Decision Making Models, Techniques & Business Intelligence

TAKEAWAYS:

• Rational vs. bounded rationality: Decisions are idealized as logical (rational) but constrained by time, data, and cognitive limits (bounded), requiring trade-offs.
• Structured vs. unstructured decisions: Routine problems (e.g., reordering stock) use rules; complex ones (e.g., launching a new product) need creative analysis.
• BI tools bridge the gap: Dashboards (e.g., Power BI), data mining, and predictive analytics turn raw data into actionable insights (e.g., NEPSE’s stock trend alerts).
• Group decision traps: Avoid biases like groupthink (e.g., Pathao’s failed bike-sharing pilot) or anchoring (e.g., overvaluing legacy systems like NTC’s manual billing).
• Ethics in decisions: BI can reinforce biases (e.g., loan approval algorithms favoring urban areas) or invade privacy (e.g., Khalti’s transaction tracking).
• Agile decision-making: Iterative models (e.g., Scrum) and real-time data (e.g., traffic apps like Pathao) adapt to dynamic environments like Kathmandu’s congestion.

1. Decision-Making Frameworks: From Intuition to Analytics

Decisions range from programmed (repetitive, rule-based) to non-programmed (one-time, strategic). Organizations use frameworks to reduce uncertainty:

A. Rational Decision-Making Model

A step-by-step approach assuming perfect information and logic:

flowchart TD
    A["Identify Problem"] --> B["Gather Data"]
    B --> C["List Alternatives"]
    C --> D["Evaluate Criteria"]
    D --> E["Choose Best Option"]
    E --> F["Implement & Monitor"]

Worked Example: Nabil Bank’s Loan Approval

• Problem: High default rates on SME loans.
• Data: Credit scores, collateral, past repayment history (from core banking systems).
• Alternatives: Tiered interest rates, collateral requirements, or AI-driven risk scoring.
• Criteria: Default risk <5%, ROI >12%, regulatory compliance.
• Choice: Hybrid model (human + AI) reduced defaults by 20%.

Limitations:

• Assumes perfect information (rare in practice).
• Ignores time pressure (e.g., Daraz’s same-day delivery decisions).
• Bounded rationality (Herbert Simon’s theory): Humans simplify choices using heuristics (e.g., "if X, then Y" rules).

B. Bounded Rationality and Satisficing

• Satisficing: Choosing the "good enough" option due to cognitive limits (e.g., a startup picking the first investor over the best offer).
• Example: Khalti’s Payment Gateway
  • Rational goal: Maximize merchant adoption and transaction fees.
  • Bounded reality: Limited to 3 months to launch; settled for basic UPI integration instead of a custom blockchain solution.

2. Decision-Making Techniques

A. SWOT Analysis

A strategic planning tool to evaluate internal/external factors:

mindmap
  root((SWOT Analysis))
    Strengths["Internal: What we do well (e.g., Ncell’s 4G network)"]
    Weaknesses["Internal: Gaps (e.g., NTC’s slow digital transformation)"]
    Opportunities["External: Trends (e.g., Nepal’s fintech boom)"]
    Threats["External: Risks (e.g., competition from Reliance Jio)"]

Worked Example: Himalayan Java’s Expansion

When to Use:

• Pros: Simple, qualitative, good for brainstorming.
• Cons: Subjective (e.g., "strong brand" is vague); ignores quantitative data.

B. PESTEL Analysis

Macro-environmental factors affecting strategy:

mindmap
  root((PESTEL Analysis))
    Political["Government policies (e.g., Nepal’s 15% VAT on e-commerce)"]
    Economic["Inflation, GDP growth (e.g., NEPSE’s volatility)"]
    Social["Demographics, trends (e.g., Pathao’s gig economy appeal)"]
    Technological["AI, blockchain (e.g., Khalti’s QR codes)"]
    Environmental["Sustainability laws (e.g., Daraz’s plastic ban)"]
    Legal["Labor laws, data privacy (e.g., Nepal’s 2018 Data Protection Act)"]

Example: Toyota Kirloskar’s Nepal Entry

• Political: Nepal’s "Make in Nepal" policy → local production incentives.
• Economic: Rising middle-class demand for hybrid cars.
• Technological: Need for electric vehicle (EV) charging infrastructure (currently lacking).

C. Decision Trees

Visualize probabilistic outcomes for risk analysis:

flowchart TD
    A["Launch New Product"] --> B["Market Success (70%)"]
    A --> C["Market Failure (30%)"]
    B --> D["Profit: $500K"]
    C --> E["Loss: $200K"]

Worked Example: Daraz’s Private Label Strategy

• Decision: Should Daraz launch its own electronics brand?
  • Success path: 60% chance → $3M profit (gaining market share).
  • Failure path: 40% chance → $1M loss (cannibalizing seller base).
  • Net Present Value (NPV): $1.4M → Proceed.

Advantages:

• Quantifies uncertainty.
• Helps prioritize high-impact decisions (e.g., Ncell’s 5G rollout).

Disadvantages:

• Requires accurate probability estimates (hard for disruptive innovations).
• Ignores qualitative factors (e.g., brand reputation).

3. Business Intelligence (BI) for Data-Driven Decisions

BI transforms data into actionable insights using:

• Descriptive analytics: "What happened?" (e.g., NTC’s monthly call drop reports).
• Predictive analytics: "What will happen?" (e.g., Ncell’s churn prediction).
• Prescriptive analytics: "What should we do?" (e.g., Daraz’s dynamic pricing).

