to leverage AceMap to understand what that hierarchy is in terms of the needs that have to be addressed—and are they big enough that they're seen as unmet? Think of it as looking at: does it drive my choice of a new product? How big a problem is it that I need it addressed now—the unmet, the bleeding example you gave? That begins to shape up things we call potential differentiators. There may be other things that are important—you need tires and a wheel, you're looking for good miles per gallon—but it's not really an unmet need, because others have solved it. That becomes a table stake, a cost of entry. You've got to have it, but just having it isn't going to be a solution people buy. So we understand those areas of differentiation today, and the ways to shape the market in the future—to educate them, where there are problems that are unmet but not everybody cares about. How do I shape the market for the future? How do I think about innovations? No idea is a bad idea. If you're an engineer and you bring an idea, it's not a bad idea. It's just that it may not be a big enough problem today that your customers are running out the door to find a solution. So if you can assess that, you can prioritize it as a market-shaping opportunity, rather than one where a solution in the short term brings you revenue. You leverage that data to shape what the roadmap should look like—where to prioritize, where to invest today, and where you might invest only after you've shaped the market about the need.

Richard Byrd: I really love that approach. It does a few things. There's so much bias a lot of the time, because those engineers were told, “Hey, we want to make a new product,” or “advance this product, and we need you to engineer these X features.” And they do, right? And they haven't thought about what the job is that people are looking to do. How many times have you seen a product go to market, and everyone thought they knew the value the customer was going to get—and then the customer finds a whole new value set on their own that the engineers never stopped to consider? If they'd known that was the ultimate value, it would have made product-market fit so much easier. But they didn't think to ask those questions, if they don't use the approach you just described.

Dr. Sukumar: Absolutely. And the tool brings a little bit of discipline. It brings the ability to put many things into it to understand what matters to individuals. Most importantly, it avoids that lukewarm-tea problem—where you ask on a scale from one to five, everybody gives fours and fives, and you think a 4.5 is a better rating than a 3.5. It just happens that the 3.5 might have been something a smaller group of people really cared about, and a solution for that particular dimension might have won you more business than going after the lukewarm 4.5 that didn't solve anybody's problem.

Richard Byrd: Yeah. Or it's already something your competitors have solved—now it's table stakes, just like you were describing. We see that a ton with clients. They really want to talk about something where they're matching a competitor. But it costs money to do marketing, and you don't want to spend it talking about things that don't differentiate you from your competitors at all.

Dr. Sukumar: Right.

Richard Byrd: What's one thing you know now in your role that you wish you had known from day one?

Dr. Sukumar: Well, I think the whole topic around the fact that what we do is a combination of both technology and people—service. You're not only selling data and technology, you're also selling domain expertise. You need a combination of those two things to win. It's important for us to have good, critical, unbiased data—data everybody can align on. It's like a data orb, if you want to call it that, where everybody says, “I believe in this, and it's going to change the way we think about the market.” That's something we do very well, but also something we need in our business. And I think AI is not going to be able to replace that—it's people who know the industry well. So Richard, you know this: oil and gas, even exploration and production, is a highly sophisticated industry. You need people who really know everything about that industry and can bring credibility—get the data to speak for itself, but also bring that credibility to those customers. That combination is very critical. I also think good-quality data is very important. There's a lot of data in healthcare, in oil and gas, in—pick an industry—consumer packaged goods. We talk about big data, but I go back to the early '80s, when the grocery business used to put out a lot of data. Every SKU at the grocery store—you've got a 65,000-square-foot store with over 100,000 SKUs, and every time somebody buys, that's data. So big data has existed for a long time. But the question we all have to ask is: is that data of good quality? Are we able to use it correctly? And how do you bring it all together? It's a combination of things critical for the success of OSG, but also for helping bring success to our customers. You need good people, good data, and the ability to actually bring all of that data together so it can help solve the problems our customers face.

Richard Byrd: No, I agree completely. Having data is one thing, but no matter how good the data is, there's always a signal-to-noise ratio. Being able to interpret that—the signal from the noise—and make sense of it, then explain it to somebody in the context of their business, where they have actual insights, so now they know exactly what to do. I've been in situations where there's too much data and no one there to interpret it properly, and then you're more confused than ever. “Well, we could go this way, we could go that way”—same decision I was at before I had all this data, and now I'm more confused than ever. What about from a sales and marketing perspective? Because I know you touch both of those worlds very heavily with your data. What's the biggest obstacle you face in your industry?

