Episode 8: Roshan Awatar

AI Transcription:

[00:00:00] Ben Culora: Hi. This podcast is brought to you by Practicus. Practicus is a recruitment, consulting and advisory business specializing in change and transformation. We hope you enjoy the podcast.

[00:00:13] Barry Panayi: Hello. Welcome to another episode of The Data Files. You’ve got me, Barry, and Tony. And today we’ve got Rash Awatar, who’s currently the group director of Data at Sky. Previously the group chief data and analytics officer at Lloyd’s Banking Group and spent a lot of his career in consulting, Accenture and EY, delivering embedding data strategies and has been at the heart of this big data phenomenon over the last couple of decades. Thanks for joining us Rosh

[00:00:42] Roshan Awatar: Thank you Tony Barry for having me here.

[00:00:44] Barry Panayi: This episode’s all about, governance and ethics, although I suspect we may stray into other items. I know, Tony, you’ve got a particular love for governance.

[00:00:54] Tony Cassin-Scott: I do. Some people think that data governance and data management’s dull as dishwater. I’m obviously not one of them and I hopefully nor you, but on that note, Why should I be excited by it?

[00:01:05] Roshan Awatar: Interesting. So it’s not, um, an unfamiliar perception that I’ve come across here. So, um, with data governance, people think about policy and control. I think about enablement and data governance is hard and it’s typically being seen as, I guess, less exciting because it’s attached to regulation versus, a value driven mindset or business outcomes mindset.

So it’s very much on the defensive angle typically, but you know, the narrative is changing. And how I think about it is attaching it to those business outcomes. If we think about what’s hot topic in all different industries, personalization. If we think about how do we get operational efficiency, how do we, um, how do we increase customer engagement?

All of these are great. People associate analytics, machine learning. But all of that is enabled by governance. You can’t get the data right. But typically we don’t, you know, we haven’t really matched those cases up. It’s always been in response to a regulation, in, in response to a policy. So it’s not something that’s being proactively driven because the business think they need it.

It’s because it’s been, some people see it as forced upon themselves.

[00:02:17] Barry Panayi: And that’s interesting. Cause of course you have, you’ve been on, two sides of the fence at Lloyd’s Banking Group. Well, we both have had the same job at Lloyd’s Banking Group, and of course that’s very much the defensive re responding to regulation, whether it be to making sure there’s no ICO complaints or the FCA regulation, or looking at specific ones like BCPS 239.

The words I wished I never had to say again, since I left. You, you talked about the concept of defensive strategy and offensive strategy, and of course now you’re out of financial services, albeit in a different regulatory environment, but I suspect, is it easier in your view to make the, the case for governance outside of financial services? Or are you finding the same things?

[00:03:01] Roshan Awatar: Again very, very interesting, uh, because the regulatory backdrop really makes a difference. Okay. When you’re in it, you think something, when you’re out of it, you think a different thing. So being in financial services previously, oh God, there’s so much effort, There’s so much policy, so much regulation, but actually it’s a blessing in disguise, because it’s forcing you to think about the stuff that actually he’s put on the back burner a bit. But when you step out of a highly regulated environment, it’s more around, well, value driven. How do you focus things on delivery? Right? Delivery driven organizations, value based organizations, you know, profitable organizations, they don’t wanna be slowed down, right?

But they do wanna do things quickly. They wanna do things right. So how can you actually attach governance, within delivery. And that is quite difficult if you don’t wanna be slowed down. So you no longer have a policy stick, you have less of a regulatory stick. So to your question, it’s not easier, it’s just as hard, but a different context because it’s, I guess the story is different.

The story isn’t about regulation. The story is much more about enablement and why you actually need it. So I think it’s actually easier in a highly regulated environment.

[00:04:13] Tony Cassin-Scott: I think that’s true. I mean, I’ve seen it used as a kind of keep out of jail type policy, which most governance is, is born out of. The, something which I came across recently, I was talking to a client about is the, the project sitting in a, uh, autonomous vehicle.

And I asked them what they trust and they said, Well, we trust the technology. . Good. What else do you trust? Well, we trust the algorithms. We said, Very good. Do you trust anything else? And they said, Well, no, there’s nothing else to trust. And I mean, the big miss for me was, do you trust the data and the governance around the data? The data management of that data? Because obviously all the algorithms and all the tech are being powered by the quality of the, of the data. I mean, and that was a non-regulatory need, non-financial, but certainly a safety one. I mean, have you come across similar? Similar use cases to that.

