This is an audio conversation from the CIO Talk Network podcast. The timestamps in the chapters and transcript below refer to the audio recording.
About this conversation
A CIO Talk Network conversation with Vijay Venkatesan (Providence Saint Joseph Health), Sherif Mityas (TGI FRIDAYS), hosted by Sanjog Aul.
What the conversation established
The conversation established that building a high-performing data and analytics team starts with agreeing on objectives and realistic expectations, not with hiring a chief data officer. Sherif Mityas of TGI Fridays said data is meaningless unless it drives decisions, described stitching POS, guest relations, and social media into a single view of the guest, and argued for early wins, starting with 1,000 guests rather than 50,000,000, treating that team as the voice of about 500,000 guests a day. Vijay Venkatesan of Providence Saint Joseph Health compared a chief data officer’s arc to Star Wars, said organizations must run, expand, and transform at once, and built a mobile app giving executives daily hospital performance using a minimal viable product mindset. Both kept data and IT closely linked, moved data engineering from ETL work toward data catalogs, and sourced scarce talent through external partners, university internships, and startups.
Host Sanjog Aul talks with Vijay Venkatesan, Senior Vice President and Chief Data Officer at Providence Saint Joseph Health, and Sherif Mityas, Chief Information Officer and Chief Strategy Officer at TGI Fridays, about how to build a high-performing data and analytics team. They explain why many leaders rate these teams low, why objectives and expectations must be set before investing, and how far a chief data officer can meet the hopes placed on the role. They compare how data and IT should relate, with Mityas keeping the function embedded in IT and Venkatesan describing his own job as an interoperability catalyst. They also weigh whether data engineering and analytics should be one unit, and how both leaders find scarce talent through external partners, university internships, and startups.
Contributors
- Vijay Venkatesan, Senior Vice President, Chief Data Officer, Providence Saint Joseph Health
- Sherif Mityas, Chief Information Officer and Chief Strategy Officer, TGI FRIDAYS
Host: Sanjog Aul, Founder and Host, CIO Talk Network
On this page: Key takeaways · Chapters · Explore more · Transcript
Key takeaways
- Sherif Mityas, Chief Information Officer and Chief Strategy Officer at TGI Fridays, said organizations swim in data but struggle because it sits in different spots, formats, and systems. He argued the first job is connecting the dots into one usable repository that gives a single view of the guest or the business.
- Sherif Mityas of TGI Fridays said data is meaningless unless it drives decisions, so leaders must agree on the end goal and how to measure return before investing in a data effort. He warned that without alignment at senior levels, individual functions push back and ask why they should change.
- Vijay Venkatesan, Senior Vice President and Chief Data Officer at Providence Saint Joseph Health, said a chief data officer must change how an organization thinks about data, not just organize it. He used a Star Wars arc of hope, inertia, and recovery to describe moving from information monarchy to information democracy.
- Vijay Venkatesan of Providence Saint Joseph Health said data teams must run, expand, and transform the business at the same time, like building an aircraft while flying it. Because executive attention spans run about 90 days, he shipped a mobile app giving leaders daily and predicted hospital performance using a minimal viable product mindset.
- Sherif Mityas of TGI Fridays keeps data and IT embedded and connected at the hip so insights reach marketing, operations, and guests as a usable experience. He keeps data engineering and analytics as separate skill sets in tight linkage, and treats the analytics team as the voice of about 500,000 guests a day.
- Sherif Mityas of TGI Fridays said the scarce data talent pool forces him to rely on external partners and act as a chief startup officer who finds new ventures first. Vijay Venkatesan of Providence Saint Joseph Health closes the gap with university internships and startup partnerships, and by appealing to millennials who want to improve healthcare.
Chapters
- 0:00 Introduction: building a high-performing data and analytics team
- 2:11 Why data and analytics teams rank low
- 3:58 The chief data officer role and expectations
- 6:16 Setting objectives before investing
- 9:16 Organizational maturity and execution
- 12:10 Setting expectations and early wins
- 18:44 Run, expand, and transform at once
- 21:34 How data and IT should relate
- 28:33 Key roles for a data team
- 34:16 Rethinking data engineering
- 43:31 From reactive to anticipatory
- 46:26 Building the talent pipeline
Explore more
- Data Analytics as a Team Sport
- Building Data and Analytics Driven Culture
- Empowering High-Quality Customer Experience with Data and Analytics
- Talks on Team Building
- Building Confidence in Your Team
- The increasing Role of Data Analytics
Transcript
Introduction: building a high-performing data and analytics team 0:00
Sanjog Aul [00:00:23]:
Our topic for today is Building High Performing Data And Analytics Team, and our guests for today are Vijay Venkatesan, Senior Vice President, Chief Data Officer with Providence Saint Joseph Health.
