In a brief interaction with Express Computer at the GCC Summit 2024 in Bangalore, Nithya Subramanian, Head of Data & Analytics – AMEA, Kellanova (Kellogg’s), shares insights into the significant data challenges in the consumer goods industry. Subramanian discusses issues such as data availability, integration, governance, monetisation, and adoption. She highlights the complexity of managing data across federated IT ecosystems, the difficulties of integrating siloed systems, and the challenges of ensuring comprehensive data governance. She also discusses the impact of GenAI, the gender disparity in technology leadership, and the evolving support systems for women in the workplace.
What are the most significant data challenges you face in the consumer goods industry and how do you address them?
The first data problem is the availability of data from sources in the right shape and format to drive analytics. It all starts with the complexity of the process itself and how fragmented the business is. In most CPG organisations, you see they are highly federated, even their IT ecosystems. This means dealing with highly complex processes and IT landscapes. One market may use a different set of tools and process nuances compared to another market. The complexity lies in identifying the right sources and bringing them together in a unified data platform to drive analytics, GenAI, and more. That’s the first challenge.
The second challenge is integration. Once you have all this data, integrating it becomes a huge challenge. Siloed systems follow siloed nomenclatures, and none of them naturally talk to each other. You need to build additional complexity to make them talk to each other, link the data sources, and get them on a unified model.
The third challenge is governance. On a unified data ecosystem, governance includes multiple aspects like data privacy, data security, data quality, data lineage, cataloguing, and data retention. Rather than handling these aspects in a piecemeal manner, they are brought under one governance umbrella. Even though different roles handle different aspects, it is all governed. However, this is still a huge challenge. It’s like moving an elephant because you’re dealing with so much data, including personal and proprietary data, which raises data security and privacy concerns. A very strong framework is required here, and it is quite challenging—easier said than done.
The fourth challenge is monetisation. After spending a lot on data acquisition, integration, and storage, how do you realise the benefit of all this? How do you monetise the data? You have all these tools at your disposal, from fundamental machine learning and business intelligence to GenAI, but how do you convert this data into money? What are the right use cases? What do you prioritise? This challenge involves dealing with multiple stakeholders who want their use cases taken care of. It becomes crucial to justify the return on investment, prioritise, and manage stakeholders accordingly.
Finally, once you have built everything, how do you ensure that it is adopted? This involves change management. Many people are still using old-school Excel processes. You have fancy solutions that you believe will work, but they need to be adopted. Change management is the last mile, but it is the most important because, without it, your entire effort could be lost.
There are two classes of people regarding AI – those who see it as a Terminator and those who see it as Tony Stark from Iron Man. You have to manage both sides. The Tony Stark guys want everything, regardless of whether it generates money. They are super excited about AI and GenAI. The Terminator guys are scared of everything and are happy with their Excel sheets. Managing this change is crucial because it ensures that your plans for monetising data are effectively implemented.
Do you think this lack of adoption is more of a budgetary concern or is it the mindset?
It’s more of a mindset than a budgetary concern. One leads into the other. If something has to start, it’s the mindset that leads to the superficial notion of a budgetary constraint. Every organisation has a baseline for a certain percentage of IT spend. Even within that IT spend, there is a certain apportionment for data and analytics, right?
So it’s not much of a concern about the capital investment. It’s about the mindset change regarding whether I should be putting this capital investment into GenAI to do X, Y, Z. Then you have to deal with the mindset problem, saying that if you give me 10 bucks, I can deliver back 20 bucks to you. You try to address the mindset issue and get the budget right.
The budget is always a baseline at the beginning of the year for every organisation. People don’t go around with money in their pockets. It’s always a certain percentage of your revenue or profit pegged right at the beginning of the year. Then it becomes a problem of mindset about how that can be invested in the right, generally, use cases.
Since GenAI has emerged in the past 2-3 years, has it made your work easier or more challenging due to perceived security concerns?
Honestly, a lot of organisations have not gone beyond the POC stage in GenAI for business use cases. However, in terms of productivity improvements, GenAII has brought significant savings and productivity gains, especially for developers. We can roughly translate these improvements into dollar savings, but many organisations, particularly in CPG, are still in the POC stage. Have they been able to realise the full benefits? I would challenge that, as only a small percentage have. Despite this, the outlook is quite positive.
Regarding data privacy, I believe the slow pace of productionalisation of these models is beneficial. It allows the industry time to figure out how to address these challenges. We have some time, though not a lot, to deal with issues like privacy and quality. As POCs are ongoing, organisations are setting up frameworks and committees focused on compliance with data privacy and security regulations. These committees include experts who understand global regulatory frameworks and legalities, ensuring that when these models are released into production, they are efficient and compliant.
AI has significantly impacted data analytics. Initially, we began with simple business intelligence, moving to linear equations, regression analysis, and classification problems. We then transitioned to machine learning and now artificial intelligence, including neural networks. AI has simplified life for consumers and fostered high innovation within companies, allowing them to innovate according to consumer preferences. Innovation cycles have reduced, leading to product releases every six months instead of years. This innovation is highly consumer-centric, with hyper-personalisation possible due to AI’s ability to crunch and synthesise data, find patterns, and take automatic actions. AI has been instrumental in productivity savings and improving a company’s market standing against competitors.
Now, every company is leveraging AI, but the competitive edge lies in who uses it better and faster.
In all my interactions so far, I’ve noticed a ratio of about 1:10 between women and men leaders. Why is this discrepancy there?
This discrepancy is definitely worrisome and something we need to address together. I completely agree with your data point. If you look at cities like Bangalore, Mumbai, or Delhi, there is an equal proportion of male and female students in classrooms. Companies in these cities also maintain diversity. However, in Tier 2 cities, college-level diversity is a problem. Even then, the proportion isn’t as bad; it might be around 40% to 60% female representation. In Tier 2 cities, women often don’t attend college or choose certain fields.
As women progress in their careers, they face various personal milestones, such as marriage, childbirth, and elderly care. Due to societal pressure, it’s usually the woman who has to sacrifice her career, not the male partner. Thus, women naturally fall victim to these lifestyle milestones. So far, companies haven’t done much to address this issue.
Today, companies are realising the benefits of diversity and are putting practices in place to support women during these critical milestones. For example, they are providing extended maternity leave, work-from-home options, and even sponsoring nannies for up to three years. There are also benefits related to elderly care, with work-from-home being a common solution. Companies are also starting to sensitise men to support diversity.
We’re on a positive trend now, though the changes won’t happen overnight. Considering an average career spans 20 to 30 years, it will take a couple of decades for the current changes to significantly impact the final diversity equation. However, the situation is already improving.