How to Make Shared Data Costs Transparent at Scale ft. Oliver Milke, TD | Ep #69
How do you make cloud data costs transparent and actionable when thousands of users share the same platform?
In this episode of FinOps in Action, I sit down with Oliver Milke from TD, who shares how he built a Data FinOps practice from the ground up inside a large financial institution, starting as a team of one and scaling into a dedicated function.
Coming from a background in economics and finance, Oliver offers a unique perspective on cloud cost optimization. Rather than focusing solely on infrastructure, his team goes one level deeper: tracking and allocating costs at the data layer, down to individual tables, workloads, and user groups.
The conversation explores why traditional FinOps models fall short for shared data platforms, how transparency drives accountability, and what it takes to operationalize cost awareness across thousands of users. Oliver also breaks down the role of education, chargeback, and engineering collaboration in building a sustainable FinOps practice.
Here’s what we talked about:
What “Data FinOps” is and how it differs from traditional FinOps
Why shared data platforms require a fundamentally different cost model
Building cost visibility down to the table and workload level
Moving from showback to chargeback and how it shifts user behavior
Platform vs. workload-level optimization strategies
The role of education and change management in FinOps success
“We recognized that this cannot be a one-off, but it needs to be a permanent ongoing thing.” - Oliver Milke
Connect with Oliver:
In this episode of FinOps in Action, Oliver Milke draws on a background in economics and project management not engineering to make the case that FinOps for a shared data lake is an entirely different discipline than traditional cloud FinOps. He breaks down why a driver based chargeback model changed the way thousands of TD users think about cost, how combining billing data with telemetry is the key to unlocking true workload level insights, and why education isn't a nice to have for a mature FinOps practice it's a core pillar that needs its own dedicated headcount.