Nikhil Varma (HEC Montreal) PhD Thesis Defense Presentation

Welcome everybody. Welcome to this thesis defense by. I'm the president of the jury. My name is Jacques, professor with the Department of Logistics and Operations Management. And I'll introduce the other juror members. Professor Goit Verma from the Cornell University Business School. Or should I say Cornell Business or it's brand new. Welcome. Anyway, so he had a nice drive from New York, enjoying the nice colors. So we're very happy to have him welcome the other two members. He did a couple of you all know Claudia de Bolido and Gillette Paquette from our department. So we're going to have about 45 minutes to make the presentation. Then we'll have questions, questions from the jury members. Oh, and I forgot to mention the representative from the school there a couple of us seats who is making sure.

That we do everything properly. Okay, thank you.

So welcome everybody and the floor is yours.

Good afternoon. My name is Nikhil Varma. I've already been introduced. So I would firstly like to start by thanking Professor Verma to take the time to read my thesis and my external examiner. I really appreciate the effort that you put in. Thank you so much, very much. But I would also like to thank all the members of my jury much for coordinating all this for directors and thank you so much for being a member of my jury. I. I can't thank my record directors list and I think this is the moment to thank you all for all the help, all the support, all the guidance that I received from you. I would not be here if. If it was not you supporting me on this time. Thank you so much for that. At the same time I would like to thank my department, I call it my department because I really feel it is my department for helping me out, for every little thing that I needed. Right from guidance, right from support. In fact, my department also made sure that I look good. There is. Thank you so much. Thank you very much for all this and thank you everyone for all the others for coming in to support me. And it really means a lot. It's a big day for me. But at this point of time I'd like to thank one person in this room who has been with me throughout this whole journey. That's my wife who's right there. It's difficult to be a wife of a PhD and I think many people know that. And she's supported me throughout. There have been four big events. I was just trying to recollect what are the big events in my life. This is the fourth one. The first one was when I got married to her. The other two was when my two beautiful children were born, Bir and Alia. And today is the fourth event. And I hope that the fourth event also goes the same way as the other three. And with this note, I would like to start my presentation by giving you an overview. Of course, I will be discussing much more than just a few bullet points, but I would introduce the core idea of my research. I will talk about my research focus. Then I would introduce the three different essays that I have, which essentially is the title of my topic, and how they build from one to the other, and eventually conclude with the overall picture of what I have achieved in this research path so far. So when I was starting my research, I looked at a model which is commonly known as production consumption. This is the traditional model where what used to happen is that the manufacturer would manufacture a product and the customer would buy the product. Over here we have a hot water tank. So a customer would buy a hot water tank from the manufacturer and pay money to the manufacturer. And in this process, manufacturer would do nothing and the customer was responsible, responsible to install the product, maintain the product, as well as dispose. Now, this was a very transactional oriented kind of a mindset where the manufacturer was just interested in making money selling a product to the customer, and then the customer would get stuck with it. Of course, there were a few other things like maintenance and warranty, that was still well known at that time. But overall the picture was more or less like this. But then people thought of a new business model, innovation. Instead of putting the product into the customer ecosystem, why not share the product? Why not have the product in the customer ecosystem, but not quite so the manager. So the manufacturer would still have the product, but would take care of all the functionalities. So disposing of the product, maintenance of the product, installation of the product. But then how would the manufacturer get money? So you create something which is a value to the customer. And the value is, I will take care of your hot water needs. And to take care of your hot water needs, you pay me money. Now, this is in very simple terms, explaining the core idea of what servitization is, how servitization is set up. Of course there is more to it, but if I have to explain, like what is servitization, I would kind of allude to this example that I just presented. Now, servitization over the years has taken a lot of transformations. So it's not to say that servitization did not exist before 1988, but the term servitization was Coined by Vanderbilt and Radha in 1988. So they talked about adding services to products to gain competitive advantage. That was the whole starting point of servitization. Then there came an article in MIT Sloan from Mohrbir Sani in 2004 which talked about temporal and spatial changes. Now this directly did not refer to the word servitization, but it talked about expansion on the service side. So what it what the most. The biggest. The idea that came out of this article was that there is a product, there are some services which are alongside the product. So these are the temporal services. But you can expand the horizons by