ROI Formulas

All formulas are in mcp-server/lib/constants.js. They are deterministic and documented.

Components

Productivity Savings

seats × avg_developer_cost_annual × velocity_improvement_pct × 0.6

The 0.6 is a utilization factor — not all dev time is code-writing.

Bug Remediation Savings

seats × 12 × bug_escape_reduction_pct × avg_bug_remediation_cost

12 = industry benchmark for production bugs per developer per year.

Onboarding Savings

new_hires × (avg_developer_cost_annual / 260) × (14 × onboarding_time_reduction_pct)

• new_hires = seats × new_hire_rate_annual
• 260 = working days per year
• 14 = baseline onboarding days to first meaningful commit

Scenarios

Scenario	Velocity	Bug Escape Reduction	Onboarding Reduction
Conservative	20%	25%	35%
Base	32%	38%	50%
Aggressive	45%	45%	55%

METR Caveat

The METR 2025 RCT found experienced devs 19% slower with early-2025 AI tools. Rebuttal points:

• Study tested pre-Agent Mode, pre-Subagent tools
• No Cursor Rules or deployment strategy in the study
• Salesforce/NVIDIA results are post-strategy, not just post-tool

Always acknowledge the METR study — it builds credibility. Then explain why deployment strategy changes outcomes.