University of Arkansas

Walton College

The Sam M. Walton College of Business

Episode 95: Dinesh Gauri

Dinesh Gauri is a Professor of Marketing at the Sam M. Walton College of Business at the University of Arkansas. Dinesh has expertise in marketing science as well as retail strategy, pricing, analytics, big data, omnichannel, and shopper marketing. Read More

More About This Episode

Dinesh Gauri is a Professor of Marketing at the Sam M. Walton College of Business at the University of Arkansas. Dinesh has expertise in marketing science as well as retail strategy, pricing, analytics, big data, omnichannel, and shopper marketing. He is recognized as a thought leader in retailing and was ranked 3rd in the world in the retailing journal influence index from 2009-15. Gauri holds the Walmart Chair in Marketing and is the Executive Director of Retail Information for the Walton College.

Read more about Dinesh Gauri and view his research.


More Episodes

Listen on Apple Podcasts
Listen on Spotify
Listen on Google Podcasts
Listen on Amazon Music
Listen on iHeart Radio
Listen on Stitcher

Episode Transcript

[music]

00:07 Matt Waller: Hi, I'm Matt Waller, Dean of the Sam M. Walton College of Business. Welcome to Be EPIC, the podcast where we explore excellence, professionalism, innovation, and collegiality, and what those values mean in business, education and your life today.

00:27 Matt Waller: I have with me today Dinesh Gauri. He's a Professor of Marketing in the Sam M. Walton College of Business. He has expertise in retail strategy, pricing, analytics, big data, omnichannel, and shopper marketing. He's also from an academic side, he's a real expert in this topic called marketing science, which we're gonna talk about today. In fact, in 2018, he received the Walton College Excellence in Research award. And Dinesh has been at the University of Arkansas for about four years now. He was previously at Syracuse University. But Dinesh is an expert, as I said, in marketing science, and he's actually helping to build our strength in that. Dinesh, thank you for agreeing to be interviewed today. I appreciate it.

01:22 Dinesh Gauri: Well, Matt it's been a pleasure and an honor. As you said, I've been here for four years, and these four years so far has been, it's really good and much better than the previous 15 years or so that I had in the academic world, because as you mentioned, I do a lot of this retail-related work, and this area in Northwest Arkansas, this is probably the dream place for me to be here. There are a lot of these professionals who are in various parts of consumer packaged goods, retailing, supply chain, and those are the topics I generally focus on for my research purpose.

02:05 Matt Waller: When we hired you, one of the reasons I was really a champion of hiring you is because I thought, you know, in Northwest Arkansas where we've got so much data, nobody's using it, and so I thought, "Gosh, we need to build strength in this area." So Dinesh what is marketing science?

02:25 Dinesh Gauri: Yeah, so marketing science in a very simplistic way is basically solving marketing problems using some data science tools and techniques. Now, data science is a very, very broad term, so I'll just specify that it includes econometric models, mathematical models, computer science models. So when we take these marketing problems of interest to either the manufacturer or the retailer or the consumers, and we use these data science tools and techniques, that essentially becomes marketing science.

03:04 Matt Waller: Dinesh, I know you teach both at the undergraduate and the graduate level. When you teach at the MBA level, what kinds of marketing science techniques do you teach to the class?

03:17 Dinesh Gauri: Yeah, so I teach or have taught courses like marketing analytics, brand management, and then retail strategies, what I've been doing for a few years now. And then introduction to management or introduction to marketing. So in terms of data science skills, and we go through things like choice modeling, so different type of logit, appropriate or multivariate components of these models, conjoint analysis, time series analysis, logistic regression, factor analysis, discriminant analysis, some sort of forecasting of demand. So these are some of the things that we go through in the classes with different assignments. And I provide some data to the students also to run it because I realize that many of these companies that they work in, even though they have access to a lot of data, which they should be running some of their own models, but the companies do not want that data to be shared. So I provide them with some data with my assignments so that they can run these different models.

04:32 Matt Waller: How about at the undergraduate level?

04:34 Dinesh Gauri: Yeah, so undergrad level, we do not go in that much depth, but at the same time, we go through some of the basics of what you can do, some statistics, some regression, basics of choice modeling. Also, we focus on things like factor or cluster analysis, and then again, the basics of the forecasting of demand and things like that. So we do the same things, but probably not in so much depth. But some students, I get surprised all the time. There are some students in the undergrad class also who want to know more, and then I encourage them to either take some more classes or to do some graduate program, or even a PhD program, for that matter, if they are interested in going deeper into these techniques and models.