adding more services which would be complementary to the product. So they talked about spatial changes into the product setup. After that there was much more transition or much more expansion to the definition of servitization, where it was where customers came into play for the first time. So far it was talking about gaining competitive advantage. So now it is slowly converging to the idea of value. So you talk about gaining competitive advantage, satisfying customer, as well as talking about performance. So there are a lot of things that came in this particular definition. But what I consider to be one of the latest developments or definitions of summitization is by Andy Neely. And what Andy Neely talks about is that servitization is not a one time kind of a change, which actually has been the core founding principles of most transitions in organization theory. So what he is talking is it's a journey. And then he brings in the concept of capabilities within the organization. So you have to go through this journey by building capabilities in the organization. And of course he talks about service alongside the products, which has been the core idea. There is another little thing that I want to show is that of course I said my department of Operations management. But servitization is not restricted to one department. Serbitisation has a multidisciplinary approach. It has to have all the departments taken into account. We've been talking about that in management for a long, long time. But when we talk about building capabilities, when we talk about customer satisfaction, we have to make sure that we take into account the whole organizational dimension. And that is why sometimes research in servitization is a bit difficult because. Because you have to take into account a lot of dimensions. Also I showed you the definition. I just want to show you on a timeline how servitization has kind of evolved. So we, back in the days prior to 1988, we mostly used to have a product driven economy. So everything was product. You want to have something tangible, you Pay for it. And that's essentially how everything was set up. But then the core concepts of servitization was mobilized in 1988 and for the last, for 20 years after that, there were various things that came into play. Service infusion, solution driven mindset, sustainability, supporting products. So there are many keywords that came into play over the 20 year Spanish. But in 2008 there is an interesting article that came up and this article was not actually bashing servitization, but questioning servitization or reporting something about servitization. And this is by Andy Neely and he reported servitization paradox. And when he studied small and big organizations, he kind of found, found out a very intriguing thing, that big organizations actually tend to get bankrupt because of servitization. And that was a very interesting thing and that has taken a lot of research. And I would also try to address in a very small form in one of my essays. But moving on from the subitization paradox then we have been talking more on the value. So service dominant logic, which is coming from the marketing domain has spurned, and I think it is much more appropriate to call it value dominant logic, where essentially we talk about the value side of things. And that is what is the next few things that servitization is taking into account. And I hope to be adding one keyboard sometime down the time in my life in this area. This is my area of interest now coming back. So so far I was explaining about what subitization is, what this domain is. But the whole purpose of us coming in here today is to look at the thesis. So the idea of what I want to present the three different essays is a build up and they are linked in some way or the other. So the first piece of work which I do is when I set out to look at servitization literature. It was very interesting, but there were still many things that seem to be undeveloped, underdeveloped. So to a researcher who wants to start in servitization, just to answer the question, what is the current state, state of art in servitization? What are the theories in servitization? Where are the theories coming from? That was my motivation for the first essay. Now the second essay, although we don't call it an antecedent model and I'll explain why, but we are trying to look at what are the organizational parameters which are influencing servitization. To be able to understand organizational parameters, we looked at lot of theory and the theory was very confusing. So we came up with the Typology of how to analyze it. And I will present that as well. And so far, these two parts were mostly looking at the organization, looking at the servitization phenomena. The third part is answering the question how? And I will elaborate on the third part, which essentially talks about having some kind of an adaptive delivery to enhance the customer experience, which is the customer value, so the customer derives that value from that experience. And how would you have some kind of an adaptive delivery process for that? So now drilling into each essay in far more detail. So as I mentioned, my essay one essentially has all, I would not say all, but most of the literature that we could find through an extensive search in standardization. And the idea was to answer three key questions. What are the trends of trends in literature? What are the research avenues that we could pursue? And how is the theory or how is how is the theory developing? Or how is the theory evolving from all this literature that we are looking at? Now, to do this, for the first two things, which is trends and research avenues, we conducted a content analysis. And I will be elaborating on how this content analysis is done was done rather. And