Empowering thousands with data: Haleon’s data literacy journey

Tarush Aggarwal (00:00)

Welcome to another episode of People of Data, where we get to highlight the wonderful people of data. These are our data leaders from various industries, geographies, and disciplines, discover how companies are using data, their achievements, and the challenges they face. Today on the show, we are thrilled to have Kshitij Kumar, the former chief data officer of Haleon, the leading everyday consumer health company, where he oversaw all aspects of data analytics and AI.

‍

His expertise spans a wide areas of industries, including CPG, retail, fashion, luxury, sports, media, and communications. With previous stints at Europe's largest brands like Farfetch, OneFootball, Zalando, KK has also founded and built several startups and helped raise over 85 million in funding. Welcome to the show, Kshitij

‍

Kshitij (KK) Kumar (00:51)

Great to be here. Thank you for having me, Tarush.

‍

Tarush Aggarwal (00:53)

Of course, are you ready to get into it?

‍

Kshitij (KK) Kumar (00:55)

Absolutely, let's do it.

‍

Tarush Aggarwal (00:56)

Awesome. So very quickly, does Haleon do? does the company make money and what role does data play in helping Haleon make money?

‍

Kshitij (KK) Kumar (01:05)

Absolutely, Haleon is a FTSE 25 company, which is one of the largest companies in the UK. You may know them by brands such as Sensodyne, Advil if you're in the US, Centrum, Enofruit salts or Iodex if you're in India, and many other brands worldwide. They used to be part of GSK GlaxoSmithKline, it was the consumer health part of it.

‍

and the split happened a couple of years ago. I used to be the chief data officer for Haleon till a couple of weeks ago, and so happy to talk about any public information on those topics.

‍

Tarush Aggarwal (01:44)

Awesome. How does data play a role? You have a multi -brand strategy. What are the interesting things data is playing in helping Haleon with all of these brands?

‍

Kshitij (KK) Kumar (01:57)

Absolutely. So if you think about the brands that Haleon has, think about a Sensodyne. Think about how we understand what a consumer actually needs. How do consumers actually use these? How do third parties that actually sell these products understand what's working, what's not? So I'll give you an example. Companies like Haleon are companies that sell

‍

through third parties in your country, wherever you are, you'll go into a pharmacy and you'll see prominently displayed Sensodyne packages. But there may be Sensodyne that's a fast relief Sensodyne. There may be another Sensodyne that's a slightly different angle to sensitive teeth or to toothpastes. And so the question becomes, how do we understand what that consumer's behavior is?

‍

versus those brands? What's working? What's not working? A lot of people purchase or get recommendations on products like Haleon's products through dentists as an example. How do you understand what that dentist finds works and what doesn't work? And then how do you take all of that information back and put it into the research and development, put it into the marketing? How do we make sure that there is impact for all of these things?

‍

for the consumer, but at the same time, how do you balance it with the risks of using machine learning, of using data, of using AI? So data plays an important part across the board. The consumer part is one, the risk part is another, but think about delivering a certain package to a certain place. When I was in South Africa a couple of years ago, my son got sick.

‍

And when we went into this pharmacy in the middle of a jungle, literally in the Kruger National Park, the pharmacist gave me a sachet of Eno fruit salts, which is a Haleon product. And it was a proud moment for me to find that product there. But when I asked the person for a few more packets, the person wanted to hold them back. He said, I don't know when I'm going to get the next ones. I want to make sure that they are available for the next customer.

‍

And so that demonstrates a challenge for companies like Haleon, which is you're delivering products through third parties. How do you make sure that the products are there at the time when they are needed, in the quantities that they are needed? And it's a great positive experience to have. I'm glad to see that companies like Haleon are able to do it. This is all based on the data.