[00:05:00] Roshan Awatar: For me, I think the use cases that I’ve come across have always been attached to a specific need.

The business objective is to, let’s say for example, personalize customer interactions. You have got a specific agenda to make sure you can hyper personalize, make sure you understand every customer interaction, join their journeys up across several channels. You need data governance, data management, data architecture all over that to make it work.

So you have a reason. Whereas typically you have, well actually, you’re trying to preempt things going wrong by having a data governance layer. So those use cases such as personalization as an example, are perfect because as an immediate blocker that is stopping you doing things. And truth behold, you can get some of the way without data governance.

You can cobble things together, but it’s not necessarily reusable. It’s not scalable, and it ends up being costly in the long run. And that’s, I guess, that’s the battle because you can do stuff without it. But the consequences down the line aren’t very good.

[00:06:05] Barry Panayi: In Interesting there. Tony, your example, and Rosh using the word consequences.

I’m gonna bring up a subject which maybe I’m not qualified to speak about, but it seems like lots of people that aren’t qualified to speak about it are. So let’s give it a bash, I guess, and that’s data ethics. I don’t wanna lead the witness Rosh but this can we versus should we. Is data ethics a thing in your opinion, or is it data folk hyping stuff up that already exists in a lot of hot air?

[00:06:36] Roshan Awatar: I’ll spin this one on its head a little bit. So take away the word data. You’re left with ethics. So for me, the principle is the same. Are you taking ethical decisions? And actions, but in this case, specifically with the use of data. So typically it’s based, based on, I mean, the focus has been on automated decision making, but actually for me it doesn’t really matter if it’s automated or it’s by a human.

You have got an ethical, you know, responsibility to make sure you make an an ethical decision or take an ethical action. So I think absolutely it’s a thing. But it’s just, I guess another topic area you, you just need to govern. So I believe it’s a thing and it’s, And the word data is just providing us some focus because data’s getting hot everywhere and AI is becoming hot everywhere.

So it’s more about our ability to actually control things that are becoming more prominent. So machine, machine learning, a lot more volatile, a lot more un predict. And that’s something that needs a thing to govern it, a thing to control it or oversee it. And that’s where data ethics comes into play.

[00:07:43] Tony Cassin-Scott: But isn’t it the case that ethics, as you said, if you remove data, has always been there.

So this isn’t anything new. It’s, it’s always been whether it be automated or not. So do you think there’s a lot of hype around the use of data ethics or should this be like business’ usual activity?

[00:08:01] Roshan Awatar: I think it should become business as usual activity, but to actually mobilize, get momentum, get profile in an organization, you make it a thing.

Was there a chief data officer? But data’s always been a thing. Now there is, It’s one of the biggest roles you get out there. So for me it’s, it’s, it’s always there and many things have always been there, but it’s just emerging as more important and just needs that focus. So we’re just helping to focus the eye.

[00:08:25] Barry Panayi: How do you think you can cut through then, a lot of the, the chat, which we’re kind of agreeing is necessary to get it on the map to actually drive some action. I mean, having worked with you a few times, I know you are super delivery focused. If it ain’t delivery, you ain’t doing it. Very pragmatic. How does someone with such a strong delivery track record, like you actually build things like ethics in which can be a bit of a wormhole and people can pontificate and have meetings and do slides?

Have you ever seen it? in action work, Well, kind of ethics principles or something like that.

[00:09:03] Roshan Awatar: For me, it’s, it’s a little bit like data governance overall, right? It works really effectively if it’s targeted. So if you think about ethics, if you apply it in practice for something that you are doing, it is effective because people see the purpose.

If you try and I guess, overlay something, think, oh, you’re trying to preempt it, which you should do cuz you have got, You’ve have governance, preemptive governance, preemptive control. But at the moment, we need to prove it out in the first place, prove that it works. So actually choosing specific deliveries, specific things, and actually proving the model before you actually scale and industrialize. We could say, Let’s set up all these governance processes. All these governance forums. Another one. Get everyone to attend. And it’s like, Oh God, we’ve lost interest. But if you say, we are building an algorithm that is gonna touch every single customer in this organization then people will pipe up and say, Right there is a reason to make this decision.

We are trying to prevent something here and now. Once people get on the train, then it’s a lot easier to actually integrate processes.