Dr. Sukumar: The biggest obstacle I face today, if you put it into perspective—the last three to four years have seen a skyrocketing use of new technologies, artificial intelligence particularly. And now more and more we talk about not only generative AI, but also agentic AI, and digital experts and bots doing everybody's work. It's become everyone's feeling that these bots will begin to do everybody's jobs. But these bots and these AI agents are only as good as the data they sit on. The traditional thinking is that there's so much data out there, why not just use it and use artificial intelligence to answer your questions and bring insights? The problem is: how do you educate your customers that those bots and AI tools are available to you and to your competitors? And most importantly, they sit on a shallow data stack of things from the past, and are not yet reflecting what your customers may want in the future. So how do you bring the future and the foresight of data into the now—which reflects all of the past? And how do you educate your customers that this is important, and that you need something unique and competitive to meet and beat your competitors? Because if everyone goes to the same AI tools and builds marketing or sales engines on that basis, it matures very fast. It plateaus very quickly, and there's no efficiency, no effectiveness in how you win when everyone has the same information and the same tools. So how do you educate your customers that bots are good—they bring speed, they bring cost efficiencies—but they're only as good as the data they sit on? The shallow data it sits on happens to be one that everybody has. So you really need to think about how to disrupt the market by thinking differently. I don't know if I'm making sense there, but yeah.

Richard Byrd: No, that makes so much sense. We're seeing the same thing, and I think everyone can see it. If you just go look at your LinkedIn feed—so many people are using AI to write their posts, and it's just slop. Everyone sounds the same. They're all talking about the same thing. It's just an echo chamber. And I can tell a post written by a human, and I can tell when it was generated using ChatGPT—even if you train it on your voice, it's just very superficial, like you said. It's not deep, and it shows. Everything starts sounding alike, looking alike. We were at a trade show for one of our clients, and I was walking around, and you could see so many of the graphics on the booths were done by AI too. And you're saying, “Man, everybody looks the same. What happened?” The bar for low-quality stuff has gone up, but now it's all low.

Dr. Sukumar: No, absolutely. There was an oil executive in Houston who once—we were having a chat about why products fail. He said, “Sukumar, I think there are three reasons for that,” and I think it stands the test of time. He said the first thing is not having customer insights. The second is not having clear differentiation—why you, and how are you different? And the third reason it fails is a lack of effort. You think about lack of effort as: do I throw more money at something? Do I spend more on sales and marketing? Do I buy the nth martech stack and send out more emails and more blog posts? But the reality is that effort isn't only money. That effort is also knowing you're starting on the right customer, and speaking to them to convert those behaviors. Because today, if 10 customers buy it, the hundreds after them start to follow. So knowing you started at the right end of the market, and not the wrong end, is very critical. And that data—helping you shape where to begin—is also very critical. So we really have to think about new innovations, new products, as well as existing growth from existing products, continuing to reflect customer insights: knowing how to create that differentiation, knowing the customer journey and where you interact with them, and most importantly, who those people are that you begin with—and being able to spend the right amount, very effectively, to deliver that growth engine. So the depth of data needed for that is far different than what you can get today in an LLM.

Richard Byrd: Oh man, well said. And that kind of halfway answers my next question. But what is the biggest misconception people have about your industry?

Dr. Sukumar: I think it goes back to what we were just talking about, which is that AI is going to solve the whole problem. If there's one word that describes the outcome of what we do, it's improving engagement. Customer engagement, patient engagement, human engagement. We're talking about how to drive that behavior change, and the feeling that AI is going to be the magic bullet is the misconception out there. How do you get people to think differently? I'll give you an example in healthcare. There's a general feeling that you'll have agentic AI nurses that address all your health problems. But how does the agentic AI nurse work if it's going to use data that isn't necessarily complete and appropriate? If the data is weak, the agent's responses aren't there. And you only need one instance of something going wrong to potentially create challenges. That doesn't mean those bots and agents aren't valuable—they are, there are many things they can do to alleviate the pain of shortages in healthcare workers, just as they can in engineering or technology, to alleviate workforce shortages and create more efficiency. But people have to be educated in how to use it, where to use it, in a manner that brings not just efficiency in cost and time, but is also more effective at driving the outcomes a company is looking to drive. That education is going to take some time.

