In the Digital era, Data and Analytics should be at the very core of every business. They should drive strategy, innovation, and operational efficiencies. It needs a supporting culture adopted by employees, customers, and partners. What does such a culture look like and what does it take to build it?

Contributor

• • Somesh Nigam, Senior Vice President and Chief Analytics & Data Officer, Blue Cross and Blue Shield of Louisiana

Download Podcast
Apple Podcast, Google Podcast, Spotify, Pandora, iHeartRadio, SoundCloud, TuneIn, and Stitcher. Find other syndication channels here or search CIO Talk Network podcast on any other app.

Explore More

• • Creating a Winning Data and Analytics Strategy
  • Steps to Future-proofing Data and Analytics Environments
  • Building High-Performing Data and Analytics Team
  • Have Enterprise Data and Analytics Programs Delivered on the Promise?
  • Is Data Analytics helping increase the impact of K12 Education?
  • Data Analytics as a Team Sport
  • The increasing Role of Data Analytics

Transcript 

Sanjog Aul [00:00:28]:

Hello, everyone, and welcome to this segment on CTN. To learn more, please visit ciotalknetwork.com. And our topic for today is Building Data and Analytics Driven Culture  . The guest is Somesh Nigam, who is the Senior Vice President and Chief Analytics & Data Officer with Blue Cross and Blue Shield of Louisiana. Hey, Somesh. How are you?

Somesh Nigam [00:00:50]:

Alright. How are you, Sanjog? 

Sanjog Aul [00:00:52]:

Very good, sir. A beautiful morning here in Chicago. And, how’s Louisiana treating you?

Somesh Nigam [00:00:57]:

Louisiana is beautiful too. I can see outside. There’s no cloud in the sky. Beautiful weather.

Sanjog Aul [00:01:02]:

Awesome. Awesome. So the reason we got together is because, in the digital era, as we know, we are all trying to move at 100 miles an hour, trying to touch and or change or optimize every possible, nook and corner of our organization. And as we are doing it, we know we would need data as the underpinning or the support of data and analytics because that goes without saying. But then in order to make digital successful, this foundation itself should have, should be put properly which means you’ve got to build a culture which will help drive, whether it’s innovation or operational efficiencies or even strategy and execution, you name it, everything should be driven by data and analytics. That’s a should. But without having a proper culture, without having your customers, your employees, your partners, or any other entity or even including your management, if they don’t adopt it, then it will be a wannabe effort, and there will be a lot left to be desired. To make this happen has been an interesting challenge, and to some, it’s been a chronic challenge. We wanted to discuss and dig deeper, and no better person than someone like you who’s living it, and I’m assuming you’re being made in charge of not only tackling the analytics and data, but also to support building of such a culture. Am I correct?

Somesh Nigam [00:02:38]:

You’re absolutely right. I think just look at this. We now have a title in many of the organizations, and Blue Cross and Blue Shield of Louisiana is an early leader in that, a title called chief analytics and data officer. This is a new title. It recognizes that data and analytics is central. It’s core to company strategy. As well as, the journey, to your point, is not just creating a large warehouse or anything like that or getting a few cool analytic toys, but actually making the organization into an analytically advantaged organization. Which means that whatever insights we gain have to work in driving the business across different functions. And that requires a new way of thinking. I mean, you obviously realize, Sanjog Aul, that what we are able to do now with new era data and analytic technologies is a leap, a huge leap beyond having data present in multiple silos and some retrospective reports, etcetera. So it also causes business and functions to think with their thinking caps on. What if? And that itself is a change in culture. So I would say it’s actually, sort of almost a circle. People begin to think, and many of us are exposed to new analytics in our consumer space. I know when we run a Google query or buy a book on Amazon or listen to a song on Pandora or what have you. We are experiencing the cutting edge AI and data processes behind the scenes. And we start thinking that what if I could apply some of the same things in my organization? Why can’t I predict diabetes if I can predict the book that I like? Why can’t I predict the best care pathway for a patient, if I can predict how to go from point A to point B on a GPS, etcetera. So a lot of people have already been thinking, but tools and processes, in many of the organizations are still the old era, static data marts and reports. So as new technologies are brought out as proof of concepts, people get excited. And they say, well, if you can do this, I can perhaps apply it to my area as well. So I do think there’s a synergy that you get. People start thinking about new possibilities. As you bring new technologies and proof of concepts in the organization, folks get excited. And then it starts creating a wave that then becomes unstoppable. So I think all of it can happen, if it’s appropriately implemented, communicated, and enough excitement is built about the potential of these new data and analytic technologies. And we’re finding it becomes fairly organic if you do it right. Now if you do it in a silo and in one corner of the company and it leads to a few cool PowerPoints or something, and then it dies out, of course it won’t thrive. So it’s much like creating a garden and tending to it.