Sanjog Aul [00:00:44]:
We also have Sherif Mityas, who is the Chief Information Officer and Chief Strategy Officer with TGI Fridays.
Sanjog Aul [00:00:55]:
So the reason we wanted to talk about this subject, which is building a high performing data and analytics team, is because of some direct conversations I’ve had with global leaders on this front because, of course, we are trying to go digital and having a bit better control on data is digital. So whether you talk about the upstream data engineering or you think about the downstream churning, which is the analytics, both are equally important, and we are finding that there are challenges in both, and especially when you’re trying to get so much of data and volume variety and velocity to be handled and make sense of it and make it to good use for the company and your partners and ecosystem. That’s what we are up against. So how do we handle it? So that said, Sherif, I’ll start with you as the first question. With the range or rather the ranking that was given to the data and analytics team by many, many leaders that I’ve spoken with is around 2 to 3 out of a scale of 1 to 10. How come we are struggling with this and we are that low in terms of the ranking?
Why data and analytics teams rank low 2:11
Sherif Mityas [00:02:11]:
I didn’t really think it comes down to a couple reasons, and it’s a little bit of playing catch up because I think every organization out there basically swims in data. I mean, we have lots of data, especially if you’re in retail, hospitality, or really any sector that deals with consumers. The problem is that the data is all sitting in different spots, and it’s really hard then to understand, well, how do you bring that data together? How do you connect the dots? Because looking at just one set of data or one repository may give you a very different and potentially inaccurate view to draw insights from to actually drive decisions for the business, and so the first piece is really just trying to understand how do you connect the dots? How do you pull all this disparate data into one usable and informative repository that’s either the single view of the guest or it’s the single view of what’s going on in the business? And that’s really the struggle because all this data has sat historically in different spots, in different formats, in different usable fashion, and it’s really going back almost in time to understand how to put and stitch that data together.
Sanjog Aul [00:03:29]:
So when we talk about the Chief Data Officer’s role, while it would look like bring a Chief Data Officer and the problem solved. Now when you are looking at things, Vijay, and when such expectations are set when you arrive at that office, how do you manage that expectation, and to what degree can any Chief Data Officer live up to that expectation?
The chief data officer role and expectations 3:58
Vijay Venkatesan [00:03:58]:
That’s a very interesting question and a very thoughtful question, Sanjog Aul. So if you’re a fan of Star Wars and the old version, not the brand new iterations, I used to say it feels like that Star Wars movie. The first movie was called Brand New Hope. So when a Chief Data Officer comes in, there’s always hope and optimism. He’s going to be the one resolving all the ills of the organization around data. Then I call it that the second version of it is that The Empire Strikes Back, and The Empire Strikes Back quite hard. So that’s the inertia of an organization that says, you know what? This thinking about democratizing data is very hard. We were very happy with what we used to have, and we are fine, and then we have to say, the third incarnation of that movie was The Return of the Jedi. In some ways, we have to say, how do we take back the colonies over? How do we make sure that now we are living in a hopeful world? So in some ways, when a Chief Data Officer walks into an organization, they have to be able to make sure that there is this idea that you are there to change culture. You’re not just there to find pockets of data, organize it, but really understand an organizational culture about how they think about data, how they use data, and how they harness the data. Is data a utility or is data an asset, and what does that mean? So that’s how I think about these conversations is that it’s not a simple equation. It’s really a matter of how do you convince people? What can you do with the data you already have? And how do you extract value from that in a meaningful way that improves your business outcomes or operations?
Sanjog Aul [00:05:52]:
So, Sherif, not every organization even today has agreed or decided to bring a CDO, and when you don’t have a single top-level owner and the department that supports it, what are they thinking in terms of getting a handle on this data and analytics problem?
