Shivanshu Mathur
AI & UX Strategist
Blending AI thinking, product strategy, and user-centered design to craft experiences that actually move the needle for businesses.
India
AI & UX Strategist
Blending AI thinking, product strategy, and user-centered design to craft experiences that actually move the needle for businesses.
Projects that showcase the range and depth of my craft — from enterprise strategy to platform development.
Led the end-to-end transition from fragmented Excel-based financial planning to a scalable, trusted enterprise platform.
Excel was the primary tool for financial planning at Novartis, but at global scale it became slow, inconsistent, and difficult to govern.
Finance teams relied on fragmented models for cost management, planning, and forecasting, creating operational risk and heavy dependency on manual reconciliation.
I led the transition from Excel-based planning to a scalable platform, structuring fragmented workflows into a system designed to support global teams and build trust at scale.
This wasn't a cosmetic issue; it was a systemic breakdown. Planning, pricing, and forecasting lived in silos owned by different people with zero version control or audit trails.
Every cycle, teams lost days just trying to get global and local numbers to agree. Because there was no way for users to help themselves, our subject-matter experts became human bottlenecks.
Knowledge was trapped in people's heads rather than integrated into the system, making high-stakes scenario planning nearly impossible.
Discovery revealed that teams didn't use Excel because it was efficient—they used it because it was flexible and trusted. To drive adoption, we couldn't just "digitize" spreadsheets; we had to replicate that flexibility while solving for what Excel lacks: Global Governance, Scalability, and Predictive Intelligence.
Next Steps: I architected a three-layered roadmap to transition users responsibly:
Design plan shared with PO for next steps
Instead of designing a single solution, I led the work as a platform initiative, sequencing capabilities to reduce risk and drive adoption incrementally.
We mapped legacy Excel workflows end-to-end to understand where trust broke, where manual effort accumulated, and where decisions actually happened. This led to a layered platform strategy — rather than a monolithic product.
I led this initiative end-to-end — across discovery, strategy, and experience definition. This was leadership through structure, decisions, and alignment — not just execution.
Standardized cost data to build trust and governance. Centralized corporate and central costs with consistent structures and reliable reporting.
Forward-looking planning and forecasting capabilities. Demand and pricing scenarios with revenue impact modeling and strategic planning.
Intelligence and optimization for cost management. Granular cost views with optimization opportunities and predictive signals.
RAG-based chatbot reducing support dependency. Answers repetitive finance and tool-related queries with contextual guidance using documentation, dashboards, and definitions.
After launching Phase 1 of the platform, we conducted a survey to understand how users were adopting the Market Model application. The survey was distributed to 50 users via Microsoft Forms.
The survey responses made it clear we needed to rethink our approach. Then, in Phase 2, Novartis shared new brand guidelines. This gave us the opportunity to revamp the project—addressing what users actually needed while aligning with the updated design system.
The platform eliminated days of manual reconciliation work. Teams shifted from getting numbers to agree to using those numbers for analysis and strategic decision-making. Subject-matter experts are no longer bottlenecks, and knowledge is now embedded in the system rather than trapped in people's heads.
Synthesized capability index for illustration — not literal measured units.
Feedback from key stakeholders who worked closely on the project.
Replacing Excel isn't a UX problem: it's a trust, ownership and system-design problem.
Going in, I assumed the fix was obvious: give people a better interface than Excel and they'd switch. What I didn't get until we were deep into it was that nobody was using Excel because it was good, they were using it because it was theirs. They trusted it, could bend it however they needed and no tool we built was going to win by being stricter than that.
So the real work wasn't removing that flexibility, it was earning the right to replace it building the governance people could actually rely on, the intelligence that made the tool worth the switch and enough self-service that people stopped needing someone else's permission to get an answer. Skip any one of those and you just end up with a nicer-looking spreadsheet that nobody adopts.
That's the thing I carry into every platform or AI project now, the interface is rarely where the real problem lives.
Open to product design and platform strategy roles, or a conversation about your own data platform problem.