[00:10:06] Barry Panayi: So you, you think the data ethics work can fit quite nicely into an existing. Data governance framework and doesn’t need to be separate. Is that right?

[00:10:15] Roshan Awatar: I do think so, but I’d probably say dependent on the maturity of the organization.

So if you take financial services, data governance is a thing, right? It’s been around for quite some time now and quite mature in terms of the chief data office agenda and the data governance agenda. Data governance already has that profile, already. Has that put foothole? So I’d say actually it could be a neat, you know, a, a neat entry point.

Somewhere else where data governance isn’t as strong. Data ethics probably needs its own thing or vehicle to actually raise the profile, get notice, get, get people’s attention. So I really think it depends.

[00:10:49] Tony Cassin-Scott: Isn’t it the case that ethics itself is a corporate wide responsibility and not just sitting with the chief data officer?

[00:10:57] Roshan Awatar: Indeed, indeed. I, I see the chief data officer as a facilitator. I think we should. As an organization, still have a foothold on what is happening on the organization, but it should just like data. I don’t think the chief data officer should be accountable for every single thing in the organization. It is a responsibility that everyone shares, right?

So we facilitate that and make sure we’ve got the, the foundations in place, the framework’s in place to make sure everyone can do their job safely and as effective as possible.

[00:11:27] Barry Panayi: Thank you. Oh, you’ve touched on how you implement stuff there, and I wanted to change tack a little bit and talk about strategy because we’ve had other guests on this podcast talking about, uh, data leadership and some myth busting around data already, and they spoke about the importance of data strategy and aligning it to the business strategy.

And, you know, strategy’s clearly important. I’d be interested not just from a governance point of view, but from someone who. Has embedded data strategies in a number of industries and also delivered the op model and stuff around it. How can you make a data strategy work and actionable and not PowerPoint?

I, I’ve seen you do it firsthand, but how do you approach developing a data strategy for an organization? .

[00:12:15] Roshan Awatar: Okay. There are some cliche responses that you know I might touch on, but I will try, try and, uh,

[00:12:20] Barry Panayi: If this was QI The thing would go off every time you said them. That we haven’t got the technology, the budget’s gone.

[00:12:26] Roshan Awatar: Right. So So, although data strategy is often a good thing, it has to be in response to, so, It’s got to be in response to a need a business team need, whether it’s near term or longer term, it has to be attached to an outcome. The business should feel it’s their strategy. Too often I see a data strategy developed by the data folk, but actually it should be the residual of the business strategy as do we wanna do this?

How are you gonna help us? How are you gonna enable us? Similarly with operating model, it’s exactly the same thing. Can’t develop in isolation, it interlocks with everything in the organization. So how do you build it? So for me it’s is upfront integration across the organization around what you wanna do.

And I probably say be a pragmatist. A data pragmatist, not a data purist, because there are compromises that are gonna have to be made because data is hard, right? As a data practitioner born and bred, I’m like, Oh God, I’d love to do this. But actually there are some things that you just need to prove incremental value along the way.

So for me, the strategy must be, along the way, showing the value, showing the benefit, showing an outcome, just to make sure people stay on the train. And then I’ll probably say importantly, money. Money makes the world go around if the data strategy isn’t backed by money. It is just a nice idea.

[00:13:57] Tony Cassin-Scott: I’ve come across instances where, um, unfortunately the, the organization has committed quite large sums of money into technology. That may, may not be appropriate for the data strategy to support the business strategy. So there’s lot of Brownfield sites there where they’ve bought some tools to do something which is inappropriate, but by that point of course they’ve invested the money. Have you come across those examples and how do you go about correcting that?

[00:14:22] Roshan Awatar: Indeed, it’s probably quite closely aligned to, uh, build it and they will come, sort of mindset, which is so, so common, Um, going into all these different organizations and play the hand, You’re dealt, you’ve invested some money. You can either say, Oh, this is wrong. Let’s you know, we a new one. Right? Tech is tech, right?

There’s, there’s some are better than others, but they’ll all probably do a job. So for me it’s around make the call. We are where we are and stitch up the pieces together. If we have to pivot some of the strategy around data, then do so. But for me it’s, there’s been too many iterations of a data strategy, a technology strategy, the age old, we’re building stuff the business don’t want.