Setting objectives before investing 6:16
Sherif Mityas [00:06:16]:
I think the key piece here, and I think it was just mentioned, is you have to set upfront what you’re trying to achieve. So even before putting in a Chief Data Officer or putting a focus on data analytics, you have to step back and go, what’s the end goal? What’s the deliverable? What’s the outcome I’m looking to achieve by getting a better handle, a more complete view of the totality of the data information I have at my disposal, and then what am I gonna do with it? So data in and of itself is actually meaningless unless it drives decisions. Unless you can extract the appropriate insights and analyze that data to make real decisions that impact your consumers, your shareholders, or your organization, and I think at the beginning of that journey, everyone has to be aligned on what are those objectives, and how are we gonna measure the return on the investment to go get those objectives. Because none of this is free. Putting the data together. Understanding how to create it. How to hold it. How to analyze it. How to draw out insights. All those pieces require investments of people, usually vendors, and obviously time and energy internally, and so it’s that stepping back and making sure the entire organization at the senior levels are aligned upfront really paves the path. Really starts to create the opportunity for either a Chief Data Officer or just even an internal team that’s focused and dedicated on creating a very different and new and informative set of insights that will help drive the organization forward, but unless you have that, I think you’ll always have disparate, potentially misaligned and to the point made earlier, a lot of pushback from individual functions in terms of why are we changing? Why is this going to help me? Those fundamental things have to be solved or at least aligned upfront to pave the way for this process going forward.
Sanjog Aul [00:08:31]:
So we have been at it for some time. This data issue when we had the MDM, the master data management related issue where we had business users or business. Yeah. I’d call them business users who were not taking ownership of the data that they produce. They were not the custodians, and we could not get everyone to agree to that one version of the truth. Now comes this whole digital age where the same thing is happening, but you got way too much data, way too much variety and coming way too fast for us. So, Vijay, when you look at this, would you say that we have a lot of different flavors of strategies that we can paint on a document, but is the devil primarily in the execution? That’s where we are failing?
Organizational maturity and execution 9:16
Vijay Venkatesan [00:09:16]:
I would tend to agree with you. So I think one of the things that happens in organizations is, as I said, when you come in, you’re putting the right strategy on paper. You’re thinking about the right way to approach the problem, but where we often not pay enough attention is, is the organizational maturity there to execute on that strategy or the vision? So are there capabilities that are available within an ecosystem? So when I think about high performing analytics team as an example, in this brand new paradigm that we live in, which is data is 24/7. It’s all around you. It’s being captured in many different places. How do you harness that? What platforms you’re gonna use to harness it on? How do you make that available? And what are the skill sets you need in this new way of working with data and working with information to create insights that enables action? So to me, missing that equation of what is the army I need for the future? And so you can’t fight tomorrow’s battle with yesterday’s army. We haven’t really spent much time. Our organizations typically don’t spend much time thinking about what are the skill sets? What are the pieces I’m missing to execute? And that I have seen is the first barrier that you have to overcome is understanding the execution mindset that is required in this new paradigm of data, digital, and so on. So that’s the way I think about it, and the other part of it is to what was mentioned earlier is you have to get the organization to believe that there is value in these data assets, and we have to democratize data for us to see the value together. I always use the expression we have information monarchy in most organizations, and the idea is how do you create a constitution to create information democracy? And that really is also the heavy lift that you have to go through.
Sanjog Aul [00:11:21]:
So let’s talk about what the flavor that you gave, Vijay, was for some organizations which are struggling through this, and that is self-inflicted because they may not be investing or they don’t have clarity. So, coming to you, Sherif, I’ve seen organizations where they have a very senior sensible set of leaders who are trying to drive this. They have invested in it. They got the sponsorship from the executive management, but still barely they inch along, and even when you’re trying to execute, even when you’re trying to get people to buy in, they’re saying, yes, things are happening, but, man, that is so slow, or it is stalled. What can you point your fingers on?
