CDO Magazine Podcast Series · CDO Magazine

PODCAST | The Data Model Behind Mars’ Analytics Evolution

March 12, 2026·21 min·2 clips

Mars executives reveal why their initial conceptual data model failed and how a common data model became a game-changer.

The episode opens with host Sachin Prabhak of Tiger Analytics framing the problem: organizations have accumulated data but leaders cannot act with confidence due to a gap between data and insight. Rachel Bellino, HR Data Officer at Mars, begins by describing her early approach of building conceptual and logical data models, which she found less useful than expected when the team needed to deliver product-specific solutions. Over time, as reusable shared data assets were built, the common data model became valuable for enabling cross-domain analytics and data discovery. The team then faced a new challenge: historical traversal and unified data on their platform. Ujwal Sehgal explains the user-centric design principle Mars adopted, where end-user goals and decision needs are identified before data requirements are determined. He describes a lesson learned from low product adoption caused by building data-first solutions without user-centric design. The principle of 'building the front end while the back end catches up' is introduced as Mars's operating model to balance near-term business value delivery with longer-term foundational investment. Rachel adds that this approach makes it easier to justify infrastructure investment by demonstrating tied value through front-end products. She also notes that deferred back-end work allows the team to plan a scalable, business-driven roadmap based on observed product priorities rather than speculation. The conversation shifts to agentic AI, with Rachel expressing enthusiasm for understanding what is inside the black box. She identifies three components: the context layer, prompt engineering with full visibility, and persona-based orchestration via an AI framework with an orchestrator and specialized agents. Sehgal builds on this by arguing that the real power of AI is surfacing cross-domain leading indicators that users within a persona silo would not know to request. He uses the example of a talent acquisition manager who would benefit from turnover forecasts but would never request them unprompted. Sachin links this back to the 'adjacencies' concept — looking at insights at the intersections of personas and their related domains. Both guests agree that the shift requires making users somewhat uncomfortable, as it challenges them to think beyond their functional silos. The episode closes with a brief product mention by Sachin for Tiger Analytics' Zero Shot Profiler, which rapidly generates persona profiles for AI-ready analytics platforms.

As heard by us

A grounded enterprise case for getting the data foundation in place before chasing the next layer.

This episode stays on the tension between enterprise data and confident action, with Mars as the concrete backdrop. It keeps circling the same point: the foundation has to be in place, the common data model matters, and the back end sometimes has to catch up before the front end…

Read the full review in PlayNext →

Why you'd press play

Want a clear enterprise case for fixing the data foundation before chasing AI buzz?

Read the full recommendation in PlayNext →

Listen to the show on

Apple Podcasts Spotify

Play clips from this episode More from CDO Magazine Podcast Series