Data makes the world go round. And with the rise of artificial intelligence and machine learning, data is likely to become even more critical in business. In this episode of the Data Point Of View Anshuman Banerjee, a Domain Chapter Lead of Commercial Data at Spark New Zealand joins us to discuss why data plays a key role in understanding customers and how to fill in the gap between business and data.
Data makes the world go round. And with the rise of artificial intelligence and machine learning, data is likely to become even more critical in business. In this episode of the Data Point Of View Anshuman Banerjee, a Domain Chapter Lead of Commercial Data at Spark New Zealand joins us to discuss why data plays a key role in understanding customers and how to fill in the gap between business and data.
Mobilewalla - Data Point of View - Anshuman Banerjee - Transcript
[00:00:00] Laurie Hood: And we will get ready to go. Thank you for listening today. I'm Laurie Hood, CMO at Mobilewalla, and this is Data Point of View. Data Point of View is a podcast for anyone interested in using machine learning and consumer data to achieve their business objectives. Joining me for this episode is Anshuman Banerjee. Anshuman is Domain Chapter Lead, very interesting title, and for Commercial Data Spark New Zealand. Spark is New Zealand's largest telecommunications and digital services company with a strong commitment to using technology to be truly useful to their customers. Anshuman, welcome to the program and thank you so much for joining me today. I'd like to start talking a bit about your background and what led you to your current role, and then focus on how you and your team are driving change at Spark. Then we'll close with your thoughts on, on what you're thinking of for 2020, [00:01:00] which I, it's frightening to think that we're talking about 2020, and it'll be here in a minute. So let's go ahead and jump right in. So your undergraduate degree is focused on technology. Is that something that you've always been interested in?
[00:01:17] Anshuman Banerjee: Hi Lori. Firstly, thanks a lot for inviting me for the session. It's all my pleasure to be here talking to you today.
[00:01:24] Laurie Hood: No, thank you for your time.
[00:01:26] Anshuman Banerjee: No problem at all. Really excited for the session. My, my background, an undergraduate degree. So, the way I came about it, it's not actually too much of a choice back in, in India when I started off where technology is something that you go into if you are moderately good with your studies. And that's how I went into technology to an engineering degree. I ended up though in software. Like, a lot of people in my generation have gone to engineering, have gone into software. From then on, I worked in software for [00:02:00] quite a bit. Technology, really loved it. Loved coding, loved going into clients and, and talking to them and solving technology problems for them. After a bit of time I felt that I really wanted to specialize on the business side. And so I did an MBA and, and from then on it has been a journey which has been a combination of business and technology. So yeah, I'm passionate about technology and I like to use it to solve real business problems.
[00:02:28] Laurie Hood: Talk a little, we'll talk a little bit about the decision, because there was, there was a little bit of a gap between getting your undergraduate degree and then going back and getting your MBA. And as you just said, you know, you wanted to kind of broaden and look more at business knowledge. So talk about having the technology background and then this business focus has really helped you with kind of your career decisions and what you do today.
[00:02:57] Anshuman Banerjee: Yeah, sure. A lot of things I think that we [00:03:00] do is, when we look back at things and connect the dots, then we see some value in them. And that's pretty much how it has happened for me as well in the sense when I look at solving business problems, and I'm really interested in using technology data, et cetera, to use, to solve those problems. And, I think if you work for a bit of time in the technology space, it brings in US the amount of ability to understand what's possible and what's not. At an overall level, I wouldn't probably term it as like a no on because the world is changing too fast with a lot of venture and new, new companies coming in so it's difficult to predict what's going to happen. But a general sense of it, as to how to solve it using information, using technology is what gets ingrained into you after a point in time my fee. And, and that, that really helps, to, to understand how you, how you approach a problem and what are the different avenues that are possible to [00:04:00] do to solve a business problem. So looking back at it, I think these are the two things that I love, or three things that I love, which is business, which is data and technology. And somehow I feel that they have, they have come together in some way, shape or form.
[00:04:18] Laurie Hood: No, that's great. Well, also, I wanted to talk a little bit about, you know, you have deep experience in the telecommunications industry. So what is it, what is it that you find so interesting about Telecom?
