 Right. My name is T.S. Biology. I've been in this business for about 15, 20 years. Today we're going to talk about a little bit background about who we are. So the company that I represent is Cox Communications. I lead a group of user experience designers as well as research and analytics professionals. And what we want to talk through here today is what we're doing from a business of analytics and big data standpoint, but how does that affect user experience, alright? So we'll go through who we are, our philosophy, and give you some foundation material, and then we'll get into big data and how we really use it on a day-to-day basis. I got about half an hour, so we'll go through about 20 minutes of slide work here, and then we'll get into 10 minutes of questions. And I'll be available for a few more minutes after that, so you guys can contact me at that point, alright? So the way we approach design, right? So we are solving customer problems and using research and analytics while we are making sure that the experiences that we deliver either meet or exceed customer expectations. And we have different components through which we measure whether a particular experience is meeting or exceeding customer expectation. When I talk about experiences here, it's not about just UI design. It's about creating an experience which UI design is a component of, alright? So what you're seeing up there on the slide is the group that we have, we have about nine design professionals and we have about four research professionals. So roughly one is to two ratio, so that's what the group is comprised of. They come from a variety of different backgrounds, art school, industrial engineering, psychology, computational science, human-computer interaction, as well as we have data scientists that come from psych and stat background, alright? So the way we approach design, so we have physical design as well as digital design. So when I say physical design, think of it in terms of store spaces, think of it in terms of paper bill, the bill that you receive, think of it in terms of anything that's non-digital that you might be interacting as a customer inside of a workspace or a call center or in terms of something that comes to your home in terms of a set top box or something like that. So just to give you some background, I forgot I mentioned, Cox is a cable company, so we provide telephone, digital data products as well as TV communication products as well as internet home security and home automation products. What that means is, I don't know if there's anything in India that compares to something like that, if there is, I'm sorry, but essentially we provide all of these four different services bundled together and you get a discount on the bundle. So that's the business proposition there. So most of the designs that we provide range from physical to digital, but we are providing that from a discipline of thinking about data and the data that we acquire either through interviews or through interactions that customers have with our different products. And when we think through those things, I'm going to talk a little bit about how my group is structured and how we think about data and how we translate that into design principles. So from a design standpoint, it's very typical. If you're part of a design group, you have something like a UX strategy, prototyping, visual design, content strategy, depending on whether the emphasis on content strategy is if you're a digital group, if you're not, if you're a product design group, there's probably less emphasis on content strategy at that point. Our design process looks something like this. It's very iterative. You start at some point where you have an understanding of what concept that you're trying to build and you get into an understanding of what customers are expecting through either research or existing data and then we go into an understanding of what is it that we need to build. We go deep into a particular concept and then we eventually launch something and then we learn from that and then we go through this process all over again. Fundamentally, this is how most design groups work, so I'm not going to go into great details. The kind of deliverables that we go through, we have something called as blueprints. It's essentially different components of what a digital design could look like. Everything from topography to color to patterns that we're going to employ within the digital space. If it's a physical product, then we have CMF or other components that we need to consider. From a process standpoint, this is what we're doing. We think about the customer's journey through different stages, so everything from learning awareness all the way to engagement, purchase, use, and then finally serving and paying through our processes. We go through these journeys. We understand whether customers are meeting or expecting different things in each part of these journey and our designs are either surfacing components of that or we're trying to address fundamentally how we run our business from a business process standpoint. Let's look at research and analytics. This is a little bit interesting. The way we approach research and analytics, we think about it in terms of health. Health is how good of a product, how good or how well a particular product or service is doing in the marketplace, and then we think about it in terms of diagnostics. There could be a problem that comes in that says customers are having problems understanding our bill. That's a statement that comes in from some place, either through a call center or through some executive someplace, and then we go deeper into that, and then the last piece is the strategic. What we do here is we think about it in terms of if I have an investment of 100 bucks that I need to make, let's assume that capital is not unconstrained and most companies' capital is somewhat constrained, and