the research avenues and theory evolution used different kind of a mindset where we use an iterative approach to understand what are the evolving theory theoretical constructs. Now, before getting into the details of what is how all this is done, I would like to present what was the contributions of this whole research. So, as I mentioned, there are a lot of starting points in servitization, but it is still very unclear. So one of the first few researches, which was done in 2004 by Baines and Lightfort, Benedictini MK they had. Sorry, they presented. They looked at the articles that were present that were the state of the art or whatever they could find at that time. There were about 58 articles that they pulled out and they looked at what were the key findings. So it was more of an exploratory kind of research. But they were trying to organize things from that research where they found or they presented some research direction and they were trying to focus on, on why companies servitize. So they were trying to probably that was one of the research questions, to look into all these articles. But it gives a fair amount of research direction. But there was another systematic literature review done a few years later which looked at a different kind of model, which was looking at how is the knowledge diffusing? What are the themes that are being built out? How is servitization being set up? What, what are the different groups that are formed and Then they also discussed that these different groups of servitization seem to be going off their own path. They should collaborate more to be able to help the development of theory in the servitization. They they looked at 148 articles and then one of the recent work from Turman, which looked at strategic advantage from service infusion, which is also a form of summitization, where they looked at articles only, which we talked about strategic advantage. They looked at 53 articles and found that there was a lack of rigor. So they didn't find a lot of constructs that would link the strategic management to the service infusion theory. So this was what we kind of looked at as the current state. And this is our sales pitch. So what we have is about 166 articles. Now this 166 articles came out of nowhere, came out of somewhere rather, where we had 424 articles which were pulled out from a very extensive search. And after looking through each one of these 424 articles, we finally selected 166 articles based on a particular rationale, wherein we saw that there was an end to end discussion that was covered. So it was important that if it was just pure service, it is interesting, but we wanted to study the whole phenomenon. We want to see if the like, you know, there is some kind of a strategic definition, there's some kind of an organizational definition, there's some kind of a measure which talks about what is the effect of servitization. So that is where we used an operational management sentence to select the articles. There was a model that we used, and I will refer to that model. We came up with a lot of trends, we came up with a lot of future research directions and avenues. And all this we believe could be a good starting point for someone who wants to go further into research in standardization. So that is the whole idea of our work. So when we were looking at how to code this. So so far there are various approaches to coding literature. Systematic literature review has a lot of directions. We looked at lot of articles in IJMR as to how we should actually start our coding process. And since we were looking at a very holistic approach, like from an end to end approach, we used the contingency theory. Now contingency theory essentially talks about being in sync with the environment. If you are in sync, if the organization parameters are in sync with the environment, your performance would be better. So we used all these parameters to be able to build our coding boxes. So essentially we had coding boxes like coding blocks, which had external environment, which had organizational strategy, which had organizational factors. And then we have outcome measures. So this is how we build the process. Now, after having identified this coding framework, it was very difficult to identify what to code here. So on what basis would we initiate our coding process? So we developed an inductive approach to be able to evolve our codes. So we started with some kind of a base code that came out from organizational change literature and service literature. So suppose we are looking at outcome measures. So we looked at outcome measures. What are the different outcome measures which service literature and outcome organizational change literature are talking about? And during our coding process, when we were looking at outcome measures, when we got or encountered new measures which were very different from our set, we either added them or we refined our base measure. And if I'm not making much sense with what I just said, I will show you an example. So over here, these are the organizational factors that we started with. So based on organizational change literature and service literature, we came up with five different initial codes. Now, these five different initial codes, we started coding the articles. And while we were coding these articles, we encountered a lot more new codes. And this is how the evolution process actually happened. You have all the parameters in white, which we started with. And so I'll give you one example, which is organizational knowledge. So while we were coding for organizational knowledge, we saw that there are two different things which people are talking about organizational knowledge. And these two seems to be, you know, it's a good idea to differentiate between them. So we separated experience