[00:04:35] Anshuman Banerjee: Again, I, when I went into telecommunications, very early on in my career, it was as an IT consultant, but I wanted to specialize in the telecommunication space because of the inherent nature of connectivity, kind of excited me. And from then on, it has been, it has been a choice to stick to the sector and I've seen quite a few ups and downs that have happened. [00:05:00] At some point of time telecommunications got highly commoditized. And, but then today I see that a lot of innovation is happening in the telecommunication space. So, yeah, it's been, it's been quite a nice journey and I don't intend to, if I can stick to the sector, I intend to be in this area, with all of the new things happening in 5G IoT and the enormous amount of information that exists in the telecommunication space, is something that really excites me. And I think we have kind of, only exploited the top layer of it. There's a huge amount of information and data that can be exploited for good in the telecommunication.
[00:05:41] Laurie Hood: And even though you've said, there's more that you can do. Telecommunications feels like an industry area that has really embraced technology, embrace the use of data. You're seeing a lot of sophisticated applications coming out of Telco. And, and [00:06:00] I really like your previous statements about kind of that combination of understanding the technology, but having the business perspective. And, as we sort of lead into talking about your current role in Spark, I would expect in, in a data science role, that to really be important because everything you're doing is really to serve reaching the business goals. It's not technology for technology sake. So when, you know, you've been at spark now for over seven years, you know, talk a little bit about the company and, and what some of their focus areas are in terms of growing their business.
[00:06:45] Anshuman Banerjee: Just give you background about, and how it, how it leads into my time at Spark, is that I have, I have typically been on the consulting side prior to this mostly, and have typically [00:07:00] work at, at a specific client for shorter durations. Six months, max one year, et cetera. Now, Spark is a place where I've stayed back for about now 7+, 8 years. And, and one of the reasons which just kept me at Spark is, I, I've understood that my thinking is that I want to work with people who are smart and smarter than me, and, and people who love to, love to experiment and try out new things. And I've had the opportunity to kind of do both of that at Spark. My role at Spark has changed significantly overtime, and with working on different things. At some point I was doing some kind of internal consulting at Spark, trying to solve problems. And then from then on, moved into data. And what I also see is, the leadership believes in changing things around quite a bit. They have got the vision, the people I talk to, my direct leadership has that vision. [00:08:00] And I like working with them because I feel that there's always something that's there to challenge me at Spark. Now this role on commercial data and a domain chapter lead position, the way it came about was typically what we saw at Spark is, we had these groups of people who understood data quite well, and we had a group of people who understood business really well. But the interconnect between business and data was, was missing. It was more business detecting data to create dashboards. And that's what data teams used to do. And, I thought that that potential of what's possible via data was not clear either to the data group or to the business area. And then this, this kind of was a gap that was identified not just by me, but by a few other people, by my management as well. And that's how this, this area got formed, called Commercial Leader, [00:09:00] which was about how do you use data to drive commercial benefits for Spark. And, and we work on internal Spark projects and drive commercial outcomes. Each of the projects that we, that we work on has either in your benefits linked to it, or longer term incubation projects which have benefits, which should be available, it should be, should come in in the next few years. The two different kinds of projects that, that we work on. And it keeps me on my toes from a quick immediate initiate delivery, as well as, an explore perspective to it as to what's possible in the future and putting my time and effort into, into some of those things, long-term incubation projects.
[00:09:45] Laurie Hood: No. That's awesome. So I want to touch on a couple of things that you mentioned. I love how you talked about the importance of your team and the leadership. And I think, you know, we're, we're also influenced by our day-to-day at [00:10:00] work and, and that ability to enjoy who we work with, who we work for, and the projects that we're working on. So I've just really appreciated that when you were talking about your team and again, thinking back to your earlier comments about this melding of technology and business, you know, I don't know how unique it is, but I think it's genius to bring kind of both sides together. Because nowadays it's hard to know everything and, and your data teams and your data scientists go so deep in one area and the business people go so deep somewhere else. So it seems like it's just this all peanut butter and chocolate, Reese's, peanut butter cup combination of, you know, how do you bring the two teams together and get more out of it. I wanted to touch on your, your title of Domain Chapter Lead and is that kerman common terminology at [00:11:00] Spark? How, how did, how did that title come about? I think I understand domain. I'm not sure about the chapter part, so I'd love some insight into that.
