Ross Rondeau

Why I Built This: The Hidden Payroll of "Free" In-House Software

In enterprise architecture, teams often justify building custom in-house software or self-hosted AI pipelines by comparing server bills directly against vendor SaaS license quotes. They overlook the real expense: ongoing maintenance engineering payroll.

This calculator calculates the true break-even volume by factoring in fully loaded engineering salaries, maintenance headcount, schedule overrun risks, and serverless compute floors.

Architectural Assumptions & Mechanics

1. Fully Loaded Payroll: Base engineering salaries are loaded at 1.4445x using U.S. Bureau of Labor Statistics (BLS) ECEC data to capture benefits, taxes, and overhead.
2. Schedule Slippage Reality: Software projects frequently exceed initial timelines. The model defaults to 1.00x with 1.27x mean industry overrun and 3.00x tail risk options based on empirical research (Bent Flyvbjerg).
3. Maintenance Floor: Maintenance is modeled in real engineering headcount (FTEs). Even a single dedicated maintenance engineer creates a significant ongoing payroll baseline.
4. API Rate Cards: Vendor costs reflect current enterprise LLM token pricing (Anthropic Claude rate cards) with prompt caching and batch execution discounts.
5. Serverless Unit Costs: Internal execution costs are modeled at an AWS serverless baseline floor of $0.00000128 per transaction.
6. The Break-Even Threshold: At low or moderate transaction volume, maintenance payroll alone usually dwarfs third-party API invoices. Custom builds only yield positive ROI once transaction volume scales past the calculated break-even point.