Production Finance Automation for a Media & Entertainment Company
The Problem
A film and television production company was running its entire production finance workflow through ad hoc spreadsheets. Line producers submitted financials in inconsistent formats. Production accountants manually reconciled general ledger actuals against budget codes, then hand-updated Excel cash spend grids, literally bolding cells to flag actualized spend. Budget vs. actuals reporting required consolidating data across projects one at a time.
Each accountant could manage five to nine projects at that pace. The work was brittle, repetitive, and impossible to scale. As the production slate grew, the finance team hit a ceiling that no amount of hiring could fix efficiently. The bottleneck was the workflow.
The Solution
We replaced the spreadsheet-based process with a structured data pipeline built on straightforward tooling. No frontier AI models. No complex machine learning. Solid data engineering applied to a well-understood operational problem.
Standardized intake. A web-based form (mobile and desktop) replaced the inconsistent spreadsheet submissions. Line producers enter data in a structured format that feeds directly into a Postgres relational database. The form enforces consistency at the point of entry rather than relying on accountants to clean it up downstream.
Automated reconciliation. Python scripts pull GL actuals and match them to production budget codes, handling the messy source formats (Excel, CSV, inconsistent templates from different production entities) that previously required manual interpretation.
Automated cash spend grid updates. The system updates Excel cash spend grids automatically, including the actualization flags that were previously done by hand. The bolding step that used to eat hours of accountant time now happens programmatically.
Exception-based review. Instead of accountants touching every transaction, the system generates an exceptions queue for mismatches and ambiguous coding. Humans review only the edge cases. Everything that matches cleanly flows through untouched.
Reporting layer. Budget vs. actuals views and rollups live outside Excel, so finance leadership can see status across all active productions without waiting for manual consolidation.
The Result
Accountant throughput increased roughly 5x. Team members went from managing 5 to 9 projects each to handling 20 or more. The annual labor savings landed between $200K and $300K against a total build cost of $118K. Year-one ROI came in between 170% and 250%.
The project is a clean example of a principle we apply across engagements: the work that delivers the most value is often the data engineering, not the AI. Get the pipeline right, and the intelligent layer becomes a natural extension. Get it wrong, and the AI has nothing reliable to work with.