[00:11:08] Anshuman Banerjee: Okay, sure thing. I think, I do have difficulty explaining it sometimes. Especially when I put it on my LinkedIn profile and people think, "Okay, what is this Domain Chapter Lead? What does this even mean?" So as far, so we've gotten, we've we, we went to Jay about three years back and almost 80 to 90% of the company works in the J mode. And we've followed the Spotify to coach. Now in Spark the teams are classified as, as Tribes at the very top level, Tribes and Chapter Areas. Tribes are more areas where there is a delivery. Associated with that, the delivery objectives. But the Chapter Areas more of skillset and people management areas. Now, the Data Automation is a Chapter Area, as well as a Tribe. Because as a Tribe, we have deliverables from an infrastructure [00:12:00] perspective, as well as incubation of projects why incubation of new initiatives on data, why a chap Data Automation is also Chapter Area. And within the Chapter Area we've got four domains. One of the domains is data engineering, there is data analysis, there's one on automation and the fourth one is on commercial data. So I lead the Commercial Data one, and within commercial data there are two chapters, two types of chapters within it. One is data science, which we all know what data science is. And the data intelligence is another Chapter Area, which is about the business and data interface. These are people with special business backgrounds and an ability and a love for using data to solve. So Chapter is more of the skills, et cetera, part of it, while Tribes is the delivery part. And, I leave those that, that domain on commercial data. [00:13:00] I don't know if that made it any clearer or confused everything.
[00:13:05] Laurie Hood: No, it, it, it makes sense. And it's an interesting perspective. So when you look at your team and you think about goals or those goals shared across the sides of the team, then potentially with some other kind of independent goals on each side, how do you, how do you drive kind of that working together?
[00:13:26] Anshuman Banerjee: Can you repeat that question please? Sorry.
[00:13:28] Laurie Hood: So when you look at kind of the specific goals and objectives for your team, are they shared across the team or are they separate depending on which Chapter you're, the person's in?
[00:13:45] Anshuman Banerjee: I think a lot depends on, a lot depends on the area of work, whether involved in. Now the area of work that the involved in is, is a game, there's another terminology to it that we use called Squads, where [00:14:00] these are small teams, which are small entrepreneurial teams of eight to nine people where a delivery happens within a Tribe. And people from the Chapters are part of these Squads. And a Squad can comprise of people from data intelligence, data science. It can have people from marketing, supply chain, depending on the kind of initiative and the outcomes that they're driving towards or working towards. So their objectives are set at a, at a Squad level in terms of what they want to achieve so they're working towards a goal. In terms of Chapter standards and, and ways of working, et cetera. the Chapter plays a role in terms of hiring the right people. That's where the Chapter plays a role.
[00:14:45] Laurie Hood: So the Squad brings kind of another dimension into maybe kind of these cross-organizational team to accomplish a goal?
[00:14:55] Anshuman Banerjee: Correct. Exactly.
[00:14:57] Laurie Hood: So, so it sounds like it's, [00:15:00] it's some kind of a matrix organization or at least working on these projects. What do you think is the key, because so many companies are now highly matrix. What in your mind is kind of the key to, to success with kind of, with these cross-functional teams and, and how they get things accomplished?
[00:15:23] Anshuman Banerjee: I think that's a very interesting question and the answer can be on so many different dimensions on that. One of the key things I believe is, is the people competent of it. And who are the people who are part of a Squad, who are the people that we are hiring. And, I personally am of the view that we need to increase talent density, because a lot of the time I've seen that if I hire great people, it lets me do a lot more meaningful work than I otherwise would be. And so, there's a component to the kind of people that we hire for these positions, and for a [00:16:00] Squad kind of a structure to also work. I feel that people need to have a certain degree of drive and, and ability to be self driven rather to, to achieve and try out, try out an experimentation approach to things, which is what we follow in these, in these areas. And, it depends on the Tribe that they belong to and the Tribe Lead and how they drive it. It also depends at the top-level, at leadership level, in terms of the degree of faith that they have in this kind of structures to make them work. Because a lot of thinking and decisioning about how we achieve something rests with the Squad, and rather than being told from the top as to how you need to achieve something.