we think about it in terms of where is the best way we could invest that $100. That $100 could be invested in new product, or it could be invested in improving an existing product, but you're also looking at it in terms of what are the returns based on that investment. We'll get into ideas of how we utilize big data and trying to drive some of those components. Within our research and analytics practice, we have about six different programs that we run. Designed research, pre-launch assessment, post-launch assessment. I'll start with design research. It's anything similar to ethnography or going into a customer's home trying to understand what is it that they do, what are they trying to achieve through the products that we provide, or if you're trying to get into a new area, what are the things that we are trying to achieve through that. Pre-launch assessment. This is pretty much risk associated with a launch of a particular product or service or a new component. Think of it in terms of if I launch this particular product or service, would customers interact with it, adopt it, and use it on an ongoing basis. Post-launch assessment is our typical large-scale survey. Once we launch something in the field, we go into the field, we try to understand how well a particular product or service is performing, and then we use that data back into understanding how we can change designs or how we can approach the service in a fundamentally different way. Modeling. Modeling, think of it in terms of mathematical models that come together to help understand value of user experience design. And when we think about modeling, we're going to get into a little bit more detail here in just a few more minutes. We use different statistical techniques, and I'll talk about those a little bit in detail. Cross-channel analytics platform, this is a platform that we have where we are able to merge together data from our call center, from our retail stores, from our website, from our mobile applications and mobile site, all the way into whatever customers are using in their homes, either through a modem or through a set-top box or whatever it might be. And then the last piece is 360 analysis. This is a better view of the different research methods that you could employ. So what this tells you is what people do, what people say that's essentially behavior versus perception, why and how to fix something, and how many and how much is essentially a quantification or a qualitative approach to the same thing. This is nothing new. Everything that I'm talking about here is not brand new. It's already been done someplace else. This is a representation from Christian Rohrer's work, if you're familiar with it. So with a 360 analysis, what we do is a complete view of the customer's interaction. So we get data from our ethnographic work, which is a pre-launch assessment or a usability testing. We could be getting data from analytics, and then we are getting data from surveys. So we bring all of these data pieces together and we try to understand what customers are doing, how they feel about it, and why they're doing what they're doing. And then we use that to employ, once again, changes to either design or different components of our business process or service. All right, so now let's get into the framework component. So when we measure our customer's reactions to our products and service, the bottom of the pyramid is usefulness. So that's essentially a measure of ease of use of a particular product or service. Usable, it's sorry, usefulness, whether a particular product is useful to the customer or not usable, it's the ease of use of measurement. We use a scale called SUS and usefulness we use a scale called CSS. In delight, it's essentially how pleasurable a particular product or service is, and we use a scale called BERT and that. And then with trust and loyalty, we essentially use NPS for loyalty and then we have a customized scale for trust. So we use all of these different components, and essentially what you get out of it is if it's a survey or something like that, you're able to essentially model how these things relate to a particular product or service that we have. And then when we report out all of these scores, this is how we typically report out. We say whether a particular product or service is meeting expectation, below expectation, significantly below expectation, or exceed expectation. So by normal curve, right, only 10% of the products or services are going to meet, sorry, be and exceed or way below categories. The rest are going to fall in between the two. We'll get into big data here. So that's a quote from Eric Schmidt from Google. I'm not going to read through it, but essentially there's a lot of data that's being created. And I'll go on a tangent here. A little less known story, the data that he presents there is not fairly accurate. So essentially what he says is it's about five exabytes of data that gets produced every two days. It's about seven exabytes of data that gets produced in every three days. So if you Google it, you'll find out the truth behind that. But essentially exabytes, in reference to how big that data is, think about it in terms of the state of California, the state of Washington, and the state of Utah. So essentially it's as big as those three states. If a single byte is a size of a grain of rice, you can essentially spread seven exabytes of data across all of these three states. That's how big it is. Depending on which company and what you guys do, your big data solution is going to be different. Your big data is going to look very different. For example, the company that I work for, we produce roughly, just from our website, we produce about 12 gigs worth of data every single day. So that's information that we're getting from our customers. And big data is essentially defined in different ways. And this is a fundamental view. You