Sanjog Aul [00:06:16]:

So to your response, a couple of things that you mentioned about us trying to play with the data or people like yourself who appreciate what the possibilities are. But then you are not able to do this if the people at the LEAF level, which is at the operational level, don’t say that it’s my responsibility to help create clean data or be the guardian of my data which essentially rolls up. And they neither have the incentive or they would say, okay. Unless it is part of how I’m measured, that’s something which I’m gonna do because I’m just gonna generate data, but the quality aspect, which is what becomes critical, doesn’t happen. And this has been a challenge because of which MDM failed in the past.

Somesh Nigam [00:07:11]:

Yes.

Sanjog Aul [00:07:12]:

And even now you’re gonna have other struggles. So how do you I mean, you can go and do all the cool things in digital.

Somesh Nigam [00:07:17]:

How do you handle it? Yeah. I mean, let me divide your question into two parts. So the first part really is what you refer to folks at the operational level. Do they feel excited about it or not? And I actually would say that they do. All of us are experiencing digital revolution in our lives. So, in fact, we find some of our greatest ideas come out from the line level, who really are looking at the processes very closely and ask this question, what if I could change something? Now clearly, the data that you need to drive that change needs to be of certain quality. And I would say, Sanjog Aul, that it needs to be different quality levels for the kind of things that you are targeting. Clearly, a financial report that has to go for the board meeting or for quarterly earnings release needs to be exactly tied down to the decimals to make sure it’s accurate. However, many of the decision making processes we are trying to build, and apply with data and analytics are what I would call probabilistic. We are trying to assess the probability of an event happening. So for example, if we are trying to figure out in our world who is likely to be hospitalized in the next 30, 60, or 90 days, we’re really trying to assess the probability. Is it 70%, 80%, 90%? Right? Processes like that also can take a lot of less than pristine data. So you can actually get, for example, we found our customer service calls and our calls into care management appropriately digitized with word clouds and so forth. So very probabilistic here. Right? Can actually feed into, be combined with structured data that we have. And together, the two, with appropriate AI algorithms can predict who is likely to be hospitalized. So the you mentioned MDM. That probably has the highest quality standard. Right? You obviously have to manage your master data management to pristine levels. But then a lot of other data sources don’t have to be at that level. Data derived from social media, data derived from calls. But what we find is as we recognize multiple types of data, join them appropriately together, and run them in these AI algorithms, the results essentially give you a very high probability event, which we can act upon. And that’s really what the whole battle is. How do you improve the signal to noise ratio so that what you are going to address is not false positives and it is really a relevant signal. Whether that’s signal around disease, whether it’s signal around complications, whether it’s a signal around impending hospitalization, or a signal about cost escalation. It could really be anything.

Sanjog Aul [00:10:32]:

So based on what you just said, it definitely makes sense. Now the thing which you mentioned is signal to noise. Now this relevance of the output of that, more the probability of that being a signal versus a noise, to some extent, is also directly proportionate to what you put in. And what do you really focus on? The two part question that comes up, number 1, can we realistically work on literally everything that gets generated and put it in a hopper and somehow try to manage it? And then secondly, what is the quality of where it gets generated? And you are right about it that we will be able to only control so much as it goes in. But then how do you prioritize your efforts, your energies, your dollars so that at the point of entry of data into that engine, if you will, you’re ensuring that you’ve done the best you could to have the best possible data quality to produce for the most useful type of data.