05:26 Matt Waller: In the past, we had marketing science, I mean, marketing science has been around for a long time, but we didn't have the data. So a lot of models were being created that couldn't even be used. Do you think that this proliferation of data bodes well for the future of marketing science?

05:47 Dinesh Gauri: So a lot of the data has been generated only in the last few years. So there was a famous saying, which IBM CEO Ginni Rometty came up with a few years back, and she said that 80% of the world's data was created in the last five years. Now that saying has been also like I think she said this in five or six years back. But I was doing some search and I found out that there are about 2.7 zettabytes of data which exists now in the digital universe. And Matt, can you estimate how many zeros is in zettabytes after one?

06:28 Matt Waller: [chuckle] I have no idea.

06:30 Dinesh Gauri: [chuckle] And so I had no idea too, so I looked it up and there were 21 zeroes after one. So 2.7 zettabytes which exists right now, and by 2025, they estimate that about 0.46 zettabytes of data will be created every day globally. And how much is it like today, one day, like 500 million tweets are sent per day, about 295 billion emails are sent one day, five billion searches on internet are made in a day, and now since we are in Northwest Arkansas, where Walmart is headquartered, so about 218 million weekly customers are what Walmart entertains every day around the world. So just to put things in perspective. So there is a lot of data which is being generated not only in the United States, but around the world now.

07:31 Matt Waller: Continuing with this thought of, what is marketing science, what are some of the kinds of questions that you can answer with marketing science tools?

07:43 Dinesh Gauri: Yeah, so I'll give you a few examples. So a few years back, I did a project where, let's say, a company like P&G, they have about, I think, 20 plus billion dollar plus brands. And let's say they have this Tide brand, which we focused on, and that's their primary brand, but they also have these other brands like Downy or Febreze, which are in a different product category. And let's say they wanted to come up with a new product called Tide with Downy or Tide with Febreze, which is exactly what they came up with. And we wanted to study things like what happens after the introduction of the product, like who buys that product, which brand benefits the most, whether it's the loyal users of the brand would just buy this new product or whether this new brand is attracting some new shoppers which are new to that category. So which is ideally what the company would like. So we used various marketing science methods like choice modeling, we had some data on before the time which the product was introduced and after the time the product was introduced, and then we ran some of these marketing science-based choice models to answer these questions.

09:05 Dinesh Gauri: I'll give you one more example from the practice side of things. So shopper marketing is a very growing field in both research as well as practice. And these shopper marketing managers, as well as the chief marketing officers of any company, would like to know what is the return on investments on various programs that they are running to attract the shoppers. So they do things like special displays, they use influencers, they use social media, they have some retailer events which they run at different retailers. So they would like to know how much is the incremental sales due to these shopper marketing programs. Now getting to return on investment requires building some mathematical models, essentially, which are modeling the incremental sales. If you get these incremental sales to be higher than the spend that you're making in that program, then your, of course, return on investment is more than one, which is what they would want, otherwise, they are losing money in that program and not getting enough incremental sales.

10:16 Matt Waller: I think a lot of times, practitioners, whether they be in marketing or sales or executive leadership positions, I mean, I would think executive leadership positions could utilize marketing science analyses and input, but my guess is that many executives aren't even aware that this exists. They probably think, "Oh, we need people who have good data science skills," but just because you can manipulate data around, it doesn't mean you really utilize marketing science. Why would you need someone specifically with a marketing science education or background to solve certain kinds of questions in business versus just someone who's just good at computer science or math?

11:08 Dinesh Gauri: Yeah, that's a very good question, Matt. So what has come about is this, that you can find a lot of statisticians and folks who can run the model, but if you do not find people who can interpret the results of the model and come up with some type of implications for the practice, "Okay, that if I find this return on investment, then what does it actually mean, or how can I shift my dollars from one activity to other activity?" So you need some storytelling ability, you need some ability to derive insights out of it, and that's where I think a person which has a combination of both these modeling skills or at least understanding what the model is coming, as well as interpreting the results, would be the full marketing science person. Just running the model and getting some of the results will not help much.

12:08 Matt Waller: Yeah, and I would imagine if you had someone that was just strong quantitatively in a marketing science position, they may not be as familiar with the terminology of marketing, but they also might not see opportunity. So senior executive might ask for some kind of analysis and someone that has a marketing science background, they'll be able to do the analysis, but they might even be able to go a step further and say, "You might also wanna look at these things," and they know that from their marketing understanding, not just their quantitative skills.