Well, if I’m new into an organization and there’s already, there’s already something being spent, I says, Okay, let’s make the most of it, let’s move the dial. We don’t necessarily see, oh, what, what’s the life gonna like in five years? Which it absolutely should be a vision directing you, right? That, that you’re aspiring to, but it’s like well, how can we just show value no matter how small it is? It is a win.

[00:15:27] Tony Cassin-Scott: So taking what you just said in, in in mind there, do you think CDO is a data leader or a business leader?

[00:15:33] Roshan Awatar: I think it depends on the industry and the organization and the remit of the CDO. The CDO is evolved over time. Some CDOs are still governance, data management heavy.

Some are expanded into analytics, some own data platforms, some don’t. So it really depends where the main focus is. Personally, I think business leader, because you’ve got to be influencing the top table, right? You need to be in those conversations, understanding organizational context, understanding business context, and being close.

So it’s, it is a very strong partnership with the business. But being a, a data leader, I’d say, well, What springs to mind is that yeah, you might be a data geek, which is fine, right? I mean, I, I used to be a programmer, which is fine, but it’s about the blend of being able to, I guess, help lead a data organization, a group of data specialists.

But your second job is be business savvy. So I think it really depends. I don’t think there’s one or the other people would say, Well, absolutely a business leader, I think yes. But there’s still a benefit to be a data leader if you wanna inspire a bunch of data professionals as well. You know, it’s beneficial.

[00:16:41] Barry Panayi: Going back to that business leader, commercial acumen. Embedding a data strategy. Money makes a world go round comment, you made. Businesses demand ROIs on these strategies. What is your strategy to demonstrating that? Do you attach an ROI to this governance and ethics staff, or do you draw a line and say, Yes, there are these AI things that are gonna return X million, but as tickets to the game on data governance, don’t make me do an roi.

I’ve seen both approaches taken. What’s yours?

[00:17:13] Roshan Awatar: Here? If I have the choice, I say no. It is really hard, right. But when presented well, I have to, because that’s how investment cases work and you get money be targeted. There are specific use cases where, you know, it can be tangible and can it be associated right to a business outcome.

But where you fall down or it becomes difficult is when you try and industrialize it and say, Well, at scale for all your, you know, data work, you need to justify everyth. Because it’s so, data is so indirect, It’s so proliferated across an organization. So touches many processes, many systems, many people.

How do you get a real consolidated view on roi unless you are literally spinning up a program to actually identify your roi? It’s sort of a, that’s gonna take a lot of time. So, so for me, I’d like not to, but when you have to be targeted, it has to be targeted because you, you can do it, um, as an example.

Um, when I was in there in consulting, you could see a direct correlation between the data quality underpinning your RWA calculations. A tweak in your data quality was actually a tweak in the amount of capital buffer that you’d have to hold. So there was a direct correlation, but you can get left in, I guess, analysis paralysis.

[00:18:31] Tony Cassin-Scott: So isn’t it the case that it’s it’s a pay to play? It’s, it’s a, it’s an enabler for the ROI of the main business.

[00:18:40] Roshan Awatar: Absolutely. I mean, it’s not too different from, I guess, tech to an extent, right? Tech is an enabler, but unfortunately a lot of it is associated with operational costs as well in terms of, you know, switching old systems off, et cetera.

Um, and data sometimes gets looked in that way. So, um, it is difficult.

[00:18:58] Barry Panayi: Thank you so much. At the end of each of these podcasts, we ask everyone the same question. What’s the best bit of advice you didn’t take?

[00:19:09] Roshan Awatar: Well, that’s a good one. First time I’ve ever been asked that. I’d probably say an old boss of mine quite early on told me, Don’t try and don’t expect to be liked by everyone.

[00:19:23] Barry Panayi: I can vouch that he isn’t.

[00:19:26] Roshan Awatar: Yeah, I don’t think. Doesn’t like being liked and as you progress in your career become more senior, et cetera, it, you know, that’s even harder. It takes a lot of effort, but I try and detach being liked for with you can work with somebody. So over time, you know, I wish I, I, I adopted that earlier in my career cause I did spend a lot of time investing in relationships, which is great, but they’ll come a point in time, you can’t do that across everyone and you just have to, there is some level of acceptance there.

[00:19:55] Barry Panayi: Thanks very much for joining us.

[00:19:57] Roshan Awatar: Glad to be here and thank you very much.

[00:19:58] Tony Cassin-Scott: Yeah, thank you.