Setting expectations and early wins 12:10
Sherif Mityas [00:12:10]:
The issue is, I think, as what was just mentioned is, it really comes down to sometimes setting the right expectations and having the right organizational executional capability. So you can in a lot of these, for instance, even with us here at TGI Fridays, we can see what we want to do with the data, with the insights. We have a team in place. We have partners in place. We can see the decisions we want to get to by understanding the complete view of the guest, for instance, in a more democratized fashion, to use the term just used, and I think the issue, of course, becomes there’s always complexity that you’re not prepared for. There’s always kind of the bumps in the road, so to speak, in the journey where you have to adapt and you have to adapt quickly. Or to the point just made, you might start losing parts of the organization. Well, that was an interesting exercise you guys tried, but clearly, it doesn’t work. Or clearly, it’s not, it’s only 80% accurate, and as soon as certain parts of the organization start seeing chinks in the armor, you have to resell the whole thing again, and so it’s really about setting the right expectation that’s saying, this is not a silver bullet. This is not gonna happen in 30, 60, 90 days. This is a journey, and it’s an evolution around really understanding and getting a handle on the information that we have, the data that we have, and again, for the purpose of making better decisions. So even in our world today, the information that we can use, for instance, we pull in data from, obviously, our POS. We know what people bought. We pull in from guest relations. We know what people talk to us about. We pull in from social media, so we know what they’re tweeting and posting on Facebook about us. All those disparate points of data help us get a more complete view and a very specific view of an individual guest. Now what we do with that, how do we partner with marketing, for instance, to create a very personalized message to an individual guest? That’s really when you start unlocking the value, and again, you don’t want to start saying you’re gonna solve world peace with this. You don’t want to say, I’m gonna be able to do this to 50,000,000 guests tomorrow, but how about if I could do it to 1,000 guests? How about if I just start there and start showing the value of what we’re doing? And understand that people need to go on the journey with you, and you can’t just unleash it at the end of the day. People have to walk down the path because there will be bumps. There will be things that go wrong. People just need to understand the North Star that you’re heading to, and to get early wins along the way will start enabling that kind of growing buy-in and momentum to getting to your final end goal.
Sanjog Aul [00:15:09]:
Sherif, to your point where we have to set the right expectations and get people to understand what the bigger picture and go towards that, go rather slowly but surely, let’s talk about a situation where this is all being done, but eventually the net result is a super slow movement, much slower than anybody’s expectation. It’s not that someone overreached and then later on, it’s coming and haunting them. It just is pure slow in absolute sense. What is it that could be causing it if we were to, basically deconstruct the problem related to execution or to build a data analytics organization or data and analytics team? What are those common issues that could be causing if you were a consultant trying to solve this problem? How would we tackle it, and how would we uncover or deconstruct the problem?
Sanjog Aul [00:18:10]:
So, Vijay, many organizations have had this issue, which is, of course, almost every progressive organization has this issue. There has been time spent on refining the strategy, getting common expectations, setting realistic expectations, but then, still, the things are inching along, and the rating, if you will, of the data and analytics organization is not improved. Correct.
Run, expand, and transform at once 18:44
Vijay Venkatesan [00:18:44]:
Organizations, I think there’s this thinking that everything is a waterfall approach to problem solving. In the world of data analytics, the way I describe it is that you have to be able to run the business, expand the business, and transform the business all at the same time. It’s like the old analogy of building an aircraft while flying in it. In some ways, that’s the mindset you have to have, and the reason you need to have that mindset is that people’s attention span are about 90 days or less. So anything over a 90-day cycle feels very long in terms of accomplishment. So one of the things that I did at Providence when I came on board was we started to change the discussion from retrospective view of information. So that’s like if you’re looking at management reporting from a week ago, a month ago, quarter ago to prospective analysis or action based information. So we did simple things like asking our executives, tell me something you want to know when you wake up tomorrow morning that you don’t have today, and they would tell us, I want to know what’s happening at our hospitals. I want to get an early alert system of how are we really performing. So we took that as a kind of a use case to say, can I build a mobile app that our executives can turn on, download the app? And every morning, it tells them, here’s my hospital performance, and, oh, by the way, here’s what it would be next week if these are the predictive variables that are part of it. So it was a way for us to change this concept that data can be very valuable, can tell you where you’re attempting to go, and how you’re going to get there, and that was the way we tried to transform the organization around this conversation. What we found was, if you do this fast and you do this with an MVP mindset, which is building a minimal viable product, people will start giving you the attention you need to focus on the strategies you have. So in some ways, it is having that entrepreneurial mindset to begin this conversation, which is how we did it at Providence Saint Joseph.
Sanjog Aul [00:21:02]:
So Sherif, when we look at the data organization as a whole, and I know it is since it’s data while it is not an IT function, but somehow it is assumed to be part of IT or in some way a big thick dotted line is supposed to be connected to IT. So say in your organization or in general, what do you think should be the kind of relationship which will allow the best functioning of a data organization? And what’s its relation to the IT department?
How data and IT should relate 21:34
Sherif Mityas [00:21:34]:
For us, it’s embedded, and if it’s not in other organizations, I think they have to be Siamese twins. So they have to be really literally connected at the hip. Because, to the point just made, without operationalizing the data. So giving that information to the folks so they can understand the hospital operations, for example, and in the example just stated, the technology team had to create something. They had to create an app, or create an interface or a dashboard that makes the data and the insights from that data usable. It’s got to connect to an actual experience for the end user, and so if those two functions aren’t literally connected at the hip, and understanding not only how I’m doing something, or what I’m doing, but how it’s being delivered, that disconnect is going to basically make all the effort meaningless, and so to me, personally, and what I’ve done in the organization is it’s embedded, it’s an integral part of everything that we look at, and as it relates to if we have data, how am I using it, and how is it being delivered so that the technology enables our end users or our guests or whatever function to look at and to use, at the end of the day. So to me, it’s literally got to be the same shade or the same color.