and innovation. And for all the codes that all the articles that were coded with organization knowledge code were revisited with experience and innovation to the coding. So it was a very iterative, rigorous approach that we adopted for our coding process. Now, besides the key codes that came out from the contingency theory model, we also had some contextual parameters. And these contextual parameters were. There were a lot of thought put in to be able to have some kind of, you know, a trend that we could identify. Now, you would see that the third group is not very symmetric as compared to the first two. And the reason is that this study was conducted in 2014, and we did not have that many years, but we split the articles into these three periods. We also coded for the name of the journal, all the journals that had their name change. Also we found a few of them, so we had to go through the process of even correcting that. We coded for the research methodology, which this article was, the article that was quoted, and we also coded for the service type. Now, I Just want to take one moment to explain what is tightly coupled and loosely coupled. Now this is flowing in from Mohanbir Sani's temporal and spatial expansion model. Over here we are calling it tightly coupled and loosely coupled because we feel that that kind of experience is much better. Now, tightly coupled services are services which are tightly coupled to the product, essentially which are just adding value around the temporal line. Now, loosely coupled services which are services which would be giving a complementary support to the product. So that was the whole idea of looking at, okay, what is the kind of service that they were having? And this was the only. Actually all of them were mutually exclusive to each other. Of course, from this particular set that we had, by the way, I just want to also mention that the organizational parameters and the other parameters that I just discussed, except for the contextual ones, there were some articles which had like more than one. So when you see 35 or 36 number, it doesn't mean that there were 30 like or it shows that it was just that particular aspect in that article. So that takes me to the research dashboards which would present or highlight the key findings. So the first finding that we had was that over the years we've seen a huge increase in the number of articles in servitization. So surveitization is becoming important. That is the core interpretation that we could have. And I think we were in the right track. So this was a good enough finding. But when we look at the number of journals, to look at how this whole phenomena is diffusing across different journals or acceptance level of journals, we also saw that the number of journals, of course every year, that is a moving target. The number of journals also increase in the, in the macro environment. But that is still being said. We could see a lot more journals were receptive to the servitization articles. When we looked at a few key journals which published servitization articles, we saw a lot of different names coming up. So we see service, we see marketing, we see manufacturing. So we can see that there is. It's not that it is just diffusing in one area. Even if when we selected articles only from an end to end perspective. So Journal of Market Marketing probably would not feature as much. But even then we could see that it is still showing a lot of different domains. And that's very interesting. When we looked at the methodology, and this is where one of the motivations for the second essay, we saw that case studies are interesting methodology. So don't get me wrong when I say that only case study is not good. But what is happening over here is that we are seeing a lot of case studies, but less confirmatory research. Now less confirmatory research, to my understanding, is that maybe the theories are not well developed. So that is why we have lot of exploratory. We have lot of studies which are going and studying the phenomena, but there is no hypothesis. Not many hypothesis testing was done in this area. So this was one of the motivations for the second one when we looked at the strategic focus. Now I think I did not highlight this in the previous slide, which we also categorized the articles based on the strategic focus. So the three strategies that we. And that is why we call it task typology, because we define them ourselves. So organizations which were driven by profits or had a self interest are called producer centric. Organizations which are driven by customer needs are called customer centric. And organizations which collaborate with customers are called ID in our knowledge. Now what we see is that we have a high number of producer centric articles. So that could also be a bias on our side. So this is a finding, but we have to take it with caution because our article selection was mostly from an operation management perspective. When we look at the external factors and we carefully define these factors, we look at a very reactive approach in servitization literature. So servitization literature is not exactly looking at proactive measures. And when we looked at the service types, we see that tightly coupled is mostly featured, which is maintenance, aftercare kind of thing, but not loosely coupled. So this is a few interesting things that came out from a trend perspective. When we look at the organization factor, we see more mostly organizational structure, hr. These are the things which are figuring. And when we look at the outcome measures, we see mostly financial investment measures which are kind of easier to comprehend or study are being done. Customer loyalty, customer satisfaction. Again, that could be a bias based on our article selection. But just to kind