Cost Estimation & Financial Management in Software Projects

TAKEAWAYS:

• Cost estimation predicts effort, time, and budget using models like COCOMO II, adjusted by project complexity multipliers.
• Financial management includes budgeting, cash flow analysis, and ROI calculation to ensure project viability.
• Risk factors (e.g., scope changes, resource shortages) must be quantified and mitigated in cost plans.
• COCOMO II uses post-architecture multipliers (e.g., product attributes, platform factors) to refine estimates.
• Earned Value Management (EVM) tracks project health by comparing planned vs. actual costs and progress.
• Contract types (fixed-price, time-and-material) impact cost control and risk allocation.

1. Why Cost Estimation Matters in Software Projects

Software projects fail 70% of the time due to poor cost estimation (Standish Group). Unlike hardware, software costs are intangible—effort, not materials, drives expenses. Key challenges:

• Uncertainty: Requirements evolve; scope changes.
• Hidden costs: Training, maintenance, and third-party tools.
• Stakeholder pressure: Underpromising leads to lost contracts; overpromising risks bankruptcy.

2. Cost Estimation Models

sequenceDiagram
    participant Client as eSewa Mobile App
    participant Server as eSewa Server
    participant DB as Database
    Client->>Server: POST /api/transaction (10,000 KDSI)
    Server->>DB: Query (RELY=1.2, CPLX=1.1)
    DB-->>Server: Response
    Server-->>Client: Success (ACAP=0.85)
    Note right of Server: COCOMO II multipliers applied
    Note over Client,Server: Effort = 2.4 × (10)^1.05 × 1.2 × 1.1 × 0.85 ≈ 35.5 PM

A. COCOMO II (Constructive Cost Model)

Developed by Barry Boehm, COCOMO II estimates effort in person-months and cost based on:

1. Project size (measured in KDSI: thousands of delivered source instructions).
2. Effort multipliers (adjusted for complexity, team experience, tools used).

Formula:

Effort (PM) = a × (KDSI)^b × ∏(EMi)
Cost ($) = Effort × Salary Rate × (1 + Overhead)

Where:

• a, b = model constants (e.g., for Organic projects: a=2.4, b=1.05).
• EMi = Effort Multipliers (15 factors, e.g., RELY for reliability, CPLX for complexity).

Post-Architecture Multipliers (Key for Exams):

Worked Example: Estimating a Mobile App for eSewa

• Scope: Payment gateway app (10,000 KDSI).
• Multipliers:
  • RELY = 1.2 (high reliability needed).
  • CPLX = 1.1 (moderate complexity).
  • TIME = 1.0 (no strict deadline).
  • ACAP = 0.85 (experienced team).
• Effort:

  Effort = 2.4 × (10)^1.05 × 1.2 × 1.1 × 1.0 × 0.85 ≈ 35.5 person-months
  

• Cost (assuming $500/month salary + 20% overhead):

  Cost = 35.5 × $500 × 1.20 = **$21,300**
  

Advantages of COCOMO II: ✅ Accurate for large projects (used by NASA, IBM). ✅ Adjustable for agile/iterative development. ✅ Quantifies risk via multipliers.

Limitations: ❌ Requires historical data (KDSI). ❌ Overhead for small projects (<5 person-months).

B. Parametric Models (Function Points)

• Function Points (FP): Measure software size by user functions (inputs, outputs, queries).
• Conversion: 1 FP ≈ 100–200 lines of code (varies by language).
• Example: A Daraz order-tracking system might have:
  • 50 user inputs (e.g., "Add to Cart").
  • 30 reports (e.g., "Order History").
  • Total FP = 80 → Estimated effort = 80 × 5 PM/FP = 400 person-months.

Comparison Table: COCOMO II vs. Function Points

3. Financial Management in Projects

A. Budgeting

• Top-Down: Management allocates funds based on high-level goals.
• Bottom-Up: Teams estimate tasks and sum costs (more accurate).
• Example: Ncell’s app development budget:
  • Salaries: 60% (developers, testers).
  • Tools: 20% (Jira, AWS).
  • Contingency: 15% (unexpected risks).
  • Marketing: 5%.

B. Cash Flow Analysis

• Key Metrics:
  • Net Present Value (NPV): Discounted future cash flows.

    NPV = Σ [CFt / (1 + r)^t] – Initial Investment
    

    Where CFt = cash flow at time t, r = discount rate (e.g., 10%).
  • Return on Investment (ROI):

    ROI = (Net Profit / Cost) × 100%
    

  • Break-Even Point: When cumulative cash flow turns positive.

Worked Example: Khalti’s Loan Disbursement System

• Initial Cost: $50,000 (development + servers).
• Annual Revenue: $20,000 (transaction fees).
• NPV (5 years, 10% discount rate):

  Year 1: $20,000 / 1.1 = $18,182
  Year 2: $20,000 / 1.1² = $16,529
  ...
  NPV = $73,582 – $50,000 = **$23,582 (profitable)**
  

C. Earned Value Management (EVM)

Tracks three key metrics:

1. Planned Value (PV): Budgeted cost of work scheduled.
2. Earned Value (EV): Budgeted cost of work completed.
3. Actual Cost (AC): Real cost incurred.