Sanjog Aul [00:23:07]:
So, Sherif, the question I’d have in this case, especially when you’re trying to build the upstream data function, which could be a data lake or any other form where you’re trying to make sense of the data, and, of course, there is nothing wrong with us having IT joined at the hip because we want to make sure the operationalization happens successfully. The challenge or the risk, and tell me if that is the right type of risk, that if you start thinking too much about the interface on how it’s going to connect to people, and that starts influencing what data ought to be for it to serve the business. Is that then not the tail wagging the dog?
Sherif Mityas [00:23:52]:
Personally, I believe it’s almost gotta be the same dog. Because if one group is just divorced and saying I’m just putting together data and data lakes, that should serve some sort of purpose without connecting directly to those that can help enable the actual utilization of that data, I think we miss an opportunity. I think they have to be in lockstep. It’s not to suggest that one drives the other. It’s to suggest that they are literally singing off the same hymnal. Because if, for instance, in my world, if my data team is collecting and filling a data lake as it relates to, let’s say, our consumers, and there’s no way or it’s not known or enabled to allow my marketing team or my operations team or even a guest to utilize the insight and the analytics that come from that data, then I just expended a whole lot of effort for no reason. We can all feel good about it because we have all this data stitched together, but if it’s not usable, then it’s pretty meaningless at the end of the day, and so that’s to me why it’s very imperative. Yes. There can be leaders driving data and analytics and leaders of technology, but if those leaders aren’t literally walking hand in hand, I think we miss an opportunity.
Sanjog Aul [00:25:24]:
So, Vijay, when I come to you and you’re wearing the other Siamese twins hat, like, based on Sherif, the data and the IT organization. What’s your view of the relationship? What would have allowed you to get your job in the best possible way and also get the best support from technology? What relationship structure would you have enjoyed?
Vijay Venkatesan [00:25:49]:
I think, as Sherif pointed out, it is a collaboration exercise, but I think one thing I would maybe offer a slightly different perspective on is that in the old paradigm, it used to be that collection of the data was a big enough effort that you had to figure out the how and what platforms and how do you build data warehouses and all that. With the technology innovation that’s out there with big data, cloud, and other things, the collection of data can be truly a back office function of just landing data, as you mentioned earlier, Sanjog Aul, in terms of a data lake, that’s an exercise that data organizations should continue to do and should continue to harness that data. If the other side of this equation, however, is a valid one, which is how do you figure out what are the business imperatives? And how do you make sure that you’re aligned with the business imperatives in terms of what context of the data that those businesses want? And how do you enable that? And then in terms of partnership with your IS, infrastructure teams or applications team, what you have to be able to do is to build what I call a human interoperability equation. Because at the end of the day, we’re all interdependent on each other. Technology is dependent on data and applications. The business, in terms of use and operations, is dependent on them. So the way I think about my role is my job is to be that interoperability catalyst. So I have to connect these pieces both at a people, process, technology level and then make sure that the context of the data is enabled in a very innovative way from an operational perspective. So that is the real hard work of how do you keep doing this on a daily basis, connecting teams, connecting leaders, and connecting processes is how I think about it.
Sanjog Aul [00:27:48]:
So let’s talk about the people, the specific people and the key roles that we need to put in place. So this is like not data analytics 101, but there are some mistakes made in the way you bring the initial team, and then you build a whole data organization who will take care of this whole data and analytics function. So, Sherif, if you were to start drawing a brand new org chart for that suborganization, which is the data organization, I think because in your case, it is embedded within IT, what are the key roles that you brought first or you think you would have brought first to give it the best shape and structure?
Key roles for a data team 28:33
Sherif Mityas [00:28:33]:
I think the point just made, the data collection side of this equation is much easier right now from a technology perspective, and actually, it requires less actual labor. So folks that are really managing what I’ll call the data collection, the data warehousing types, is a pretty small job. I think where you have to invest is really again on the data insight and analytics side, and not every organization can go to MIT and hire big data experts. So we at TGI Fridays partner with folks that are smarter than us that help us do that, but I think you really need some folks that really understand how to look at and manipulate information. We have some folks that are focused specifically on utilizing AI, for example. So how do we employ machine learning and artificial intelligence in the data repositories that we have to, again, draw and learn from it, and create the appropriate insights for driving decisions that we need to make for the business. So to me, I would start with the foundational pieces are there. You need some people to manage it, but if you’re investing in a couple key resources, it’s really around those insight, analytical type of folks that really can drive the next level of technology manipulation, which in our case is really being driven by our AI tools.