Somesh Nigam [00:11:38]:

Right. No. You’re absolutely right about that. There’s absolutely a data hierarchy. And first of all, a very mature data governance operation, working together with all the folks who are collecting and warehousing the data, whether structured or unstructured, is required. And a lot of these issues get discussed, at least in our governance process. It is not a bureaucratic process of cataloging data sets anymore. Of course, that’s all part and parcel of it. Rules by which we will act upon, of course, in our healthcare world, HIPAA and related groups, those play a big role. Contractual obligations play a big role. But, increasingly, the question is about value. Right? What value we are trying to create. And in order to create that value, how will business processes change using data and analytics? And of course, in order to deliver that, what data domains are needed? And at what level of quality? And once you answer those questions, you actually can set up processes for that. So for example, if we are driving something with our claims data, that has to be highly accurate. This is highly curated. This data drives our business. But then data coming from social media signals will be more fuzzy, and that adds probability of events, for example. Right? So we end up creating many data domains within our Data Excellence program. And all of these data domains work together to deliver the value at the end. So, I hope that clarifies a little bit. Right? I think if you keep your focus on value and value is almost always driven by corporate strategy. Right? Corporate strategy defines where we are going, how we’re gonna create value. Then in order to create value, what are the drivers for that value? In order to then deliver on those, what type of data and analytics and business process is needed? So all of these are very integrated. And most companies don’t have that culture of integration across functions, across different business areas to deliver value. So going back to your original question, when you start on your data and analytics journey, it’s not a technical journey. In fact, technical technology pieces are relatively small pieces of it. It’s really a culture journey, and that allows you to then drive value.

Sanjog Aul [00:14:30]:

Let’s take a quick break, listeners. We’ll be right back. This response that you gave was very thorough, and it does bring up some important points. You did mention that the business has to set parameters on where it wants to create value and let everything flow from there. Now how about us trying to even break that mold of business somehow from somewhere will create the definition of how it’s gonna create value? What if the data and analytics engine itself could start giving them the insights which could then drive the strategy, the innovation, and the operational efficiency. So could we, and in order for us to do that, we do not have anything else to bank on or a throat to choke, if you will. That means something has to happen far before business even says this is the place where I create value. So how do you, it’s like a catch-22. How do we tackle it? Please stay tuned, listeners. We’ll be right back and explore.

Sanjog Aul [00:17:24]:

Welcome back. So, Somesh, the question about business telling, from their ivory tower that this is what we wanna do and rest of the world starts following, perhaps that is getting challenged, or the business at the top is recognizing. In fact, I had a show where once someone mentioned that lately in a financial services firm, they have one of the board members is actually an algorithm. And that is. Yes. So imagine when the reason they are bringing it is because we cannot definitely counter somebody’s years of experience and the fuzzy thinking that human brain allows, but at the same time, we cannot undermine what data science and data engineering and these algorithms are allowing you. If we were to fundamentally break the mold of saying, let us have the whole organization be driven or at least be empowered by data and analytics. What prevents us from doing that?

Somesh Nigam [00:18:25]:

Yeah. I mean, I think it’s a journey. And what you are describing about an algorithm being part of the board and so forth, it’s clear that even today with data science, it appears to be fodder of science fiction. Having said that, most of what we thought was science fiction a couple of decades ago has come true. So I wouldn’t be surprised if that happens. But where we are in the journey, and I’ll give you some specific examples. It increasingly happens as you bring a data and analytics culture into an organization that while the first round of suggestions as to where analytics can be applied comes from business areas, unexpected nuggets come out of your analysis that they were not even aware of. So I’ll give you an example. We had worked in my past life on a model to predict customer complaints. And it seemed logical that complaints have a pretty significant impact on your business in many ways. Right? Including reimbursement and incentives and so forth. So it seems logical that a person just doesn’t escalate a complaint on the first go. They call once and their issue doesn’t get resolved, and they call again and it doesn’t get resolved. And after 2 or 3 frustrating events, they escalate the complaint. It may also depend on what their age is. It may also depend on what particular issue they are dealing with. It could be that someone has more complaints in a home health setting versus in a hospital setting. They may have complained because they were unable to fill their prescription. It could be many things. Right? It could, of course, depend on their economic level as well. So it’s a complex problem. We said, well, let’s try to create an algorithm where we’ll predict complaints. The idea was that if you can create a hot list of people who are likely to complain, our customer service folks can get to them ahead of time and solve that complaint. So we will get an advanced signal. Who’s gonna complain? And certainly that will go a long way in improving the quality of our service and the customer experience. So we went ahead and created a model using a whole bunch of variables, including customer service calls and all that. And indeed, we were able to predict 70% of the complaints often weeks before they happen. In fact, there are many examples of where we could predict it even a month to sometimes 2 months before they happen because we could see that escalation. Very successful. We implemented a small program, had a team that started calling these folks and tried to solve them, and in fact saw a dramatic reduction in complaint escalation. So all great story. However, in that process, we found all these nuggets that business wasn’t even aware of. There was a particular event that I remember where a person’s name popped up as one of the variables automatically through the algorithm. If I were to just say it’s an agent Heidi. Well, agent Heidi was somebody who was solving a fairly complicated set of problems, and there were a lot of complimentary messages about her. Well, this is something that nobody has visibility to. Right? They could now go to this agent, Heidi, and ask her, out of hundreds of agents who we have answering phones, what is it that she is doing that is different? And what type of problem is she solving? On its own, there would have been no visibility to either their managers or senior levels as to what was the problem. Now by looking at it, we were able to solve it. We also found, for example, certain towns had a much higher complaint rate than other towns for similar background. We found that certain types of durable medical equipment that was delivered at home had much higher complaint rate. Right? Now, think about it. Durable medical equipment such as wheelchairs, respiratory equipment, other support equipment at home. Some of those are harder to manage and have a lot of complications, others don’t. So signal is getting generated that can feedback that, you know what, we don’t want that type of wheelchair or we don’t want this type of diabetes monitoring equipment because this is prone to more complaints. So you end up getting a lot of signals that were completely hidden. Now once those signals come out, it’ll logically make sense. Do we now inform our business to go and solve these problems? Right? So this becomes very much like what you said, an algorithm advising the business. So it started out with business advising us that there are complaints, and now algorithm is advising the business of what to fix and how to fix it. Right? So this type of insight becomes pretty prevalent, and you start finding more and more such examples. Now will it reach a point within our lifetime where literally there is a member of management or a board sitting there who is actually an algorithm? It might. Right? It depends, really, on assimilating a lot of complicated information and finding those nuggets and advising as to what’s the right course. I still am a firm believer in human ingenuity, and I feel that human intelligence could create a Monet or a Chagall. That question keeps on coming up. Where is innovation and that spark of innovation? Can that be taught to computers without being rule bound? So it’ll be an interesting journey that we both get to watch in our lifetimes.

Sanjog Aul [00:24:50]:

No. Totally. And to the example that you gave about getting better in handling complaints and being predictive, did you just use the data that was already floating around or did you actually prompt this to be an organization-wide initiative where the people on the front line gave you some additional data or you started capturing additional data so it enabled you the insight which you were not having earlier?

Somesh Nigam [00:25:22]:

Yeah. No. It was clearly data that was being brought in. Now if you were to say data, right, the customer service calls get recorded and if you are lucky, they’ll be transcribed, and you will have the transcription available. If not, there at least will be customer service notes that’ll be there. All of those get typically archived, especially the unstructured portion are very hard to analyze. So we essentially created a process where we brought all of this unstructured data in and started analyzing them via algorithm. What do words mean? What really says somebody’s upset versus not? And it’s a complicated problem because you realize that there are two conversations going on. There’s a member calling and a customer service agent responding. So you have to separate those out. Now someone could be calling on someone’s behalf, and that could change things. We found fairly interesting patterns. You found, for example, in a product that was meant for the elderly, when people started using that product, there were a lot of complaints, and the complaints went down as people became used to it. Then as they aged, their kids got involved in their care, and the complaint rate went up because now kids are trying to learn the new system and the process and trying to take care of their mom and dad. So it was an interesting U-shaped curve. But you do get all of these interesting insights. And to your point, you end up using data that you otherwise don’t consider data. I mean, there’s also a lot of data in the temporal relationships. How often someone calls you, what is the frequency, is that frequency getting escalated, is a call to care management also happening at the same time as a call to customer service. That can be a signal. Typically, we won’t analyze all of that. You may be able to intuitively find a few variables. But if you end up applying a machine learning approach, you would create thousands and thousands of derived variables. And within those clusters, you will find insights, which you could then present to a business person, be it a clinician or a customer service person or someone who’s managing the operation to say, well, this is what we are finding. This is correlated to higher complaint rate. Do you see this? And sometimes they’re able to scratch their head and figure it out, and sometimes they are not. So a lot of it is based on these hyper correlations.

Sanjog Aul [00:28:10]:

Let’s take a quick break, listeners. We’ll be right back, and let’s talk about the challenges because I’m sure when you came in, Somesh, it would not have been fully cooked like the culture of data and analytics, and I’m sure you’re still working on it. So what do you think are the typical challenges faced in building, or starting out and then building this culture? Please stay tuned, listeners. We’ll be right back and explore this further.

Sanjog Aul [00:30:25]:

Welcome back. So, while we’ve discussed what you were able to do, Somesh, in your organization, when you came in or when you were tasked with doing this data and analytics related effort or taking this on, what are the typical challenges you face or what you’ve seen other people facing as they get started?

Somesh Nigam [00:30:45]:

Yeah. I mean, I think the key challenge is that traditionally, organizations have been organized in a very siloed manner. And in fact, on a day to day basis, we are finding that to move the needle on anything, be it a new product, new sales, new innovation, new way to manage health, multiple business functions have to collaborate within the organization in order to make that happen. And typically, whatever you are trying to make happen is a data and analytic driven decision. So in some ways, we are already on the curve to not only integrate business processes, but also the data underneath to allow that to happen. But the old habits die hard. And almost every organization, whether it’s the CIO or CDO you talk to, will talk about multiple data marts hidden away in the organization, things on people’s desktop, little nooks and crannies that access databases here, something like that, a lot of data duplication. And you then find, of course, that leads to confusion and doesn’t give you the right answers. So one part of the culture change is just that. That’s a huge challenge and that’s what we are trying to tackle through the Data Excellence program at Blue Cross and Blue Shield of Louisiana. What we found is that we did invest heavily in enterprise data warehouse capability using cutting edge technologies, and in fact, that was adding value. But what we found did not change is what our CEO often called, if you build, they’ll come strategy. We had built it. We assumed everybody will recognize that, and they’ll start connecting their processes to this pristine source of integrated data. Well, that didn’t happen because, as I said, old habits die hard, and we still have a smattering of individual data marts flying all over the place. So this particular strategy that we have launched is to first find and then retire those marts and link everything to a well governed source of data, which we are now calling single source of truth for the company. So that’s a journey that we are on and that is a huge challenge and I would bet that many of the CDOs deal with that. Now, if you have a pristine source of data, does it all have to be in one warehouse? Some organizations may find that there can be several such sources, a handful of them as long as they are governed appropriately and serve as that single source of truth for that data domain. So that can be done as well. But going ahead, the folks who want to make a change, you can create the best algorithm in the world, but if you don’t have a business process change mentality and process, and kind of culture, the algorithm will collect dust on the shelf. For example, if we were to create a process which identifies with high accuracy who’s gonna be hospitalized in the next few weeks, the clinical management will have to change their process. Right now, they call a certain percentage by their scores, and that’s their process. Now they have to create a separate unit. They have to specifically call these folks who may not be calling in. You have to have some of your best people talking to them to figure out what their issues are and you need processes to solve them, which means it requires a business change. Change is always hard. And you have to make sure that we build a culture of change to make it happen. So that when analytics folks come up with their next best thing and they’re standing at the door of business and say, hey, you need to try that, there should be equal amount of excitement on the other side. Also, when business people come up with a great idea and they come to analytics, we need to come up with the best way to solve that problem. So that’s really the building of the culture. And I think that is perhaps the biggest thing we struggle with on a day to day basis. How do we continue to create that culture? One thing that you mentioned, we do have an innovation process that we adopted in our company. In fact, it is run by a senior executive. All of the senior executives are involved in that process. The idea there is to try to nurture innovation all across. It could be via an idea proof of concept. It could be by bringing a new idea. Often, they are not data driven ideas per se. They’re, well, what’s the right thing to do? How do we improve customer service? How do we improve our products? How do we create a better insurance product for a particular population? Could be anything. How do we improve customer experience? So people come up with that, they get excited about that. And as they come up with ideas, when we start thinking about how to implement that, very often the journey goes through the data and analytics channel. You have to kind of figure out how to get appropriate pieces of data in some manner to drive this new process that someone is creating. So I find that one of the things that chief analytics and data officers can do is to be tied at the hip with the innovation part of the company. Sometimes it is internal innovation. Sometimes it is in collaboration with major universities and academia. We are doing that here in our neighborhood, working very closely with organizations such as Tulane and LSU and other academic institutions so that we can get benefit of new research and new ideas that are incubating in academia. Another way to do that would be to really have innovation challenges. We are very fond of things like hackathons and innovation challenges. You know, concentrated amount of time when you bring people together to work on a problem which is somewhat defined, gets people excited, gets people’s creative juices flowing, and then can lead to something big. So there’s no one answer to this, but I would say all of these can play a role in changing the culture. It’s exciting times. And we do find that as we nurture these, lots of ideas come about. It’s not a top ground process anymore. The best ideas could come from any corner of the company or even outside.