of summarize about the implications, what we were able to capture with this summative analysis is that there is a diffusion of servitization research. So servidization is there and it is happening. The next thing that we found out was that, you know, there is a lot of proactive approaches that need to be done. We have to do a few confirmatory things that have to keep going. The organizations are still looking at loosely coupled services. There is a lack of research in losing couple services. Sorry. And of course this is a standard prescription. We do not see any kind of a scorecard measure. Now so far we had the articles under control. We did the summative analysis. So we were like, this is just a trend, why not look at the theory? And that was the motivation for us to do a follow up analysis where we wanted to look at what are the theories and how are the theories of servitization evolving. And this is why we conducted this follow up research where we used a process based model, which is proposed by Ann Langley, which analyzes a transition process transition based on the process view. And we looked into the three dimensions with producer, centric, customers centric and dyadic as the three strategic pieces. Essentially what we did is we had pre servitization, during servitization, post servitization. We looked at different discussion constructs, theoretical constructs or descriptive data that came out from the discussion article. And we looked at for all the different strategies that we had to be able to derive the evolving theory from the theoretical constructs. We use the grounded theory approach and the grounded theory approach has various schools of thought. We use the Goya method, which is iteratively building into the next level of abstraction and finally building constructs out of it. So there is a first order theme which follows an aggregate dimension. And then we have a generation of proposition. So this is one little example of how all these things pan out. So if you see the first order constructs, these are basically discussions from the articles. So this is written in our own words, but there are several references because of space limitations. Also sometimes we kind of highlighted just the main article. But this is all done in NVivo with multiple articles referring to one mindset. And based on the first order construction, we kind of abstracted it to the second order theme. And the second order theme was then eventually abstracted to an aggregate dimension. So just to zoom into one of these, this is two discussions which talk about customer experience and communication. And we have the aggregate theme built up as relationship management for value delivery. And based on these discussions, we come up with a proposition that organizations should have internal and external open lines of communication to reduce risk of subdivision. So this is the second part which where we developed 29 propositions. And these 29 propositions could be a good starting point for confirmatory testing. So essentially what we claim is our research develops is a typology to analyze certain with organization strategy. We have a process based approach which is slightly different from many analysis. And we have developed propositions which could be used in further for confirmatory testing in future. Now moving ahead of SA1, so what after having looked at the theory, after having looked at the evolving, after having looked at the trends and the evolving theory and derived a few, deriving a few. It was, what do we do next? So the next thing that came was that we see a lack of confirmatory research. And that was one of the findings. And based on the lack of confirmatory research, and also we found out that there are three strategic orientations. And these three strategic orientations, based on our evolving theory research, we could see that they are actually different from each other. So what we wanted to do is to have some kind of a conformatory testing. So our. One of our motivations that we saw was that the impact of influencers on servitization is really confusing individuals. And this is one of the examples. So this is an article from Gibor which talks about adding services actually increases sales or profit margins. But at the same time, as I already discussed about this article, where Andy Neely talks about servitization leads to bankruptcy. Now there are two different perspectives for a same kind of a mindset. So how is financial related to servitization intensity? So this led us to get into looking at different influencers. And this is how most people are studying it. So this is an influencer. This is the servitization. Now what we are proposing is that the organization strategy is having an influence on this relationship. And how is this organizational strategy having an influence on this relationship? You know, only confirmatory testing can help us do that. So essentially what we are claiming over here is that this is our research contribution, the current research that we see where, you know, there are some articles which are mostly in theoretical nature, like they are talking about managerial motivation as an antecedent for service orientation, where they have one parameter which is managerial motivation and they look at different effects on a theoretical basis. Again, there is an article which talks about reverse servitization path and links financial to servitization with two case studies. And you know, the third article also talks about how servitization. These are few articles which we have currently in the literature which are very case study based. Now, case study based is definitely a good starting point, but they are not confirmatory testing. So what we are bringing into play over here is that we are partitioning or we are developing a