Sanjog Aul [00:30:25]:
Vijay, let’s talk about the data engineering side. So definitely on the analytics side, as Sherif pointed out, that we need to have some solid resources who could really make a difference. Now a lot of companies have reported issues with the data engineering where they are trying to even the sources that they have, there is some issue with the consistency of what data comes in and how clean it is and then how it all marries together and then is there one version of the truth, the standard data engineering function. That’s where people have expressed issues or they say because of it, my downstream function like analytics may not function well because garbage in garbage out. So once we come back, let’s explore what would that ideal data engineering organization look like, and what can we do to remove the issues which are reported which prevents the right data, the sacred data to become sacred and which is causing downstream issues.
Sanjog Aul [00:33:22]:
So Sherif was very successful in building the data engineering team, which is, of course, helping him support his functions, but, Vijay, coming to you when you look at this upstream function of data engineering, people have reported issues, and it’s not been as straightforward to have the data which is so clean, which has got a lot of integrity, and it’s always available, and what it is supposed to do for the business. It’s not always up to par, and people have faced challenges. So if you look from a CDO standpoint, does that fall under your stewardship? And if yes, what is it that you have to do in terms of the roles, the responsibilities, the workflow to make sure that this function becomes what it’s supposed to be?
Rethinking data engineering 34:16
Vijay Venkatesan [00:34:16]:
That’s a great question, Sanjog Aul, and a very hard one. So I appreciate you asking me that, and I’ll do my best to respond. Data engineering. I think if you look at the old paradigm of managing data, we used to call them ETL developers. The idea of an ETL developer was you have requirements back. You have a tool that’s very graphical user interface. All you’re doing is just mapping data from point A to point B, and then it ends up in a database called a warehouse or a datamart, and off you go, and that was the accountability. In the new paradigm, when we think about data engineering, it is to think about the various consumers of that data in a much more expansive way, and thinking and then working backwards to what is the skill set I need in order to develop that, and so when I think about data engineering, I think of them as not only just building and optimizing data from applications, but it’s also doing all of that housekeeping activities and having a little bit of an analytic mindset. Because the datasets that they’re building now are going to be used by data scientists, are going to be used by third party apps. It’s going to have API interfaces to downstream applications and so on and so forth. So we have to now think about what’s the right way to frame the context of the data into what I’m calling data catalogs. So our idea is that we need to be able to build this version of an iTunes store of data where people can then mix and match the data galleries that they want for their downstream consumption. So that requires thinking about how do I model the data? How do I manage the data? And how do I prepare the data and make sure it’s of the highest quality? And without compromising any user experience or user interface, and that is a skill that’s hard to find because you’re now dealing with brand new platforms, brand new interoperability challenges. So we find that we are figuring out how to build this capability by leveraging partnerships and also by trying to figure out if we can bring the right skill set in. So it’s a complex dialogue, but something that we feel we have to be focused on to be able to extract value from the data.
Sanjog Aul [00:36:40]:
So would you say there is something you have to do to attract the right talent, as the talent itself is in short supply and you’re not able to attract them? Or when they do look at how you’re trying to deal with this data engineering issue and they find cludginess, that’s what drives them over,
Vijay Venkatesan [00:36:58]:
So healthcare especially, I’ll just speak from an industry perspective. Healthcare data is very challenging because the ecosystem of healthcare is very fragmented. So you don’t control all of the data that comes within your ecosystem, and so what happens is when people see that challenging, challenges of how do I interpret this data? How do I synchronize this data? How do I connect the data? In the earlier part of the discussion, we talked about do I have a one version of a customer defined? We have to define one version of a patient, and that’s very hard to do and the patients don’t come into all of our touch points. So there’s a lot of data fragmentation challenges, and most of the skill sets, if you’re coming from CPG, retail, and financial industries, you are just struck with the enormity of the data complexity challenges, and then the other thing you also find is that in order to get to that single version of something is an iteration and not a once-and-done conversation. Those are some of the issues, and then last but not least, healthcare is still, for not-for-profit health systems, our pay scale isn’t comparable to what the industry is. So we have to attract people who are enamored with doing something for others and have that selfless attitude as they come into an organization like healthcare and not-for-profit healthcare. So those are some of the things we’re tackling.