Sanjog Aul [00:37:42]:

So you mentioned part of your response that you would like to have the business tell us what they want and or come on board. Now when you start this journey of identifying what’s important for them, it may not always be, I mean, they will say that, and they have an agenda because they feel that’s important. But when you look at the complete enterprise perspective, you may not give it the same high marks as they are giving. And so on one hand, you have to take care of a universal view of data so that it becomes a centralized data creation and engineering and governance and analytics. But then that business person says, if you’re not gonna take care of me and your approach is not gonna take care of my agenda, then you can literally kiss goodbye the culture that you’re trying to develop because there’ll be a passive resistance going forward. So how do you tackle this?

Somesh Nigam [00:38:36]:

Yeah. But that can happen. I mean, one of the things I tell our analytics people, and in fact, we have changed we’re trying to change our HR processes to make sure that folks are incentivized not just to create an algorithm or a report or a new data model, but actually, they get incentivized on changing the business process, which requires then our analytics folks to work very, very closely with business. And in fact, that business outcome is what they are being incentivized on eventually, not just the path leading to it. So in small ways organizations can do that. Another thing we have done is to create business relationship managers. And we’ve grown on that path. We already have a couple of them for our key business areas, and we are adding many more. These are folks who are most experienced on our side. Sometimes they have actually come from business. They really know whether it’s an actuary or a clinician or provider network or sales and marketing. They have deep understanding of those areas, and we assign them as business relationship managers, and they literally sit with the senior management team of that business leader. They are extremely familiar with the nuances of that business. And they can also manage this process. So for example, if a business leader says, I need X, Y, Z, they actually can, in a very respectful way, debate that and say, well, you probably don’t need X, Y, Z. You probably need A, B, M, X, and that other piece is really not important, and maybe we should focus on these areas. And it becomes a very collegial dialogue, of give and take versus a relationship where someone is telling you what to do and you are doing that. It becomes very thoughtful. So I found these folks who we call business relationship managers work their weight in gold. Because they not only know the technology and the processes and algorithms and the challenges involved in it, but they also have deep knowledge of business and actually can create this thoughtful dialogue with business units about what needs to be done. Very, very useful for portfolio management. These folks, we don’t even have any operational responsibilities for them. Their job is to manage that relationship and, in fact, the entire data and analytics department’s resources are available to them to make something happen. So maybe structurally, that worked very well in my past life and we’re trying to institute it here. There are other ideas that can allow for better collaboration. But in the end, we do keep corporate strategy in mind as we allocate our resources. Some areas are nice to have, some are business as usual, and others are really driving the business to go to another place better in terms of quality and cost. In our case, a lot of it boils down to the cost of health care and quality, and how do you manage those two? And overlaid on top of that is our member experience. So very often it’s referred to as the Triple Aim, and that’s what we focus on. So I feel that helps us.