‍

Tarush Aggarwal (04:25)

Yeah, very exciting. Very to see data is playing such a big role. What culture, you know, what's the culture like a massive business like Haleon

‍

Kshitij (KK) Kumar (04:36)

It's amazingly positive. I used to be part of a team that was a digital and technology team. And to your point, it's a company that started out as a pharma company. it wasn't initially, everything wasn't data -driven, obviously. It's a pretty old company if you think about it. And so the pedigree may not have been data.

‍

But the teams understand data. They understand how important it is. And the C -level is down to the engineers working on things or even the R &D folks working on these things. Data has become a big part of what they are doing. Is it where it needs to be? think given all the things that are happening with machine learning and AI and analytics, et cetera, we will see data being used even more.

‍

But think today it is fundamentally part of the core DNA for the company. I think part of my role being there as the chief data officer was to make sure that that DNA percolated through the company.

‍

Tarush Aggarwal (05:34)

what was the culture of the data team? Like, yeah, cause you know, very typically you have data groups which are like leading the change in many ways, especially as you mentioned with AI being on the forefront. how did, know, what was the, what was the, what was the culture like in the data team and how did it play out into the bigger org?

‍

Kshitij (KK) Kumar (05:52)

Yeah, it's a good question because when a company starts the journey into data and AI, you typically start with innovation. And so when you're innovating, you're not really looking at the impact. You're trying to find use cases that'll actually move things forward. But then as you grow, you start looking at what is the impact versus the investment that we made.

‍

How productized are you? And you start thinking about things like governance, thinking about things like risk tolerance, things about responsible AI, et cetera, et cetera. So, you we had the same things at Haleon. The culture started from an innovation culture. But while I was there, we were maturing into being a forward thinking, innovative, but also

‍

well -governed organization with responsible AI guidelines, et cetera, et cetera. So it's following the normal path from innovation to becoming a more productized data organization, I would say.

‍

Tarush Aggarwal (06:52)

Yeah. Talking about normal path, what does the data platform look like at Haleon?

‍

Kshitij (KK) Kumar (06:59)

Quite broad, think again, I won't be giving away any secrets if I talk about things that are publicly well known. It is primarily an Azure shop, primarily a Microsoft shop, but there are other clouds being used as and when needed as well. Without going into the details of those clouds, there are some major platforms that are used today like Databricks. It's been a big Power BI.

‍

effort as well. So lots of analysts, people use Power BI, et cetera. I think overall, if you think about companies in this space, there's often an operational data store and there is a platform for analytics. you know, SAP happened to be the operational part of it. And then data was extracted into a data store in Azure. So again, nothing very surprising, but

‍

very, very well put together large data team that was focused on making sure there is impact and there is impact safely.

‍

Tarush Aggarwal (07:52)

Yeah. And how big is the organization from a people perspective?

‍

Kshitij (KK) Kumar (07:57)

So without going into the details exactly of how many people there were, suffice it to say it's a very broad organization. It's located worldwide with offices in Bangalore, in India, in the US, in the UK. Our headquarters are in the UK, and so a significant part of our team is in the UK. Several hundred employees as well as third parties providing resources as and when you need it. So very progressive organization.

‍

broad worldwide and working across many different time zones.

‍

Tarush Aggarwal (08:26)

Yeah. I like the mix of folks, internal and external. I always feel that creates a really healthy culture. If you have one piece of advice for someone who is looking to go join the data group at Haleon, what would that be?

‍

Kshitij (KK) Kumar (08:40)

I would say to know what your special intellectual property is. Why you? Why yourself? If you asked me that question, why was KK the chief data officer at Haleon? I would answer that the same way. I would say I bring a lot of depth, technical depth as well as cultural depth, working on all kinds of data and artificial intelligence topics across the board.

‍

So the one piece of advice I would bring is understand your value proposition. It doesn't matter if it's in the data science side or the analytics side, the AI side, governance, risk, responsible AI. Pick the area that you want to focus on unless you're the CDO, then you have to pick all of them.