Sanjog Aul [00:38:35]:
So, Sherif, when you are looking at this whole idea of the team being brought together, the downstream and the upstream function, should they be basically one unit or do you want to keep them separate because you don’t want to have analytics drive how data is harnessed?
Sherif Mityas [00:38:57]:
I think they’re separate. They stay separate in my mind more because of the skill set than the intention of what they should be doing, and what I mean by that is I think they’re very different in terms of what skill set, what talent, what experience those two different areas have to do to be successful, but again, by creating what I’ll call a very tight linkage, we make sure one doesn’t get out of lockstep with the other. Because to me, it’s still incredibly important to ensure that whatever we’re doing, the what is connected to the how. To ensure that we’re always looking at that end state goal, that use case in mind, that decision we need to make from both not just collecting, compiling, cleaning, ensuring we’ve engineered the data correctly, but then also the insight and analytics that go into creating a very usable and a very dynamic and a very thoughtful outcome.
Sanjog Aul [00:40:15]:
So in analytics, as you see, and I’m sure you would have done it in your organization, once you start giving some insights to the business, they just keep asking for more, and then analytics team can totally go crazy trying to support and keep delivering to the business. That is a totally different type of mindset, a totally different type of function than a little more regimented function of data engineering, where they should not, at the time of collecting data, be always trying to align to what the next business insight someone is looking for. Because at the end of the day, there is a limit to how many ways you can pull data out. How many ways you can crunch data is one thing, but the data stays data. So would you say the mindset of the people who are doing this, data engineering, should that be at all influenced by what the next best analytics report that your business would be asking for?
Sherif Mityas [00:41:14]:
I think it has to go hand in hand a bit though again because if we’re too pure from just a pure data perspective, if we’re trying to be too isolated in terms of don’t bother us with what you’re trying to solve, we are only focused on the data, I think you get too disconnected from the real-life use cases of why you’re even collecting that data. So to me, the kind of what we’ve done at TGI Fridays is I’ve kind of made that team the voice of the guest. So we’re very big on trying to connect the dots with all our 500,000 guests that visit one of our restaurants every day, and so if they can be in any meeting with it, whether it’s with our marketing team, our operations team, our field team, if they in the room are the source of truth. Everyone’s got an opinion in that room, but that team is the voice of the guest. They’re the voice of truth. Regardless of what insights we’re asking for, regardless of what decisions we’re trying to make, everything bounces off, well, what does that team think? Because that team is the source of truth. If we can create that, and that’s the mantra I put on that group, then all the requests are fine within reason, but understand that we’re not being biased by those requests. We are coming with the facts. We are coming with data and behind us that says yay or nay to what you’re trying to do or what you think is happening out in a restaurant or for a particular guest, and that team can remain pure, but also singularly truthful, if that’s possible. As much as the data allows, that’s really what I want that team to be able to deliver. Is that they are the voice of the 500,000 guests per day that are not in that room with our other functions.
Sanjog Aul [00:43:13]:
Wearing the hat of a data leader, Vijay, when you look at the same question, what would be the level of isolation and independence you would seek from any of the downstream functions for you to be able to do your job?
From reactive to anticipatory 43:31
Vijay Venkatesan [00:43:31]:
I think I’m going to build on Sherif’s answer and say that part of what we think of ourselves, at least in my current role and how our staff thinks as well, is we have to be anticipatory. One of the things that used to be the old way of doing things was we’re always a reactive organization. Data functions were generally reactive. So we used to call us the order-taking team. You pull up through your drive-through window. You say, I need a cafe latte with whole fat milk and blah blah blah. We have changed that paradigm to say we’re not just an order-taking business. We are a value-enabling business. By that, I mean we have to anticipate where is the business going? Where is our organization going? What data matters in that context? And how do we become a little bit more anticipatory? So I strongly believe that the voice of the patient, which is what our team should represent, and the voice of the customer, which is the caregivers who enable that care for the patient, if we start with them as the focus and how do we become anticipatory in that sense, then we will do a better job of bringing in data, whether it’s internal data or external data, and harness it and make it available so that we can constantly meet the emerging needs of the business. Because as a data function, I believe if we’re not anticipating the next 2 or 3 things, then we’re not doing our job, and that’s a mind shift of how do you change that thinking from reactive to proactive? And that, I believe, is going to be critical for people like me in my capacity or any data organization to be thinking about the future constantly and not the past.