Sanjog Aul [00:42:11]:

Okay. So based on what you just said, all of those are pretty good ideas. You at the end of the day put a bow and tie and say, okay. All of this has to be driven by the strategy, and this is a structure. Now if I were to come and try to have you go to the next level, would I not want to break that structure and say let’s do more experimentation, let’s do more sandboxing, and not let everything be just driven by strategy because that could be holding you back from finding or rethinking what’s possible?

Somesh Nigam [00:42:48]:

Yeah. Rapid analytics is a key part of our strategy. And in fact, that does involve creating sandboxes and creating data marts on the fly, joining different types of data, and providing turnaround analysis not in months, but days, to give some type of a high level assessment. And those experiments sometimes can be done with the help of analytics folks, but sometimes they are self-service. So we can take BI tools that are not just reports, but in fact allow fair amount of customization on a query and let the folks try to find their own insights. So we truly believe that more experimentation you allow, better drill down on data you allow, better predictions you allow, better what-if scenarios. In our case, predictive models that can tell you whether this strategy will work or not. You know, let’s say we were trying to push for a certain medication management or medication adherence strategy. We can take a look at our data and see whether that has been tried elsewhere and try to find those cohorts, find a control cohort, and see whether that worked historically or not and then see if that can be scaled up. So those kind of things need to happen at much faster scale. And in fact, we have set up a rapid analytics unit just for that. Their job is to provide analysis very quickly. We also have a highly curated research unit, which does outcome research that is published in top flight journals and can inform policy nationally and talk about what Blue Cross and Blue Shield of Louisiana is doing, etcetera. But that’s at the other end of the spectrum. And in the middle, of course, there’s a whole bunch of activity which I would call operational, day to day operational support. So I hope that answers your question. I think that’s pretty critical.

Sanjog Aul [00:44:55]:

So please thank you so much. Please stay tuned to us. We’ll be right back, and let’s talk about the sheer volume or variety or velocity of data that’s getting created as part of how we are tackling our business and living our lives. So whosoever is going to churn it and or manage it, whether from a data engineering standpoint or data science, data and analytics, how do we make sure that it’s not getting out of hand? So people always say fewer things done better. So when we are going to go and keep branching out into 20 different directions, try to stretch people to the max and also burn out people and other resources, is this truly going to help build that culture of data and analytics when people are saying this is gonna do nothing else but stretch me to the max? How do we prevent that from happening? Please stay tuned listeners. We’ll be right back and explore.

Sanjog Aul [00:47:38]:

Welcome back. So, we are talking about the volume, the variety, the velocity of data that gets created. And if we keep bringing it on, while it looks cool, it looks exciting, but then does that make a dent which will hurt in a way the potential of us building truly a culture, among the people, not just people at the top would be happy, but also the people at the bottom who are actually in the field staff, who are supposed to be responsible for this data creation, they should not be burning out. There should not be any burnout. How do we prevent that from happening?