can look at it, if you Google hard enough, you can find 10 different Vs of big data that someone has explained what those Vs are. But essentially it's a variety of volume and velocity at which data comes through. So the explanation that I was giving about 12 gigs worth of data produced every single day, that's volume of that data. And if you start to look at the velocity at which that data comes in, that's what that component is. And different varieties of data. So what I explained is internet data coming from our website. You could have internet of things. So if you're in a home automation business, you probably have sensors that are connected to homes in different places and all of that data is coming through. But if you're in the B2B space and if you have a warehouse, you have sensors collecting different kinds of information about whatever business that is, right? So all of that you can combine together and you need a platform essentially to help understand what is really going on in each one of these spaces. So when you think about a typical, sorry, progression of how big data works through your organization, right, you have majority of your work today and this is probably true to most companies that are fairly immature in their big data space. Majority of your time is spent in setting up of data. The next component you'll see is either reporting or some component thereof where people are taking data from different platforms and then reporting out this is what's going on. And very little time is spent on analysis and modeling. And as you mature, majority of your time should be spent on analysis and modeling as opposed to setup or reporting. I'll just say this from the way I work to me, reporting is dead. If you're still in the reporting space, you're antiquated, you're looking at things in the past. You need to be looking at real-time data, making real-time decisions. Obviously that statement holds true in certain kinds of businesses, not all kinds of businesses, right? So in our group, we're trying to get to the target state that I'm depicting here and we have a platform that helps us understand all of this information. So we have different kinds of information coming into a cloud platform and we provide Tableau as a front-end interface to most designers as well as product managers and then we use SAS as a platform for most of our data scientists. The idea here is if you think about where people are using big data and how they're using it and some of the immaturity that I talked about early on, most of the work is fundamentally basic algebra. If you have high school algebra, you should be able to answer, 100 people came into this particular portion of the site, 10 people bought something or 10 people went into payment and that's all they're looking for, right? That's where majority of the work is. But if you want to move on to modeling and analysis and you want your data scientists to move into that space where they can provide valuable information back into your design teams, then you're going to need some help organizationally to figure out how you can self-serve those customers or internal clients, how we want to look at it to get to those basic algebra questions. Otherwise, you're going to have an analytics professional who's either a data scientist or a... Sorry about that. Or someone who has programming background trying to get at data. So essentially, your data is inaccessible to most of your designers today. And what we created with this is a way to get that data accessible to our designers. So we'll look at what kinds of problems you're able to solve once you have access to that kind of data, right? So essentially, we think about this data as a customer's journey through different touch points. So I talked about call center, retail stores. We have something called as truck rolls. I don't think there's something similar here. I could be wrong. But essentially, what happens is, let's say your phone line or your data or your internet or your TV is not working. In the US, we send a truck into your home to figure out what's really going on, and we call that a truck roll. And that truck roll is essentially the most expensive component of the business from a service standpoint. If you think about calls, let's say it costs you about a dollar, your truck roll is going to cost you somewhere between five or six times that, right? So what this depicts is essentially how many people are coming to our website, of those people, how many end up calling us, and of those people, how many end up getting a problem which they cannot solve by themselves, and how many of them end up needing a truck roll at the end of the day, all right? Wow. Okay, I got five minutes. So we can surf through all of this data and we'll get to ideas of why and what is really happening, all right? The same data represented differently here is essentially what you have is everything in blue is a call, everything in orange is a Tier 2 call, everything in green is a truck roll, everything in violet is essentially a website visit. So what is happening is if you start thinking about number of interactions customers are having, customers take up to eight interactions before 97% of their issues are solved. So when you start thinking about it from our company standpoint, we think about it in terms of can we reduce it to four interactions or less, or can we get to first-called resolution or first resolution in the first step or in the first interaction, all right? When we think about modeling, I'm going to go quickly here, all we're doing is capturing data in different ways, representing in mathematical models and then we use some kind of prediction to understand whether if we do X or Y, can that be different in real life or in the field, all right? The most common model that most people are looking at is the weather, right? So a weather gives you a view of high and low and it