Richard Byrd: Yeah, well said. I was listening to a podcast the other day, and a guy on there was an AI expert, talking about just what you're talking about—AI taking people's jobs. He said it's definitely going to be disruptive in the job market. But he said, let's say anything—starting a business, coming up with a product design—is a 10-step process. AI is not good at the first two steps: knowing what to build and why. That's a human skill. Some humans are good at it, some are not. But knowing what to build and why. Then the next three steps, three to seven, it's probably pretty good at—“just generate this,” or “write this line of code,” or “help me do this better.” It increases efficiency big time. But then the last two steps, it's not as good at—now we have to bring it to market, convince other people, do the change management. That's where AI is going to really struggle for a long time, and that's where humans can really do it. So if you think about every process like that: those two initiating steps get things going and point it in the right direction, then you align AI to help execute quickly, but then you have to socialize it with the other humans to make it live. I thought that was a really good breakdown.

Dr. Sukumar: No, absolutely. I think you're spot on. That front end of developing a new innovation—it's solving problems for people, for humans. You cannot do it without their involvement, or their understanding of what matters to them. And it's the same thing when you take it back out to the market. Where you can leverage AI is that in-between step. And even that requires some monitoring, so it's doing it correctly. But the end game of bringing that innovation to market—to the right customer, at the right time, at the right place—still requires humans to oversee it.

Richard Byrd: Yeah. Well, tell me a little bit about where you see your industry going over the next five to 10 years.

Dr. Sukumar: I think we're going to be inundated with more data, that's for sure. We're already inundated with complex data—structured, unstructured. We're going to see systems and technologies that can do this faster and cheaper. We're also going to see automation that addresses problems more of an execution nature—exactly like you described—to do them faster and cheaper. But the emphasis is soon going to come back to recognizing that data quality is important, and that the type of knowledge and information you need has to be strategized up front, before you go into the world of the digital experts and the automation. So I think we're going to start to see a migration from using LLMs to building LLMs on more focused and deep knowledge—deep data that reflects not only historical data, but also some of that foresight we're talking about. How do I bring in what matters to customers, what matters to people, and incorporate that along with the existing, quote-unquote “knowledge gardens,” applying them to very domain-specific things? So better use of data, better understanding of the data, and the AI and automation tools structured more and more around unique data that a corporation believes it owns—that differentiates it from its competition and allows it to be more successful going forward. I'm seeing some degree of executives beginning to think that way. I think it's going to be a lot more pervasive as you go forward. So there are certain routine things they're most comfortable with—let them be automated, whether that's website marketing or website design and collateral design. But I think they're going to come back to wanting things to be unique—exactly what you said earlier, where you walk into a trade show and everything looks clearly designed by a very broad-spectrum ChatGPT and Claude. And now we're going to come back to saying, “I want it to speak the language of my company. I want it to reflect what we uniquely bring out. I want it to reflect our brand book. I do not want it to be the same set of words and images pulled from a broad-spectrum LLM.” I think we're going to start to see a greater focus on that. Hope that makes sense.

Richard Byrd: No, that makes sense. How are you guys preparing for that at OSG?

Dr. Sukumar: I think it comes back to three things: making sure we have the right technology; making sure we're reflective of the right data sources—some historical, and some, like what we bring and add to the model, unique and differentiated for a customer; and then making sure we have the talent on our team that has the domain expertise. The three together will continue to speed us up, to deliver things more cost-effectively. And most importantly, continue to deliver solutions that are impactful—because that impact comes from knowing the customers of our customers, knowing what matters to them, and bringing that forward-thinking data into the knowledge pool, so the AI and automation is crafted more uniquely for a customer of ours, rather than generically using LLMs.

Richard Byrd: You know, I could see that being so much more powerful—bringing in that true data about what your actual customers are thinking and saying they want, and what trade-offs they're willing to make to receive that, with the power of AI behind it to extrapolate that, or improve the efficiency of how you deliver that message. I could see that being very powerful.

Dr. Sukumar: Yeah. And you know, Richard, about 20 years ago, if you'd used the word “data,” the only thing that went into people's heads was numbers. But today, when you say “data,” it's no longer just numbers. It's voices, it's text, it's videos, it's graphics. It's embedded in all of that. It's not only what people said, but the emotion with which they said it—what they portray, and how that reflects. So when I said “technology,” it's just one word, but realistically that technology is very complex now—much more complex—because it can bring and synthesize all of these different kinds of data. Imagine, 20 years ago, if you did an ethnography and went out into the field to see how a geologist or one of those engineers used the drill bits, you observed and walked away. But now you're actually recording that and making better sense of that data, besides looking at all the numbers and everything that goes along with it. So the word “data” means very different things today than it meant to people 20 years ago.