Somesh Nigam [00:48:19]:

Yeah. I mean, I think this is the sort of problem that the leading edge AI technologies were not built in companies that were in traditional businesses, but they were built in places like Google, Amazon, Facebook, etcetera. They realized that they had to manage gigantic volumes of data in real time and act on it and have to build a whole slew of technologies that automated many chunks of it, ingestion and curation and then being able to automatically apply algorithms and so forth. So I think one answer to your question is that technologically, you want to make sure you’re using things like Hadoop and Spark and new age technologies that work on those, R and Python, etcetera, to be able to run those in near real time. And we are doing quite a bit of that. The learning curve is pretty steep, and we are coming up to speed there. The second part is that the data deluge has just begun for us. So whereas a standard visit to a doctor led to a claim which was just a few fields, now we are talking about getting data that could be their electronic medical record. We could be looking at some call that they made. We could be talking about numerous conversations they had. It could also be data from their wearable devices. It could be coming from home monitoring systems or a health band that they are wearing, etcetera. Very quickly, we can see the pathway where the data that we are collecting today, as large as it is, is about to become 10 times larger and perhaps take another leap beyond that. So there will be a data deluge. There’s no doubt about that. If we don’t plan appropriately with technology expertise and processes, it can overwhelm you. Now, of course, not all the data has to be treated the same way as long as it is onboarded and you go to the right domains to look for particular insights, and integrate them. That’s really where we are right now. But eventually, algorithms themselves will be scouring different data sets to find all of the patterns automatically, and we will be able to get those insights. So it’s a journey. I think this will be a big topic of discussion at many of our upcoming meetings and conferences. How different CAOs and CDOs are dealing with this comes up a lot. The other part of your question was how do you avoid overwhelming folks at the lowest level? You have to be very careful about that. It’s critical to make sure that you don’t impose a process upon people that completely overwhelm their capacity. And how do you do it in a manner that is proportional to creating value? In the end, every year we go to our board and ask for resources to do something. Of course, that task is always balanced against the value you are creating. If you are very good at not only bringing these processes but also quantifying the value, I think that’s something we should spend a little bit of time on. A lot of organizations are very good at creating insights, but they are not great at figuring out their own value, and not just the initial value, but how that value is sustaining itself, growing, or declining. If we create those processes, then the question is never of do we need more people or more technology. It’s all about value. If you’re creating value, you’ll be able to get resources allocated to you that allow you to keep up. So if you combine it all, we can create a culture which grows sustainably and continues to inform the organization about what to do.

Sanjog Aul [00:53:06]:

30 seconds for you. If you were to give a message to the other leaders who may or may not be wearing the chief data and analytics officer hat, but are responsible directly for building this culture, what type of leadership should they bring to the table?

Somesh Nigam [00:53:30]:

I think just like ages ago we created the title CIO, and we are obviously talking to the CIO network today, we recognized that all the IT infrastructure and databases will overwhelm an organization. So you need a CIO. I would encourage very strongly identifying the leadership that can be under the name CAO, CDO, or in my case it’s both together. But you do need that leadership. If you can bring that leadership to your organization and empower it. In our case, we are very fortunate that that initiative came directly from the CEO of the company, Steve Budwej, who’s a visionary and actually was on this journey many years ago starting in the early 2000s. He recognized the value of data and analytics and how that can change your business in dramatic ways. If you have that CEO level support, focusing all of that energy on one person whose job is to bring all of the data and analytics processes together for the benefit of the overall organization, I think that’s a key step. And that will be my message back to many of my compatriots that just like we recognize CIO as a title, it’s time to think and grow the CAO or CDO as a title.

Sanjog Aul [00:55:02]:

On behalf of the show and our listeners, thank you so much, Somesh, for sharing your views on how organizations can build a culture of data enablement and a data and analytics driven culture and use it to successfully drive our strategy, innovation, and operational efficiencies. Thank you.

Somesh Nigam [00:55:20]:

Great. Thank you. Enjoyed it.

Sanjog Aul [00:55:22]:

And listeners, hope you enjoyed. Got some nuggets from Somesh. Please, like us on Facebook. Search for CTN, and be sure to follow us on Twitter and LinkedIn. Thank you again for listening to this segment on CTN. This is Sanjog Aul, your talk show host. Till next week, take care, and God bless.