typology. We actually have developed a typology based on that typology. We are partitioning the data and you will see that all of us are sudden there is a new strategy evolving that has come into play here. And this is just because we had a two by two matrix and we have producer centric, customer centric DYADIC So the other cell that did not have any of these, we called it an evolving strategy. And we use structure equation modeling to do this testing. And we see or we try to assert that strategic orientation can explain a few of these. So when we set out to do this work, and this is where Claudia was a great support, the data for this research is very difficult to collect. And we had to ideally develop the surveillance instruments and conduct a survey that would probably be 2018 when I would be defending my thesis. So it would have been very difficult. So we had. And Claudio was deeply involved in imss. So this is a manufacturing strategy survey. And they introduced one whole section on shifting manufacturing towards services. And this was the opportunity for us to be able to use this survey data to be able to do this testing. Now, of course, we all know that a survey which has already been conducted has limitations. We don't have everything that we have, but to the best of our abilities, we try to identify all important influences in the data. If there are fewer influences, which might seem to be interesting, but are not, there is a limitation of the survey. This is the best people do. And we also selectively picked up OECD countries only. And few studies showed that, you know, mostly servitization literature, also diffusion from a geographical perspective that oecon, non OECD countries, countries don't have the same kind of a mindset or the same kind of an approach to servitization. So just to be on the safer side, we picked up OECD countries only and we identified six different influencers. Now this is a cross sectional survey, so we cannot talk about causality over here. So that is why we are mostly studying the correlation between the two. And we are saying that these six parameters are having some effect on summitization intensity. So these are the latent variables that we built from the questionnaire. But we do have the two by two matrix which presents the theoretical orientation, strategic orientation of the organization. So we basically partition the data and I will show you the method in the next slide. And then studying the seritization intensity. So the research is basically set up with the data that comes from the 600 plants. So these 600 plants were hand picked from OECD countries only. We classify based on some identification that kind of matches our typological definition on which theoretical or which strategic block these plants actually are. And then finally build a structure equation model to be able to test the relationship of these six latent variables or the influences that we call here on servitization intensity. This is all done in the EQS tool. And of course this is testing the hypothesis. Following this research results, we found very interesting insights. So I would give you an overview of the insights that we have. So this is on the left hand side is all the strategic orientation. And when we look at financial performance, we see financial performance. So all the reds show that a negative correlation. The green show a positive correlation. The gray ones show that there is no relationship that we could significantly find. And we could see that we two see some kind of a variance happening here for each one of these parameters may be for some it has a common effect. But for our point of interest, which was financial performance with which I started my presentation, you see that producer centric and customer centric have a negative and a positive relationship. Now just to kind of refer back to the two articles which the first one which talked about organizations gain from servitization, the second one talking about organization lose from servitization, we don't know, maybe they are talking about producer centric and customer centric strategies and that is why they are able to see they are seeing these kind of things. So it is important for managers to take into account what is the organizational strategy before making any investment in servitization. And this also links to the definition of MELE which talks about firm should develop capabilities. Now developing capabilities also has to have the strategic orientation. So they seem to be all linked with each other. So so far we've been looking at. So essay one essentially talked about what is servitization now? What are the the things in servitization? What are the involving theories in servitization? Sa2 with the influencers presented why servitization, what are why an organization behaves differently in servitization? What are the different strategic orientation? So now we would like to change the context slightly and based on all this immense data that we have have you would like to present the how. So sometimes it appears that there is a huge disconnect between the first two essays and third essay. And so this is why I would like to reiterate that the first two essays were focusing on the organizational side of things. So what is the organization and what are the different things in the organization? But now when we have to change the context to how we have to bring in a very important person and that important person is the customer. So far we've been not looking at the customer. And when we have to bring in the customer, we have to study how the customer interacts with the system. So that is why we Bring in a multi interaction model and we try to predict the customer expectation by modifying the service. So this is the whole idea. And we want to enhance the customer