AI Cost & Chargeback Reporting

Problem

As Morningstar's AI tool portfolio grew, so did the spend — and Finance had no reliable way to attribute costs to the business units actually generating them. Licenses, API consumption, and seat-based tools each had different billing models, different reporting cadences, and different owners. Without a chargeback framework, AI costs were being absorbed centrally with no accountability, no visibility into which teams were driving spend, and no basis for forecasting.

Approach

• Mapped every AI tool to its billing model — seat-based licenses, token-based API consumption, flat enterprise agreements — and designed a consistent allocation methodology for each.
• Built paginated Power BI reports tailored specifically for Finance: precise, printable, exportable, with clean audit trails rather than interactive visuals.
• Used Python to automate the data preparation steps — pulling cost data from vendor APIs and internal finance systems, normalizing currencies and billing periods, and producing the clean input files the report consumes.
• Worked directly with Finance stakeholders to validate allocation logic, ensure it matched existing chargeback policies, and stress-test edge cases (partial-period licenses, shared accounts, bulk seat changes).
• Built a forecasting layer using historical consumption trends to project forward AI spend by business unit.

Tech stack

Impact

• Allocated $200K+ in annual AI spend across business units, enabling cost recovery and departmental accountability for the first time.
• Gave Finance a defensible, auditable methodology for AI cost attribution — replacing ad-hoc spreadsheet estimates.
• Enabled multi-year forecasting conversations grounded in real consumption data.
• Reduced the manual effort of monthly cost reconciliation significantly by automating data preparation.

Screenshots