[00:16:43] Laurie Hood: Well, and there's definitely, there's, there's a personality type, I think who's more comfortable working in with that kind of model and, and it does sound like it's so much more empowering, but you're right. Hiring the right people is so [00:17:00] much at the success of different teams. So when you, so shifting gears a tiny bit, you're in a highly competitive industry. So without revealing any trade secrets, 'cause I'm sure there's a lot of secret sauce that you guys are working on. You know, what are some of the projects that you've, you've been a part of that have really interested you, and it's sort of, what is your thinking as you lead your team to tackle these?
[00:17:28] Anshuman Banerjee: I think a lot of it is about, is about understanding customers. And I think that, that is something that we try to keep at the center of, of all of the work that we do. That how does it, how does it enable or help the customer to, to succeed? At Spark, our vision talks about to meet all New Zealanders, win big in the digital world. And, and keeping customer at the center of it. Understanding when a customer needs something. [00:18:00] And, and what do they need. And being able to anticipate that, using information and data is something that we feel can make us a lot more relevant in terms of the conversations we are having with our customers, as opposed to a world where it's driven by the company where we have our objectives to reach, and we bombard customers with, you know, numerous offers, et cetera, and without doing what really works for them. So on that journey we have moved quite significantly in terms of trying to understand customers better and then providing them with things that we think that they might need. So, well, a lot of our projects are oriented towards that. But at the same time we have, we also have a number of projects which are about, how do we make our internal processes and practices a lot more simplified and an experiment oriented so that the degree [00:19:00] of bureaucracy, et cetera, goes down. And there are a lot of just-in-time decision making that can be enabled in the company.
[00:19:08] Laurie Hood: Did you find, I think a lot of people see the telecommunications industry as being kind of a pandemic winner, if you could talk about anyone winning during the pandemic. How did, how did that, did that potentially shape or influence, you know, some of the things you were doing when you talk about breaking through the bureaucracy, you know, it just, it created much more usage and pressure on the telecommunications companies. What, what was a little bit of that internal impact for you?
[00:19:42] Anshuman Banerjee: So internal impact, we've had, so everything about the pandemic has not been great for the telecommunications industry. For example, we've lost out enormous revenues from a roaming perspective. And, and if you know about New Zealanders, they love to travel. They love to go around the globe. [00:20:00] And, we are a country where a lot of people come in as well. So, on, on the roaming, et cetera, front, there has been a substantial decline almost to a point to zero for us. But at the same time, the degree of connectivity that has been required for people to work from home and to work from remote settings has seen huge jumps in terms of the amount of data usage, that has happened. Now, does it come with increased amount of revenue? In some cases. But it also, if the challenge of the capacity utilization of our resources grows substantially, all of a sudden. So we've been up to that challenge and we've been trying to manage and maintain, make sure that we are up to it and people have absolute levels of, you know, they have connectivity during these trying times. In terms of our ways of working, we have been, as a team, thankfully being in a company which works on, on telecommunications and mostly on the IT, et cetera space. The degree of impact for our employees [00:21:00] is more about being used to work in from. So that, so the level of impact is probably lesser compared to a lot of other sectors where people. So, but the element, things that I see, which, which I see as, have impacted in the ways of working also is that, I love white boarding. I love being able to be in a room with, with a team and, and share ideas and thoughts and come up with new ideas. And I, and, it's been, it's been difficult to do that. And I'm, I'm really waiting for things to open up as our vaccination rates are rising and we are hoping that we'll be able to go back to office and hopefully January.
[00:21:39] Laurie Hood: Oh, that's great. Now we, we talk about that a lot internally. I mean at Mobilewalla we have data science teams and technologists, and just the, the, the innovation that comes out of that interaction and just being in the same room, it is, it is hard to replicate when you're not. [00:22:00] So talk a little bit about, as your team has grown, because I know you guys have really invested in, in data science, what have some of the challenges been as your number of data scientists increase, your number of models increase in sort of, how have you had to change your thinking as you've seen growth?