Sanjog Aul [00:45:26]:
Now talking about the very source of this talent. So talk about data engineering, yes, some of those functions we have had in legacy environments which we have in IT so we could leverage them, but then analytics, to some extent, again, we can say it, but then when we talk about data science and other downstream functions, those are definitely new types of animals who we need to somehow find and incorporate in our team, whether as an outsourced partner or even as inside talent, but also on the data engineering side. So, Sherif, when you were to look at a constant pipeline building, because this is not going to stop-this function data engineering as well as analytics is only going to grow-and there will always be outside forces which will try to pull good talent out of your organization. What are you doing to build that pool and maintain that pipeline so that you are never short on staff and capability?
Building the talent pipeline 46:26
Sherif Mityas [00:46:26]:
That’s the million-dollar question. Because that talent pool, to your point, is in high demand, and it’s not a very big group of individuals today, and so really for us, if you can’t go and spend high six digits on getting the best talent, you’ve got to be able to really rely on, in our case, external partners, and for us, that’s worked pretty well because we can draw in the expertise and the insight, get the knowledge sharing that we need to our internal team, but then stay relevant in terms of what’s next. So I kind of always challenge the team in terms of what are other people doing out there. There’s so many new startups and new talent out there that are kind of pushing the envelope in terms of what else we can do around data and analytics. Trying to build that internally is very difficult and very expensive, and so for us, it’s really trying to partner and really stay out in front with who’s doing the next big thing. Is there a use case that makes sense to go try it? Is there something that allows us to continue to evolve our capabilities around the data and analytics function that others are doing? Whether it’s from academia or from entrepreneurs that have new startups. It’s really that ability to stay networked and connected in this space that also allows you to differentiate. Because everyone’s trying to do this, not just across sectors, but obviously our direct competitors, and so it’s how do you create and how do you use something that’s new and different that no one else has, and that’s the challenge. Because whether you’re trying to retain these people internally or trying to find the next great partner, who gets there first and who does it faster is going to win this game, and so there’s no easy answer to that question. It’s a continual battle to make sure you’ve got the right people either internally or connected from a partner perspective to make this happen.
Sanjog Aul [00:48:36]:
So you’re saying that either the CIO also assumes a headhunter function or somehow figures out a pipeline outside because HR may not always come through when you’re trying to get this level of talent.
Sherif Mityas [00:48:48]:
Absolutely correct. As the CIO, I have to be almost chief startup officer as well. I go to places. I stay connected to networks, because I need to know who else is starting, who are even now in stealth mode that haven’t got their initial seed money yet, but they have a good kernel of an idea. They have really smart people that may have come from Google or Amazon or MIT that are starting a new venture, and you want to be their first client. You want to try something first that no one else has, within reason, but those are the types of things that continue to help drive a competitive advantage.
Sanjog Aul [00:49:31]:
So, Vijay, when we look at the kind of skills that we need at the individual level or the team level, like the benchmark which we would like to set for all the people in the room so that we do not go below, what would those benchmarks be in today’s day and age and what we see coming tomorrow?
Vijay Venkatesan [00:49:50]:
So I think the benchmark for me, at least, is where are other industries in comparison to our own? So one of the things we have to do is to be more aspirational, and regardless of the industry we’re in, we have to say who are the leaders in this space, how do we figure out kind of where they’re at, what it’ll take them to get there, and so just to kind of add on to how we’re closing the resource gap, one of the things we’ve done very well is we have internship programs with local universities. So specifically, if they have a data science curriculum program, if they have a data engineering program, we now partner with our local universities to say, can we be that internship hub for them? Can we train the talent within our industry domain? And hopefully, that’ll be the pipeline by which we hire the future generation engineers and data scientists. So that’s one deliberate strategy we have. The other one is really trying to appeal to the millennials and say, hey, do you want to change the world? And healthcare is the domain where we strongly believe we need brand new ideas. We need that entrepreneurial spirit and talent. So that’s another way-we’re really going to the hearts and minds of the young to say, can they come in and do something for the common good? And then the third thing is actively working with startups and entrepreneurs out there to say, which part of the gap can they really help us resolve? So is it a data science gap? Is it a data engineering gap? Is it an analytic platform gap? How do we think about these things and figure out where is our strength and where are their strengths, and how do we marry the two? So we’re kind of doing an all-in strategy, if you will, to make sure we’re prepared for the future.