customer experience and generates customer loyalty and profits. And this is the whole idea of my third essay where we use behavioral analytics to assist in service design. So the core research focus is that using behavioral analytics, I believe currently with big data and so many things happening around us is coming in from the digital space. So how can the digital space help us so that over a period of time we can keep that customer smiling? So we want to make sure that the customer keeps smiling over a period of time and we are going to use digital data for that. But there are various schools of thought on that. There are some people who say, continue to do what you're doing. If you are good, you don't have to change. However, I don't believe in that. So we kind of have two different approaches. So there is one approach which uses a statistical approach, a linear regression or some kind of a model where you kind of evolve with different interactions with the customer and based on that you make some kind of decisions or you use artificially trained system to help make, you know, support the customer much better. So coming to the research contribution and this is one of the articles from our sterling examiner is talking about service sequencing, which is a very interesting concept which talks about peak events and how we should should space the peak events from the end events and how that has an impact on customer satisfaction. Now this was a big motivation for us because we wanted to space things, but our spacing strategy is slightly different. But the idea is more or less the same. And then there was a piece which was built up very recently. That's why I don't have this in my article. Also in my thesis also is talking about previous experience. So building on the sequencing, but also the result of previous. So there is adding another dimension to it. So what we intend to do is instead of having a boilerplate kind of a solution, we are trying to be very customer specific. So every customer is different, every customer has different needs. How can we be different for each customer? And we use or we modify monitor. That's not the right word. But we use the behavioral data that we get from the customer interaction and based on that we adapt and modify the service. We modify different things in the service. And this data also is used to train neural networks which actually gives us what could be the best fit for this customer. And we claim that of course we have more accurate predictions and we will demonstrate that. So the research method. So the first thing, if you remember the slide two slides back was do nothing statistical methods or using artificial networks to help you. So we wanted to first compare between regression method and artificial network neural network method to see which one has a better predictable power. So this was the first step where we collect survey data. And I'll explain how we collected the survey data method. It's a standard process. We separate it into training and test data and we then feed in the training and test data into two different things at the same time concurrently. So one is the linear regression model. So we train the linear inertial model with the, we build the linear regression model with the training data and we test it on the test data to see the accuracy. And likewise we do with the artificial neural network and then we compare the results and then we say, okay, this is our best method. So this is the first part of the research. Now this is not helping the customer in any way. This is just to demonstrate that we are using a pattern method. Now the second part is where we use the better of the two methods to assist in dynamic service design. This is where we again take all the data from the customer. We split into training and test data and we use the training data to predict the weights and ranks and, and I will explain what am I doing, what do I mean by weights and ranks in a minute. And a recommendation service. And then we use that on the test data to show how accurate we are. Now this might seem like, you know, you have the data, you're using the training and test data. So essentially what we are doing over here is that by splitting the data into training and test data, we are demonstrating our forecasting ability for a new customer would come. So assume we have a new customer. That new customer is our test data to what level of accuracy we can predict. To collect the data we had a survey setup. Our survey was set up for a car purchase experiment. We have a low price car and a high price car setup. A low price car has Honda Civic and Toyota Corolla. And the high priced car has a BMW and an Audi. And what we these were explicitly stated at the beginning. So that is why we call it an experiment. And we have five sections in our experiment tool. So the first section is to collect the demographic data and we very much use this demographic data graphic data in our model to be able to see and we see a few of them actually make have an impact. Of course in neural network we don't get to see that they did a few of them didn't come out in the sensitivity analysis. So see that they were doing good, but they did have effect in the linear regression model. So the second part was that. So the first thing we collected the demographic data. The second part is where, where we actually present a search engine to the person. So this is your search engine. And then ask questions as to what would you use to search and also ask questions on whether you would like to have more options. So these are different inputs and these are all collected on a seven scale, likewise scale. So we know what are the different needs of the user. The third thing is post the first search for cars and we give them a page with assume