‍

Tarush Aggarwal (09:21)

Yeah. And do you see like a general shift in roles? know, if on one side of the spectrum, have broad skillsets, data engineering, largely been SQL and Python based versus, you know, very, very specific skill sets. As we think about AI and training of like foundation models and stuff like that. can go very specific, you know, if we have to, you know, in the last few years, we've been fairly broad.

‍

with analytics engineering was very largely SQL -based technologies. Do you see the shift continuing to, do you see this as sort of, these sort of broad skill sets continue to be very relevant or do you think that teams, especially in the Haleon context, are starting to get more specialized?

‍

Kshitij (KK) Kumar (10:05)

Yes and yes. So what I mean by that is this space is changing very, very fast. If you think about it, two years ago from today, November, 2022 is when chat GPT was launched, right? Nobody would ever, sorry.

‍

Tarush Aggarwal (10:20)

Yeah, I know

‍

the exact date of that. was December 14, December 14, 2022. And the reason I know this is that's the day 5X launched. So we both launched our product on the same time.

‍

Kshitij (KK) Kumar (10:32)

Awesome, amazing, you picked a good day to launch. And the reason for that is, think about it, that's not even two years ago from today, right? And how much of a generational shift has already happened with generative AI? Now generative AI causes some really interesting questions, not just for Haleon, but for any company in this space. Generative AI, or any AI in general,

‍

Tarush Aggarwal (10:40)

Yeah.

‍

Kshitij (KK) Kumar (10:56)

depends on the data underneath. So the question becomes the data engineering. How are you getting that data? How are you cleaning that data? How are you governing that data? How are you going to make sure that you are going to do responsible AI, et cetera, at that time while you're making impact with it? Even the way you do analytics, there are startups that are doing, making huge changes with analytics, for instance, moving things forward, making it easier. So

‍

the question is where will the skill sets actually evolve to? And unfortunately, I don't have the answer. What I do know is you're going to need both of those. You're going to need those skill sets where you can pull in that data, put it into these platforms, use these platforms. But on the other side, you're going to need people who really know how to ask those business questions in a way

‍

that these platforms, whether generative or not, are able to get you those answers. Analytics, the way we do analytics today, is not going to remain the same. Analytics is going to be very heavily artificial intelligence powered. So the answer is you have to do both. There will be people who will need to know those skill sets of data engineering, et cetera. And there are all new things. Like prompt engineering is going to become even more sophisticated. Right now,

‍

If you want answers from a GPT, from an LLM, you have to talk to it like a human being, like you're talking to a data scientist, but you have to give it all of the information that you want it to have. There must be better ways of doing it than how we do it today. What those better ways are, I don't know yet, but that's the direction we're going in.

‍

Tarush Aggarwal (12:35)

Very exciting. is one achievement during your time at Hylion you were particularly proud of?

‍

Kshitij (KK) Kumar (12:42)

I'll actually tell you one problem first and then one achievement that we had which solved that problem. The problem we have in any company, not just Haleon, but a company of the size of Haleon. Haleon is about, I think, between 23 ,000 and 24 ,000 people worldwide. And a FTSE 25 company, which is one of the top.

‍

Tarush Aggarwal (12:50)

Next slide. Yeah.

‍

Kshitij (KK) Kumar (13:08)

25 companies in the UK size -wise.

‍

And a company of that size, when you are trying to deploy data and artificial intelligence, the question becomes, how are you going to get people to even understand what it means? How are they going to work with the data? How are they going to work with AI? And it's not a Haleon specific question. This is exactly the same problem that I saw at Farfetch as an example, right? Very fast moving company.

‍

very well -known, largest online luxury retail platform. Same problem, which is how do people understand what data there is and how can I even use that data? So this was an issue for us at Haleon as well. The way we addressed this issue was we figured people need to start talking about this. They need to start using data. They need to start using AI.