Formulas:

• Cost Variance (CV): EV – AC (positive = under budget).
• Schedule Variance (SV): EV – PV (positive = ahead of schedule).
• Cost Performance Index (CPI): EV / AC (>1 = efficient).

Example: Pathao Driver App Development

Corrective Actions:

• Hire more developers.
• Cut non-critical features (e.g., AI route optimization).

4. Risk Management in Cost Estimation

Top 5 Cost Risks in Software Projects:

1. Scope Creep: Uncontrolled changes (e.g., adding biometric login to eSewa).
2. Resource Shortages: Key developers leave mid-project.
3. Technical Debt: Cutting corners on testing → future fixes.
4. Vendor Delays: Third-party APIs (e.g., payment gateways) fail.
5. Inflation: Rising salaries (e.g., 15% annual hike in Nepal).

Mitigation Strategies:

• Contingency Reserve: 10–20% of budget for risks.
• Prototyping: Validate requirements early (e.g., Daraz’s MVP).
• Agile Buffers: Short sprints allow course correction.

Probability vs. Impact for software projects (Image: Peter Gladdish, CC BY 4.0, via Wikimedia Commons)

5. Contract Types and Cost Control

stateDiagram-v2
    [*] --> FixedPrice
    FixedPrice --> Delivered : On Time
    FixedPrice --> Dispute : Late/Defective
    TimeAndMaterial --> Billed : Per Hour
    TimeAndMaterial --> Adjust : Scope Change
    FixedPrice --> [*]
    TimeAndMaterial --> [*]
    note right of FixedPrice
        Risk: All on vendor
        Cost: Predictable
    note right of TimeAndMaterial
        Risk: Shared
        Cost: Flexible

Example: NEPSE’s Trading Platform Contract

• Type: Fixed-Price ($250,000).
• Risk: Vendor must deliver on time; penalties for delays.
• Contingency: 15% reserve for market volatility.

In the Real World

1. eSewa’s Cost Estimation

   • Used COCOMO II to estimate their digital wallet app (20,000 KDSI).
   • Multipliers applied:
     • RELY = 1.4 (high security needed).
     • CPLX = 1.3 (integrations with banks, NFC).
   • Result: 45 person-months → $225,000 budget (saved 20% via agile sprints).

2. Khalti’s Financial Management

   • NPV Analysis: Delayed launch by 6 months → lost $500K in transaction fees.
   • Solution: Used EVM to track progress and reallocated funds to marketing.

3. Daraz’s Risk Mitigation

   • Scope Creep: Initially planned a simple e-commerce site but added logistics tracking.
   • Fix: Used agile buffers (2-week sprints) to manage changes without blowing the $5M budget.

Exam Tip

How to Score Full Marks:

1. COCOMO II Questions:

   • Always list 3–4 multipliers with their values (e.g., RELY, CPLX).
   • Show the effort formula step-by-step.
   • Compare with function points in 2–3 bullet points.

2. Financial Management:

   • For NPV/ROI, show calculations with time value of money.
   • For EVM, plot a simple table with PV, EV, AC, and interpret CV/SV.

3. Contract Management:

   • Define fixed-price vs. T&M and give a Nepali example (e.g., NTC vs. Pathao).
   • Link to risk allocation (e.g., "Fixed-price shifts risk to the vendor").

4. Worked Examples:

   • Use real companies (eSewa, Khalti) and local scenarios (e.g., "Estimate cost for a school management app").
   • Assume salary rates (e.g., $300/month for a Nepali developer) if not given.

Avoid: ❌ Vague answers like "COCOMO is used for estimation." ✅ Instead: "COCOMO II’s post-architecture multipliers adjust effort for factors like RELY (reliability), where a high value (1.4) for eSewa’s payment system increases effort by 40%."

sequenceDiagram
    participant Client
    participant Vendor
    participant Bank
    Client->>Vendor: Fixed-Price Contract ($250K for NEPSE platform)
    Vendor->>Bank: Loan Application (Collateral: IP Rights)
    Bank-->>Vendor: Approval ($200K, 10% interest)
    Vendor->>Client: Milestone 1 (Login System)
    Client->>Vendor: Payment ($50K)
    Vendor->>Vendor: EVM Check (EV=$45K, AC=$60K → CV=-$15K)
    Vendor->>Client: Escalation (Delay Risk)
    Client->>Vendor: Approve Contingency Fund ($20K)

In the real world

• eSewa uses COCOMO II with RELY=1.2 (high reliability) and TIME=1.0 (no strict deadline) to estimate payment gateway development costs, adjusting for Nepal’s regulatory compliance needs (hidden cost: 15% contingency).
• Khalti applies Earned Value Management (EVM) to track loan disbursement system progress, where a CPI < 1 (e.g., 0.71 in Pathao’s example) triggered reallocation of developers to backend services to meet the 2-year execution deadline (TIME multiplier).
• Ncell’s app budgeting allocates 60% to salaries (reflecting Nepal’s high developer rates) and 20% to AWS tools, using bottom-up budgeting to align with COCOMO II’s ACAP=0.85 (experienced team) multiplier.