Scaling FinOps at Workday: Turning fragmented cloud costs into actionable insights

Workday is a leading cloud-based software enterprise that streamlines HR, finance, and IT management. With thousands of customers worldwide, Workday helps some of the world’s largest organizations plan, execute, and analyze their most critical business operations.

Like many enterprises, Workday relies on multiple cloud services—primarily Amazon (AWS) and Google (GCP). But doing cost optimization across such large, complex environments is no small task. Without clear visibility into cloud spend, it’s nearly impossible to identify inefficiencies or waste.

We’ll explore how Workday tackled this challenge by building a custom FinOps platform called Opus. We’ll also look at how dbt played a central role in helping Workday structure its data to generate trustworthy, actionable insights.

Fragmented and siloed cloud-spend data

For Workday, reconciling cloud-cost data from the two providers was incredibly difficult. Each provider reports cost data in a different way, with different schemas, naming conventions, and levels of granularity.

Workday’s finance and engineering teams had limited visibility into where money was going, and it was challenging to answer basic questions like:

• How much are teams spending?
• What are the costs associated with different services and customers?
• How much of that spend is efficient or wasteful?
• What are the cost forecasts for the next quarter?
• Where should the business reallocate budget?

On top of that, Workday's previous reporting stack was slow and required a lot of manual work to use. Key business logic was trapped in the reporting tool, making it hard to audit or reuse. Because the reporting tool was closed-source, customizations were limited, leading to bottlenecks and performance issues.

“We had little to no governance, which led to silos,” reflects Pattabhi Nanduri, Senior Data Engineer at Workday. “Without a framework for development testing or CI/CD, it was difficult to maintain code and deliver timely, data-driven insights to the business.”

A FinOps platform, powered by dbt

To address these challenges, Workday built Opus, an internal FinOps intelligence platform that sits on top of dbt, Lightdash, and a custom ETL stack running on Trino and Delta Lake.

Opus brings together cost and usage data from AWS and GCP, standardizes it, and models it into clean, usable formats. Today, it powers dashboards used across finance, engineering, and other teams. Its key capabilities include:

• Unified dashboards in Lightdash. Now, finance and engineering teams have a single place to monitor cloud costs, model future spend, and identify opportunities to cut waste.
• Deep visibility into compute usage, instance types, and spending trends. Teams can better understand which workloads drive costs and where to make optimizations.
• Role-based access control (RBAC) and SAR compliance. Access to sensitive financial data is restricted based on user roles, ensuring regulatory compliance.
• Cost categorization powered by dbt. Costs are organized into dynamic, multi-level hierarchies—like compute, storage, or credits—using reusable dbt macros. The structure can easily evolve as business needs change.

“We delivered a new cost-categorization hierarchy quickly and iteratively, thanks to dbt’s templating,” summarizes Nanduri. “Additionally, dbt’s Jinja macro framework made it easy to restructure our models without rewriting large amounts of code.”

Building structured, scalable data models with dbt

To build Opus, dbt gave Workday the technical foundation to support a high-impact, high-visibility FinOps platform.

Under the hood, here’s what that looks like:

• Macro-driven model design. Workday built an internal dbt macro library to template and enforce source models, staging models, intermediate models, and mart models. As a result, Workday can enforce modeling best practices across the organization.
• Streamlined hierarchies and flexibility. dbt's templating allowed dynamic modeling of an 8-level product dimension hierarchy. Categories like compute, storage, databases, and credits were easily shifted across levels through filter logic updates—saving significant refactoring time.
• Integrated testing. With custom macros and dbt-expectations, Workday automated join key validation, freshness checks, and custom filter and logic validation. This streamlined data QA without burdening users with complex testing syntax while driving trust in the data.
• YAML generation and documentation. Workday’s macros also auto-generated YAML files and documentation from model specs, eliminating manual sync work between SQL and metadata. This ensured documentation was always accurate and discoverable through dbt Catalog.
• Execution optimization with lineage graphs. Workday used dbt's lineage graphs and model profiling to build dynamic execution plans. These plans fed into Airflow and Cosmos to scale resources up or down, maximizing performance and minimizing cost.

“What’s great about dbt is how quickly new users can get up to speed with it,” says Eric Pu, Senior Software Engineer at Workday. “They can dive right in without spending months learning how to use it. Meanwhile, advanced users can go deeper and add customizations as it makes sense.”

Cleaner data and faster answers

Today, thousands of internal users leverage Workday’s cloud-cost data through dbt powered Opus. Now, finance and engineering teams have a single source of truth for cloud costs. Dashboards load quickly, users can find data for specific services or time periods, and spend is properly attributed across departments and environments.

Teams are onboarding faster, and they’re experiencing performance improvements across the board. It’s a truly self-serve model: users can access the data they need without waiting for a data analyst to pull a custom report.

Crucially, Workday’s FinOps layer is no longer a black box. All model logic lives in version-controlled dbt code. Every transformation is testable, reviewable, and documented via dbt Docs, making audits far more trustworthy.

“dbt is a powerful and flexible tool for enterprises,” concludes Pu. “It’s easy to plug in to your existing environment and customize to fit your exact needs.”

Your stakeholders need fast answers they can trust. If you’re thinking about how to build data infrastructure that delivers actionable insights, we’d love to chat. Get in touch with us to book a live demo, or sign up for free to connect to your data warehouse and start building.