‍

Data and AI are a culture problem. The technology part, we'll figure it out. We know how to figure it out. The data culture part is really, really important because there can only be a few hundred people in the data teams. There'll be a few thousand or tens of thousands of people in the teams that are using that data and are using that AI, are using that platform. So the way we sorted it,

‍

is we actually did what we call an AI Accelerate Day. If you look online on LinkedIn and stuff, people have posted photos and details about what we did. But what we did is we did one day where we brought everybody together around the world across, I think it was like 12 time zones or something, where we brought people together to talk about data and AI. We called it an AI Accelerate Day. The plan was to do it every year.

‍

We were able to talk starting from the very basics of how is AI, how is data impacting the jobs of people around the world? How can they do their jobs better? Where do they even go to get started? Where is the data? And then we talked a lot about our responsible AI approach. What does it mean to use AI responsibly? What does it mean to be using, you know, everybody wants to use ChatGPT.

‍

How do you use ChatGpt in order to do your work better while keeping the company safe and while keeping yourself safe? And we used this eight hour time period to spread across different time zones. It was afternoon in India when it was morning in London and then the US joined at some point during the middle of the day. But during those eight hours, we started this conversation.

‍

where people got excited and people started understanding, how do I get started with data and AI? What does it really mean for me? We didn't solve that problem completely, but we scratched the surface and we connected people. So people knew who to talk to after that. So I felt like at Haleon, that was a very positive step. And I hope that culture and the cultural change that we started continues while I'm no longer there.

‍

Tarush Aggarwal (16:09)

Yeah, I love that example of using a grassroots movement to really go impact systematic change, to build a framework which can have systematic change. Do you have any, if you had to go measure the impact of that project, and I know this is particularly tricky, how you think about measuring that? what do you think would be some of the

‍

second and third level consequences of projects like this. I'm sure a lot of CDOs out there, this is not a technology problem, as you mentioned, this is a culture problem and an education problem. how do you, and I can tell that you have given this some thought, so how do you really measure impact of this? How do you know if you're on the right path and what to do and sort of what to do?

‍

Kshitij (KK) Kumar (16:54)

Yeah, well, since I'm no longer at Haleon, I can't address this directly about how we can measure it at Haleon. I suspect it'll be measured over the next few years. What I can tell you is we did exactly the same thing at Farfetch when I was there at Farfetch, slightly smaller company, another NYSE traded company. I was on the executive board of the company. And when we did this cultural change approach, and you're right, I have given it a lot of thought, I've focused on

‍

the cultural change that drives data and AI and innovation across all of my roles. And at Farfetch, what I can share is we were able to measure via surveys, just looking at just the data and AI team. We would measure how the data and AI team felt, the temperature on how they felt about their jobs, how they felt about working at the company. And we measured every year over several years.

‍

And the difference between the numbers when I joined versus the numbers when I left directly correlated to the efforts that I had instituted to create a data culture. And what we noticed is there was a 120 % improvement in the sentiment inside the data team. And what does that mean? The data teams...

‍

feeling 120 % better about themselves means that they are having more fun at their job, that their job is becoming more impactful, that their stakeholders are seeing a response and their stakeholders are partnering with them so everybody sees the impact of that data. They're feeling safer because the data governance is there. So what I would recommend to anybody who's doing these kinds of efforts is to measure them as you start. If you haven't measured it yet,

‍

measure it today, start understanding how our stakeholders are using this, how our teams themselves are using it, and then use it over a period of time so that you can actually see how you're doing. Measure what matters, right? And then you'll be able to see how things are improving.

‍

Tarush Aggarwal (19:00)

measure what matters on that note. KK, thank you so much for your time and thank you for your wisdom today.

‍

Kshitij (KK) Kumar (19:05)

Absolutely, it's a pleasure, Tarush and I look forward to hearing more interesting episodes on your podcast.

‍

Tarush Aggarwal (19:12)

Thank you so much.

‍

Kshitij (KK) Kumar (19:13)

Cheers.

‍