Richard Byrd: Absolutely. And what AI can pick up on is pretty amazing too. We use this tool called Fireflies to analyze our calls, and it detects accents as well as I can, and it can even tell when people are being sarcastic with their tone. I love it when that happens.

Dr. Sukumar: Well, it's actually a better reader of how engaged someone is in the conversation. Sometimes you move, and you're not reflecting on what the speaker is saying, or you're multitasking. It starts to flag some of those things very quickly these days.

Richard Byrd: Sure. Humans are notoriously bad listeners. My wife will tell you that about me—that I'm a terrible listener.

Dr. Sukumar: My wife will tell you the same, if she ever gets on a Zoom call. I'll get a text message from her saying, “You're not focused, you're not listening.” She just seems to know. She's my AI bot who figures out whether I'm engaged or not.

Richard Byrd: We're big-picture guys, you know. We're thinking about the next thing. I love it. What about from a sales and marketing perspective—what tactics are you most excited about at OSG?

Dr. Sukumar: It's interesting you ask what tactics. I was looking at some reports recently that said most chief marketing officers use as many as seven different channels to communicate. So if I focus on the word “tactic,” it comes down to the fact that there are many more tactical ways to engage a customer, but not all of them are effective. And there's technology that can bring speed and low cost to many of them. Think about another statistic: most consumer goods companies today spend as much as 80% of their marketing spend on digital medium. When I say “digital medium,” it could be anything from emails to texting to Google ad placements to Instagram influencers. But they only generate about 30% of their sales through e-commerce and digital means. So there's still a lot of waste in there. If someone's using seven channels, they must be spending a lot more money across every one of them. We have to study and understand the return on marketing investment from each channel. Whether it's B2B or B2C, you have to ask: are they well integrated? Are they well sequenced? Is the message hitting the same customer with the right kind of sequence to generate an effective outcome? I think we need to bring more appropriate data, more unique data, to our customers, so they can understand which channels to abandon—because you can attribute and say, “That's not very effective for you, you need to move in a different direction.” So what are we preparing for? We're preparing to continue educating our customers that you have many more tactics and tools at your disposal—don't jump to use all of them. They're all nice, flashy tools. Let's go back to understanding what matters to your customers, how you might message to them, where in their journey, and how you'd target those messages with the right sequence, across the right channels, with the right frequency, so as to convert. Let's really understand that and attribute the return on marketing investment more appropriately than what you're doing today. So there's some work there for us as well, tactically—both for our own business and for the businesses of our customers—to help reduce the many shiny tools people use but don't get returns on.

Richard Byrd: You really hit on a nerve there. There's no shortage of tools and no shortage of channels out there, and every day there's more. Knowing which one is appropriate for a client and their customers is not always easy. And what we've seen is it changes—it even changes for the same type of target audience. What worked great six months ago isn't working as well anymore. We operate in such dynamic markets these days, everything is changing so fast, competitors are responding. And some things that were very popular five, six years ago, that people said, “This isn't as effective anymore, I'm going to give up on it”—now they're back. I have a client we're helping with a paid search campaign, and I didn't think it was going to be very effective. I thought it would be expensive. It's kind of old hat to do paid search these days, with all the other fancy channels. But it was awesome. And it was very low cost, because none of their competitors were doing it, and the customers had high intent—they were at the bottom of the funnel and ready to buy. So it worked great, but I wouldn't have assumed it would. It was low on my tactics list. But that's where you have to let the data show you, because everybody has biases. I had a bias against that channel because I didn't think it would do anything. But the customer said otherwise, and they're the ones who matter.

Dr. Sukumar: Absolutely, yeah. And I think you're spot on. What's going to happen is a lot of these technologies, a lot of these different channels, are going to come under a single roof. Right now we see disparate tools, and there'll be a huge roll-up, and they'll begin to come under a single infrastructure. There'll be a lot of consolidation. And there has to be intelligent data that helps you decide which channel—or which portion of that big house of tools—makes most sense to use, for which customer, with what sequence, and how often, so you don't overload them. Because today, Richard, as you know, you could send one more extra email at no cost to somebody. But actually there is a cost, and that cost is somebody getting very frustrated and throwing you on the unsubscribe list. Because it's only a button away, with all the regulations. It's only a click: “get me off the list.” So we've got to be mindful that the cost is not just the cost of technology, or another email, or another paid ad—the excessive bombardment also leads to disillusionment, and people getting a negative perception of your brand. Being able to know what your customers really want, how to sequence it, how to personalize it, is going to be very, very important.

