Enterprise Data Strategy: 22 Datasets in 8 Weeks

When your data strategy is stuck, stop trying to change the data sources — meet the data where it is. That single shift in approach took a team from one dataset in three years to twenty-two in eight weeks.

A large team had been working on a data platform for three years. In that time, they’d managed to land exactly one dataset. Meanwhile, the business was making insurance, liability, and capital allocation decisions across an $18 billion risk portfolio — essentially blind. No consolidated view. No way for executive leadership to see total cost of risk. Fragmented, ungoverned data scattered across multiple business units.

The classic enterprise trap: endless planning, vendor evaluations, architecture discussions, nothing shipping.

Why the Data Strategy Failed for Three Years

The previous team had spent three years trying to get data sources to change — pushing upstream systems to restructure, standardize, and deliver data in a perfect format. They were fighting the right fight in theory, but in practice, nothing was moving. Every data source had its own team, its own priorities, its own roadmap. Asking them to change was asking them to reprioritize their own work for someone else’s vision.

I came in and asked a different question: what if we meet the data where it is?

Instead of forcing sources to change, I leaned into the data expertise already on our team. We pulled data in whatever shape it existed and organized it ourselves into something we could derive real value from. I built actual ETL flows to apply the transformations — not just designing the architecture on a whiteboard, but building the pipelines that moved and shaped the data. It wasn’t the textbook approach. But it shipped.

Building a Risk Data Platform in 8 Weeks

I assembled a tiger team and focused on one thing: land real datasets fast. Twenty-two datasets in eight weeks. After three years of one.

Leadership was impressed and confused. The shift wasn’t technological — it was philosophical. I call it the “meet the data where it is” approach: stop waiting for perfect upstream data, start building with what you have. Pull data in whatever shape it exists, organize it yourself, and derive value now.

With the data flowing, I built the first-ever Total Cost of Risk dashboard in four more weeks. For the first time, executive leadership could see the complete risk landscape across the entire portfolio. Insurance negotiations, self-insurance decisions, capital allocation — all of it now informed by actual data instead of spreadsheets and institutional memory.

From there, I prioritized AI use cases: legal-summary generation, risk monitoring, insurance-negotiation modeling, capital allocation optimization. Built the model-validation and data-governance frameworks to make it all sustainable.

Results: $18B Portfolio With Executive Visibility

Twenty-two datasets in eight weeks versus one in three years. That contrast tells the whole story. An $18 billion portfolio with executive visibility for the first time. The solution became an archetype — in its first month, it generated $323K in additional revenue as a replicable pattern.

But the real outcome was simpler: a team that had been stuck for three years was suddenly shipping. The data engineers who’d spent years being told their platform was failing? They were the same people who landed those 22 datasets. They weren’t the problem — the approach was. Once we changed the question, they moved faster than anyone thought possible.