Project Sizing Guide

Complexity Assessment Rules

ILF / EIF Complexity (based on DET – Data Element Types and RET – Record Element Types):

EI Complexity (based on DET and FTR – File Types Referenced):

EO / EQ Complexity (based on DET and FTR):

Converting Function Points to Effort

After calculating Unadjusted Function Points (UFP):

1. Calculate the Value Adjustment Factor (VAF) (optional; deprecated since IFPUG 4.3+ but still used by some teams)

   • 14 General System Characteristics (GSC), each scored 0–5
   • VAF = 0.65 + 0.01 × Σ(GSC)
   • Adjusted Function Points AFP = UFP × VAF

2. Function points to person-hours

   • Industry benchmark: 8–15 person-hours per function point (varies by language and team maturity)

   Technology Stack	Person-hours / FP	Notes
   Low-code / Mature Frameworks	4 – 8	Many reusable components available
   Python / JS / Modern Web	8 – 12	Mainstream development productivity
   Java / C# Enterprise	10 – 15	Includes architecture and standards overhead
   Embedded / C / C++	15 – 25	High debugging and testing cost
   Legacy System Maintenance	20 – 30	Comprehension and regression cost

Steps

1. Identify function components: List all ILFs, EIFs, EIs, EOs, and EQs
2. Assess complexity: Rate each component as Low / Medium / High
3. Calculate UFP: Sum (count × weight) for all components
4. Select conversion factor: Choose person-hours per FP based on technology stack
5. Compute total effort: UFP × conversion factor
6. Add buffer: A 15–30% management and risk buffer is recommended

Estimation Adjustment Factor Checklist

After completing the estimation, verify that the following factors have been accounted for:

Technical Factors

• Technology stack familiarity (Is the team experienced? If unfamiliar, add 30–50%)
• Technical debt (Poor legacy code quality? Add 20–40%)
• Third-party dependencies (Unstable APIs? Missing documentation? Add 10–30%)
• Performance / security requirements (Special non-functional requirements? Add 15–25%)

Team Factors

• Team size (Communication overhead increases significantly above 5 people; add ~5% per person)
• Personnel turnover risk (Key members may leave? Add 15–25%)
• Parallel projects (Team context-switching across multiple projects? Add 20–30%)
• Onboarding new members (New hires? Expect ~50% reduced efficiency for the first 2 weeks)

Process Factors

• Requirements stability (Requirements likely to change? Add 20–50%)
• Approval processes (Multiple layers of approval needed? Add 10–20%)
• Deployment complexity (Multi-environment, multi-region deployments? Add 10–15%)
• Compliance requirements (Audit or compliance processes? Add 15–30%)

Commonly Underestimated Work

• Code review: +10–15%
• Unit test authoring: +15–25%
• Integration / E2E testing: +10–20%
• Documentation: +5–15%
• Bug fixing and regression: +10–20%
• Environment setup and DevOps: +5–10%
• Meetings and communication: +10–15%

Estimation Output Template

After the Agent completes the estimation, it should produce output in the following format:

## Estimation Report: [Project / Feature Name]

### Estimation Method: [PERT / T-shirt / FPA]

### Work Package Breakdown

| # | Work Package | O (person-days) | M (person-days) | P (person-days) | E (person-days) | σ |
|---|-------------|-----------------|-----------------|-----------------|-----------------|---|
| 1 | xxx         | x               | x               | x               | x.x             | x.x |
| 2 | xxx         | x               | x               | x               | x.x             | x.x |

### Summary

- Total expected effort: X person-days
- Total standard deviation: X person-days
- 68% confidence interval: X – X person-days
- 90% confidence interval: X – X person-days
- 95% confidence interval: X – X person-days

### Adjustment Factors
- [Factors considered and adjustments applied]

### Final Recommendation
- For internal planning: X person-days (90% confidence)
- For external quotes: X person-days (95% confidence)

### Risk Alerts
- [Key risk items and mitigation suggestions]

Calculation Tool

The scripts/estimate_calculator.py script supports numerical calculations for all three estimation methods:

# Three-Point Estimation (PERT)
python3 scripts/estimate_calculator.py --method pert \
  --tasks '[{"name":"Login Module","O":2,"M":3,"P":8},{"name":"Payment Module","O":5,"M":10,"P":20}]'

# T-shirt Size Conversion
python3 scripts/estimate_calculator.py --method tshirt \
  --tasks '[{"name":"Login Module","size":"M"},{"name":"Payment Module","size":"L"}]'

# Function Point Analysis
python3 scripts/estimate_calculator.py --method fpa \
  --components '[{"type":"ILF","complexity":"medium","count":3},{"type":"EI","complexity":"low","count":5}]' \
  --hours-per-fp 10

References

• IFPUG (International Function Point Users Group) CPM 4.3.1
• PMI PMBOK Guide — 6th Edition, Section 6.4: Estimate Activity Durations
• Steve McConnell, Software Estimation: Demystifying the Black Art
• Mike Cohn, Agile Estimating and Planning