Richard Byrd: I couldn't agree more. It's something I often reflect on. People are already so inundated with advertising and messages, and it's never been more than it is now. And I think about, with AI democratizing that and making it so much easier and more affordable to produce all these things, it's going to be exponentially worse going forward, and people are going to be so burned out. I wonder about the ability to reach consumers. They're already skeptical and burned out, and this is just going to dump more kerosene on the fire. So how are we all going to adapt? Because it's not if it's going to happen, it's when. And I think we're already there.

Dr. Sukumar: No, and I think smart marketers are going to realize this very quickly. Those up there in the maturity curve are going to quickly recognize that they may have all the tools and technologies at their hand, but the real solution is to find what's going to be most effective for every customer, one at a time. I think back to the days when I first started in marketing—we didn't have any of these technologies. And when a colleague began to talk about one-on-one marketing, I was in my head going, “That ain't gonna happen. How am I going to speak to one customer at a time? What do I have?” We're talking late '80s, early '90s, and what we know today didn't exist. But now we know there are tools that can speak to you individually. But our attention spans are becoming more and more limited—so if it doesn't do it smartly, and in a manner that generates the right kind of engagement, I think we're going to have a backlash of a different kind, where money spent is money wasted, rather than money spent yielding the growth results corporations are looking for.

Richard Byrd: Yeah, it's one to watch for sure, because you can see this train headed down our path. Well, that's some good advice for marketers. What's some advice you'd give somebody just starting their career out in sales or marketing?

Dr. Sukumar: It's a great question. I still get a lot of students coming out of academia calling me every day about internships, and I speak occasionally back in the classroom. For those in sales and marketing, there are two big pieces of advice I'd give anyone. One: always do your homework before you go in front of a customer. And two: always be prepared to ask a lot of questions of your customer before you make the decisions you've got to make, or before you're trying to close a sale. If you're in sales, you're trying to close a sale. So it's very important that you do your homework and prepare a list of questions you can ask your customers. Because the more knowledgeable you are, the more likely you are to get a sale. And the more knowledgeable you are, the more likely you are to pull down the barriers—where a customer is considering you or someone else for a deal. So let's not forget the importance of asking good questions in helping remove the barriers to closing a deal.

Richard Byrd: That is very well said. I think about all the deals I've tried to close, where I thought I had a really good bead on the customer and knew what their problem was—I knew before I even walked in the room what their problem was going to be. And then I present a solution that wasn't right for them, because I didn't ask enough questions and I misdiagnosed the problem. I think you just can't ask too many questions in those meetings, for sure. Well, what's something we have not talked about that we should have talked about today?

Dr. Sukumar: No, I think you asked a lot of great questions, Richard. It's been a fantastic conversation. But I'd love to leave you and the listeners with a couple of important comments. One, we all know artificial intelligence is here to stay, in all its forms. And we all agree it's driven by the data it recognizes and is able to put out there. So it's important to step back and often ask ourselves: has the business problem changed? Because I don't think the problems businesses face are any different now than 20 years ago. So the real question becomes: am I using artificial intelligence on the right data and the right knowledge? And am I able to make the right decisions for our customers, our companies? And if not, what needs to happen? I bring it back to the three important ingredients: there's technology that brings speed and reduces costs; there's data that goes into that technology and makes it smarter—so make sure the data you have not only captures the past, but is also reflective of what matters to customers in the future; and then there are people, who—as you pointed out—the first two steps and the last steps are still domain-specific, driven by people, needing people and knowledge. So I want to leave everyone with that thought: we need to make sure we're using technology appropriately, with the right kind of data, and with the right kind of people and domain, so that automation and speed can be combined with delivering effectiveness to our customers—to the very people we work for, whether that's the end consumer, the patient, the physician, or the geologist. So you want to make sure we're doing the right thing in that fashion.

Richard Byrd: Very well said. Thanks for leaving us with that parting thought, Dr. Sukumar. Thank you for being on the show. How can people reach you if they want to get in touch?

Dr. Sukumar: Well, first, thank you very much for having me. I enjoyed the conversation today—great conversation, great dialogue. They can reach me through my email, which is r.sukumar—S-U-K-U-M-A-R—at osganalytics.com. So: opportunities, strategies, G for growth, and the word “analytics,” A-N-A-L-Y-T-I-C-S dot com. They can reach me on my email, and I'd love to have a conversation and a follow-up with them.

Richard Byrd: Perfect. Well, thanks again for being on the show. Until next time—we'll talk again.

Dr. Sukumar: Great, thank you for having me.

‍