that you get these four cars as your search option. Are you satisfied? And you know, so we collect the satisfaction matrix. And this is a very traditional way of collecting three measures to merge them together and finally measure satisfaction. So that was the third section. In our fourth section is where we add the service. And over here we introduced four different services. Maintenance, which is a traditional service. Insurance financing and a very interesting service which we hypothetically created, which talk talks about a service which would create their manage their carbon footprint. Okay, would you like. And we would give them the price and everything for each one of these services and then kind of ask them to rank which service they would want. What is more important, suppose you have hundred dollars. How would you split between these four? So kind of getting multi dimensions in our approach. Once that was done, finally in the last stage we would collect their. Oh yeah, I'm sorry. After that we would give them a recommendation service also. So they had a search parameter. And then we would say assume that we would recommend a new service, a new car to you which is maybe 10 miles more than what you had set as your tolerable distance. Would you be interested in this car? Or maybe $100 more? Would you be interested in this car? And then kind of asking them what recommendation? Of course, recommendation is an area of research on its own. We don't mean to be getting into that spin, but we wanted to see how they respond to this recommendation and eventually the overall satisfaction. So one is assert satisfaction, one is acceptance of recommendation, and one is overall satisfaction after having done this. These are the results that we got from that particular data that we had. So when we were by the way, for the first one to have the predictability power, we used the overall satisfaction, the satisfaction that was collected at in section 5 with all the parameters that we had. And we saw that for the low price bar and the high high class linear regression, These are the RMSes that we get, which were, which were pretty good actually. But as compared to artificial neural networks we were it's a multiple back propagation train network and we used open source tool to simulate this. And we got, we had a considerable increase in performance. Our rupees per errors were very less as compared to linear regression which we were really hoping that actually happened. So then we ended up using the artificial neural network to be able to help in the service design process which was our second step. So in this, this is what I was talking about. So maintenance, insurance, finance and footprint. Besides having them to rank in service design general generally it is also it is important to know whether I should focus or not with limited resources. So we introduce this with two dimensions. One is rank and the other is weight. So if the weight is very less for the fourth one, suppose it is okay if you lose the fourth one. But the first three would make a lot of sense. But if the weights are very close to each other, then the service design approach should be different. So that was why we brought in two different dimensions to this and we forecast the rank and weight. So basically eight measures over here and this recommendation service we split into two. So this is a seven scale likert. We split it into yes and no. So Basically anything over 4 is yes. Anything less than that is no. To be able to find out about the recommendation service, we were able to forecast the weight and rank to a good extent. But when we looked at this is the interesting part, we were not anticipating this. When we looked at the difference between high priced and low priced car, our recommendation forecasting ability was reduced considerably. So we were able to recommend estimate Recommendation up to 90% for low price car. But when we looked at the low price car, our recommendation forecasting ability went down to 62.5. So it was something which was interesting and was my later part of the summer where we actually decided that why not conduct a post hoc test? Is it because of our data science or is it because of something else? So this was not that we started out with. And so we conducted a post hoc test where we collected more data. So the idea was to look at accuracy versus more number of parameters. And when we collected this, interestingly we saw a further decrease in our forecasting ability when we had 200 more samples come into play. So after this, what we can safely conclude is that with the increase in price, because we were selecting the same behavioral data for both. So we had the same survey, just the cars were different. So it is likely that you have more satisfaction for when the price increases you have to collect more behavioral data to be able to predict better. That is what came out of our findings. So to conclude I would like to mention that the first essay was the whole purpose was to look into the whole literature, dig into the details, find out what is happening and present a summative analysis so look at the different trends, look at the evolving theories and then propose future research directions. This is exactly what we did in our first essay. Our second essay developed and tested a typology for of strategic orientation which would help us to explain the problem a few variations in literature that we had seen and finally in our third essay we developed a method for dynamic service design using behavioral analytics. Thank you very much.

Jury members who ask questions we have two rounds so that's just one question at a time starting with.

This is a very impressive piece of work. Thank you. Congratulations you and your advisors for helping you get here. Clearly.