12:45 Dinesh Gauri: Yeah. So you need a combination of these different skills to be able to come together, and most of us are good at one or two things, we are not good at everything, so that's why you need different types of people and a cross-disciplinary or a cross-functional approach, in this case. Now, you need some long-term vision also for this, because many of these models and things cannot be just run or implemented like a switch or a button. So you need to build up that skill in your team in the long run, so that when you have the data, and when you get some of these people who can run the models, then you can see that what comes out of it, that you can implement or not. So it's both an art and a science. And as I say in my classes to the students that models do not do miracles, right? So models are as good as the data that we are feeding them into. Now in present world that we are living, where about this zettabytes of data which exist, and about 80% to 90% of it is unstructured data. So now you cannot put that unstructured data directly into any type of model. So you need...

14:04 Matt Waller: Now, let me ask you a question real quickly?

14:06 Dinesh Gauri: Yeah.

14:06 Matt Waller: I think some of our listeners won't understand what unstructured data is, would you mind explaining that?

14:13 Dinesh Gauri: Yeah, so unstructured data is things like reviews or text, which people write, or the post that they do on various social media and other platforms or the videos that we watch and upload, the images that we, again, watch and upload in different platforms. A user-generated data, which has not been curated in any shape or form in many of these social media platforms and other platforms around the world and different websites, so that all is unstructured. Now, whereas structured would be where you have your point of sale information, which is coming from the register and you know that somebody came and they bought two units and they paid X amount of dollars, and whether they came again or not, so that is more like a structured data which a lot of companies are used to seeing, but this unstructured data which exists in so much abundance around is something really new. So that's why I was saying that you cannot just wake up one day and say, "Okay, I have now these five terabytes of data and then just run it into a model and tell me what the data is telling me." So you will have to build up expertise and skills, you will have to put some structure to it and then put up into some of these machine learning and other models to be able to give you some insights.

15:42 Matt Waller: Dinesh, I know for both research and teaching, you need data. What kind of data sources do we have available to us in the Walton College that you could use in marketing science?

15:53 Dinesh Gauri: Yeah, so we have access to both structured as well as unstructured data for both faculty as well as students to work on for, teaching as well as research purposes. So for unstructured data, we have access to a global database of a company called Brandwatch and for structured data, we have access to Nielsen's scanner and panel data, Kantar advertising data set, and we also have point of sale data from a major retail store, which we cannot name for the purpose of confidentiality, but it's a very good data set that we have. And amongst these unstructured and structured data, this is merely the best that anybody can get. And we also have relationships now with various companies and they are willing to do some experiments if some of the researchers are able to come up with some, let's say, things that they want to implement. So the companies in the area are willing to do it that, "Okay, we are going to do this experiment now and get the data and see what happens before and after the experiment."

17:08 Matt Waller: So you were the first hire in the marketing science area in our marketing department. Do you think that the future of marketing science is bright? What are your expectations about the future?

17:25 Dinesh Gauri: Oh, yeah, it is certainly bright. So if you think about any of these companies now we are living in this during pandemic world, I cannot call it post yet, but many of this work at home and other possible things that we have been able to do is because of the technology companies and think about the data that each one of these companies is generating or being able to help us with. So companies have now troves and troves of data at their disposal, and because of some competitive nature of the business, they are not comfortable sharing it with others. But then what they do is they hire people directly into their setting, into their company, and then they tell them that, "Okay, now what else can we do?" So I think future is really, really bright for students to be able to know these skills, and then there is no dearth of companies. Even now I see, there are a lot of these job postings in our local area, which keep coming off either directly at Walmart or some of these suppliers who deal with Walmart and they have the openings in shopper marketing, they have openings in digital commerce, e-commerce, data analytics. So there is a lot of requirement for these people in the area now and it will continue to be there in the foreseeable future.

18:54 Matt Waller: Dinesh, thank you so much for joining the Walton College a few years ago and building our capacity in this really important area. I think many people in our region know data analytics is important, they don't realize the opportunity in marketing science because there's just many people aren't aware of it. So I appreciate you coming here and helping to build that capacity, one, our students need to be learning it, we need research, it's relevant in this area, but also for your network and building this part of our organization. Really appreciate it.

19:32 Dinesh Gauri: Thank you very much, Matt. I have been very excited and honored to be here and thank you for all your efforts and making things happen. And I know that many folks that I interact with, they say that we didn't know that Walton College had these courses, but they are excited about the possibilities, and we get a lot of positive feedback from different benefactors and people all the time. So thank you for having me and thank you for your efforts too.

20:03 Matt Waller: Thanks for listening to today's episode of the Be EPIC Podcast from the Walton College. You can find us on Google, SoundCloud, iTunes, or look for us wherever you find your podcasts. Be sure to subscribe and rate us. You can find current and past episodes by searching beepic podcast, one word, that's B-E-E-P-I-C podcast. And now: Be EPIC.