gives you a view of what is the chances of it raining in Bangalore today, right? So when you think about what are the chances of rain, the precipitation changes is represented in terms of a percentage, right? Which is just simple probability. But the model that drives that probability is fairly complex. It takes into account historical precipitation chances the day before weather or climate or what was happening in the city or in that location and what had happened in the previous 100 years if that amount of data is available. So we are taking something very similar. So this is essentially fairly complex that they have boiled down into a high and a low which is a discrete model and a probabilistic model which is your chance for rain, right? So we take a very similar approach and we look at it in terms of so this is data for our TV product and not ours but our competitors and we're looking at an understanding of how does remote, how does usability, how does other things that we might be doing in the marketplace as well as the reliability of that service starts to affect satisfaction. So when I talked about modeling earlier, I said if I have $100, where should I be investing? So what this helps you understand is should I be investing it in the remote? If I invested in the remote, what are the chances of NPS or the satisfaction improving? If I have to invest in usability and now you start to see where design comes in so it could be a design of remote that you're changing or it could be a design of the service that you're changing which could be part of your usability factor or it could be just pure network changes that you're making which helps you understand the reliability of that particular service and you can further start going into either the set top box reliability or a network reliability and you can further break it down into depending on how the network is laid out further things into the network on where you can utilize that investment. So the idea here is once again just like in weather we'd be able to get to a discrete value or you could get to a probabilistic value of what are the chances of success or what are the chances of NPS or NPS is Net Promoter Score which helps you understand loyalty or satisfaction improving based on the investment that you make in different components. So here's another example of the same thing. So the biggest difference here is there's an education component that was added on. So here what is happening is a customer gets an installation they get education about how the system works and how the remote works and we utilize that information and the success and the satisfaction associated with each one of those steps in the overall scheme of this service and how customers perceive the product to be and ultimately the company to be and we use that in a way where we're starting to go the investment that we need to make for next year will have to make a little bit more on let's say a different kind of remote depending on the demographics or a different kind of education depending on what kind of systems that we're putting in or in the reliability and in the shorter term we then look at how we can solve usability if you have a system that you can update on an ongoing basis. So that's how we are starting to utilize big data in the workplace today and this is probably fairly unique in that very few people are doing this kind of modeling work where you can start to predict issues or system investments that we need to be making and with that I'll take maybe one or two questions I can't see sorry go ahead ask your question I got lights in my eyes so I can't see who that is raising your hand go ahead I mean where do you find resources for such kind of work I mean we have been struggling to find resources for this which kind I mean talent no which kind design or research or analytics which one of those largely on the big data largely on the big data so most of the people that I found I can tell you that I went into schools talked about this found talent that was interested in it groomed them got them to do this on an ongoing basis in my company so that's a short answer to that you don't get any ready resources come on how many of you were ready and came into this what are company you're working for and where valuable assets to that company on day one so alright any other any more questions one more one more yes one example of so the example that I showed you right so that's based on set up boxes right so if you know what's at our box it's essentially as box that connects into your TV that gives you all the different channels what we have done is we have found where we lack in terms of what we're doing compared to our competitors and we're going to make an investment based on where we lack compared to our competitors and one of those components is remotes for elderly and education so that's something that we're going to make investment next year so that's relatively small example of how you're using it go ahead you talked about big data maturity right and you talked about the actual situations where we are giving more importance to the settings rather than we give more importance to the analysis so how do you move from giving more time and effort in doing the settings to analysis because if you don't do the settings correctly you won't get an analysis of proper data reflection right yeah great question ultimately what you got to realize is this is not something you're going to do on an individual basis this is an organizational movement you and your entire organization need to move in this direction for this to be successful this is not going to be a heroic effort of one or a group of five individuals right and as you as you understand that the kind of allies that you need to make that kind of a transformation is what's going to really be telling and it's long lasting and sustaining at the end of the day alright thanks