[00:22:26] Anshuman Banerjee: That's really interesting question. 'Cause when we started out with, we had maybe one Squad which was creating these AI machine learning models. And it then went across to two or three Squads. And, we did not really know how to scale at that point. And the result of which an implication of which was that, there were data features, which are required for models, which would have been independently created by each of these Squads, and in data science, 80, 85% of the work is in feature creation. As we have grown, though, we've [00:23:00] understood that there needs to be a certain degree of streamlining of that piece and we can, we need to get faster in the way we can create models. So we've created a universal layer called of, of features, which we are calling as a Feature Store. And it's a layer where features are tagged and marked out as, as reliable across the company, which means that any new models which are coming up can immediately tap into this Feature Store and create models using those features. Already our model creation timelines have, have gone from over two months to almost a few days in which we can turn around models, and that's been an enormous benefit and has helped us scale. The other thing that we have done is as this AI models get embedded across the organization, we want to make sure that the models are, are, credible and delivering the results that we wanted and that they haven't got rogue. So we have now [00:24:00] a Squad, which we call is, AI Governance Squad, which looks into and ensures that we are tracking model health and data health for these models on all of his own basis. And that's, that's important for us, to ensure that these mod, in addition to that, we're also looking into if our models have certain degrees of bias and if they're fair, et cetera. All of those things are being put in place at this point to make sure that the degree of credibility that we have of our models amongst us, as well as among senior management, remains high.
[00:24:35] Laurie Hood: Yeah. I mean, you see so much. I was looking, it was financial services, but there was an interesting article today about a bias in AI and a degree of transparency in some of that, that kind of decision-making. Also, you know, we, we talk a lot internally about building reliable models and, but you fundamentally have to understand the [00:25:00] reliability of that data that you're incorporating in your features. And then I think it's great that concept of sharing features because when you've got a lot of different teams, there may not be that communication. And it's such a big part of the process and expensive part of the process and an Aeroplan part of the process. So why not reuse something that's working? So that's awesome. So we, we are coming up on time, so I wanted to just close with, as you look at 2022, what are, what are the three things that are kind of on your mind as you think about, about what's going on with your team and what company goals are coming out of the pandemic? What does that look like for you?
[00:25:46] Anshuman Banerjee: I don't know if I can talk about three things sorry, but, but things that that I think about is, I always try to think about what are the new things that are going to come up in, in this AI machine learning space and [00:26:00] how do we, how can we make sure that we are on top of those things and trying to bring some of those things into Spark? So we're planning for our FY20, I mean FY23 at this point. And what are the new projects in terms of what we should incubate and start working on, and creating, creating a list of, of those initiatives, which is really an exciting part for me., which is future-looking et cetera. I, I see that the importance of data and information is going to become ever more important. So how do we scale a team, and how do we at the same time? So we've got a team of highly talented people, and, and it's a challenge to bring in more people and maintain the same it costs and maintain the same degree of drive, et cetera in the team. It's also sometimes, it's also another thing that I think about is I, when I'm bringing people who are highly capable, they want to be challenged and they want to work on things which are interesting. So having a [00:27:00] pipeline of initiatives and projects so that they can be utilized, they don't get bored is, is another thing that I think about. Yeah, so these are some of the things.
[00:27:11] Laurie Hood: That's great. And, and I, I like and appreciate your statements on, you know, what's next and how do you, how do you stay current. It feels like the sort of AI and machine learning space right now is, there's a lot of investment, technology, automation and, and a lot of things going on in it that probably do make it, make that a need to really stay current. And then with all of the discussions around data, existing data, first party, third party, how do you enrich it, how do you use it. I mean, data is the fuel for a lot of machine learning and AI. So, so I know you guys have tons of it, but is it the right data, and is it usable? So there's a [00:28:00] lot of exciting stuff. Well, I want to thank you so much for joining us today, for sharing your insights. I mean, you've got this great experience and you're doing a lot of exciting things, and you're really leading, leading your team into kind of new territory, I'm sure, at Spark. To our listeners, you know, thank you for your time today and please join us for another episode of Data Point of View, brought to you by Mobilewalla. So thanks Anshuman. Appreciate it.
[00:28:29] Anshuman Banerjee: Thanks a lot, Laurie. My pleasure. Cheers.