Mars operates in more than 80 countries and is home to some of the world’s most recognized brands across pet care, food, and confectionery. With over 150,000 associates globally, the company has made significant investments in people analytics to enable smarter and faster workforce decisions. At the center of this effort are Ujjwal Sehgal, Global Head of People Analytics, and Rachel Belino, HR Data Officer, leading the People & Organization (P&O) Data Office. Their teams focus on solving real business problems with data while strengthening the foundations that make those insights scalable.

In this second installment of a three-part series, Sehgal and Belino join Shachin Prabhat, Vice President and Business Unit Head at Tiger Analytics, to outline the practical playbook Mars is using to close the gap between data and confident decision-making. While Part 1 examined the realities of the data-to-insight divide, Part 2 explores the solution levers Mars is putting in place to scale reusable data, balance short-term value with long-term foundations, and prepare for agentic AI.

Why a common data model becomes critical over time

Belino begins by explaining that early on, the team focused on delivering targeted data solutions that addressed immediate business problems.

As the organization matured, the need for reuse and cross-domain insights became unavoidable. What began as product-specific data gradually evolved into shared, reusable data used across multiple solutions.

That shift exposed new limitations. The team needed a way to analyze data across domains and across time.

“That’s when a common data model became a critical part of what we needed. It can unlock data discovery for your data scientists and enable use cases that require you to cut across data domains to bring together a more holistic package of insights,” Belino adds.

Delivering value now while building the foundation

A recurring challenge for enterprise data teams is balancing immediate business demands with long-term platform investments. Sehgal shares how Mars addresses this tension by keeping the end user at the center of every initiative.

“We’ve approached this from an empathy and understanding standpoint. What is the goal and the objective they are trying to achieve?”

This user-first mindset directly shapes product design and adoption. “We want to design solutions that are easy for the user to navigate, get the insights faster, and they don’t have to figure out what dropdown or filter to select,” explains Sehgal.

He further mentions that one of the most important lessons learned was to reverse the traditional development sequence. “We follow this principle that we’ve tried to embed into our operating model, which is to build the front end while the backend starts to catch up.”

This approach ensures that solutions are shaped by real user needs before heavy investments are made in infrastructure.

A key observation was that when initiatives began with specific requests for data for analysis, the resulting products often saw low adoption rates because users struggled to navigate them.

By designing for the user first, the team accelerated adoption and created a clearer direction for the data platform roadmap.

Why letting the backend catch up actually helps

Belino explains that allowing the backend to follow successful front-end solutions creates both financial and strategic advantages. “Building a foundation requires significant investment for your organization, both in terms of money and time. It’s important to demonstrate value tied to that investment.”

When front-end products prove value early, they make it easier to justify foundational investments. “If you are focusing on the foundation and the pilots or solutions that have proven successful, it’s easy to justify investment in the foundation.”

This approach also creates space for thoughtful planning and scalability.

“It gives me time to be thoughtful and mindful and plan and come up with the roadmap that builds the foundation to be scalable in support of what is being driven by our stakeholders,” Belino says.

Rather than building infrastructure in isolation, Mars builds it in response to proven business demand.

Opening the black box of agentic AI

As the conversation turns to emerging technologies, Belino emphasizes the importance of demystifying agentic AI and avoiding hype. “My favorite buzzword is the agentic AI framework. I hate black boxes and hype words, and I love digging into it and figuring out what’s in that black box.”

Her team focused on breaking the concept into tangible components and building internal capabilities around them.

“We owe ourselves as data professionals to get behind that and understand it.”

Belino highlights three foundational elements that proved essential:

1. The context layer: “The context layer became extremely important when we were enabling AI for BI. It’s important to have a detailed technical blueprint so you don’t have a black box.”
2. Prompt engineering visibility: “It’s important to know what’s going on in that prompt. Do you have the right technical and functional skills to understand what’s going into the prompt and validate what is returned?”
3. Persona-driven orchestration: “To get suitable responses, you need to first start off with the persona. Who is the persona that is asking the question?”

Together, these components enable scalable, repeatable, and dependable AI capabilities.

Moving beyond silos

Sehgal expands on how agent frameworks can help break functional silos and surface insights users may not think to ask for. “What the end user is looking for is the AI elevator. Many end users don’t understand the fundamentals of getting the right data and having the right workflow.”

AI’s value, he explains, lies in uncovering connections across the broader ecosystem.

“If I’m responsible for talent acquisition, I’m only going to think about talent acquisition. But the beauty of bringing in the common data model and the agent framework is now saying, let’s not just look at talent acquisition,” Sehgal says.

He provides a practical example: “Turnover has direct implications on talent acquisition. If I already know what my turnover’s going to look like, I can start prepping my talent acquisition team.”

This shift enables proactive, cross-domain decision-making.

“Don’t look at it purely based on the blinders of what’s relevant for you. There could be other aspects influencing what matters for your role. That’s where the power and the beauty of AI come to life,” Sehgal concludes.

Up next in Part 3: Sehgal and Belino share how Mars is tackling the hardest challenge of all, driving adoption at scale by building trust in data, protecting sensitive workforce information in the age of GenAI, and offering practical guidance for organizations just beginning their journey from data to action.

CDO Magazine appreciates Ujjwal Sehgal and Rachel Belino for sharing their insights with our global community.