A. BI Tools in Nepal

B. Data Mining for Patterns

• Example: Khalti’s Fraud Detection
  • Algorithm: Uses clustering to flag unusual transaction patterns (e.g., sudden large transfers to new accounts).
  • Outcome: Reduced fraud losses by 35% in 2023.

Challenges:

• Garbage in, garbage out (GIGO): Poor data quality → wrong decisions (e.g., NTC’s billing errors).
• Privacy risks: BI tools may violate laws (e.g., tracking user location without consent).

4. Group Decision-Making: Pitfalls and Best Practices

Groups often make better decisions but face biases:

A. Common Biases

B. Nominal Group Technique (NGT)

A structured approach to avoid dominance by a few:

flowchart LR
    A["1. Silent Idea Generation"] --> B["2. Round-Robin Sharing"]
    B --> C["3. Vote on Priorities"] --> D["4. Discuss & Rank"]

Example: Chaudhary Group’s New Product Launch

• Step 1: Team writes down ideas (e.g., "organic snacks," "EV chargers").
• Step 2: Each shares one idea; no debate.
• Step 3: Vote using dot voting (e.g., 3 dots per person).
• Result: "Organic snacks" selected for pilot in Pokhara.

Advantages:

• Reduces social pressure.
• Ensures all voices are heard.

5. Enhancing Decision-Making with Technology

A. Expert Systems

Rule-based AI that mimics human expertise:

• Example: Medical Diagnosis in Nepal
  • System: "HealthKatha" (a Nepali AI tool) uses symptoms to suggest diseases.
  • Rules: "If fever + rash → likely dengue."
  • Outcome: Reduces misdiagnosis in rural clinics.

B. Real-Time Decision Support

• Example: Pathao’s Traffic Routing
  • Data: GPS, weather, road closures (from NTC).
  • Algorithm: Adjusts rider routes dynamically (e.g., avoids Lalitpur’s Durbar Square during festivals).
  • Impact: 25% faster deliveries.

In the Real World

1. Ncell’s Churn Prediction

   • Idea Used: Predictive analytics (BI).
   • How: Analyzes call duration, data usage, and payment history to flag at-risk customers. Sends personalized offers (e.g., "Free 1GB data") to retain them.
   • Result: Reduced churn by 18% in 2023.

2. Daraz’s Dynamic Pricing

   • Idea Used: Prescriptive analytics + real-time data.
   • How: Adjusts prices based on inventory levels, competitor actions, and local demand (e.g., higher prices in Kathmandu vs. rural areas).
   • Example: Diwali season → prices rise 10–15% for electronics.

3. Nabil Bank’s Loan Approval AI

   • Idea Used: Decision trees + machine learning.
   • How: Traditionally, loans required collateral. Now, the bank uses AI to assess alternative data (e.g., digital footprints like Khalti transactions, social media activity).
   • Impact: Approved 40% more SME loans in 2023 without increasing defaults.

Case Study: Toyota Kirloskar Nepal’s EV Strategy

Problem: Nepal’s transport sector is polluted (90% of vehicles run on fossil fuels). Toyota wants to launch EVs but faces challenges.

Decision-Making Process:

1. SWOT Analysis:

   • Strengths: Toyota’s global reputation; existing hybrid models.
   • Weaknesses: High EV cost ($30K+); limited charging infrastructure.
   • Opportunities: Government subsidies for EVs; rising fuel prices.
   • Threats: Consumer skepticism; competition from Chinese brands.

2. PESTEL Analysis:

   • Political: Government’s "Electric Vehicle Policy 2020" offers tax breaks.
   • Technological: Charging stations are being installed in Kathmandu (e.g., at Thamel).
   • Economic: Middle-class growth → demand for affordable EVs.

3. Decision Tree:

   • Option 1: Launch now with high price → 60% chance of slow sales.
   • Option 2: Wait 2 years for tech improvements → 70% chance of success but lose market share.
   • Choice: Pilot a low-cost EV model (e.g., $20K) in Kathmandu first.

Outcome: Toyota partnered with Clean Energy Nepal to build charging stations and launched the Toyota Prius Hybrid as a stepping stone.

Exam Tip

What Examiners Look For

1. Definitions with Examples:

   • Don’t just write "SWOT analysis is a tool." Explain how Nabil Bank used it to enter the digital banking space.
   • Marks lost: Vague answers like "BI helps in decision-making."

2. Comparisons:

   • Always compare two techniques (e.g., SWOT vs. PESTEL) in a table with Nepali examples.
   • Example:

     Criteria	SWOT	PESTEL
     Focus	Internal/External factors	Macro-environmental factors
     Example	Himalayan Java’s coffee expansion	Toyota’s EV market entry
     Limitations	Subjective	Ignores competitors (use Porter’s Five Forces for that)

3. Worked Examples:

   • Must include:
     • Problem statement.
     • Data used (e.g., "Ncell’s call logs").
     • Decision made (e.g., "Offer discounts to high-risk customers").
     • Outcome (e.g., "Churn reduced by 18%").
   • Avoid: Generic examples like "a bank loan."

4. Diagrams:

   • Decision trees, SWOT mindmaps, and BI dashboards are high-mark questions.
   • Label every node (e.g., in a decision tree, write "Launch Product (70%)" not just "Success").

5. Ethical/Social Issues:

   • Always link BI to ethics:
     • "How might Khalti’s transaction data be misused?" → Privacy concerns.
     • "Can Daraz’s dynamic pricing be unfair?" → Discrimination against rural users.

6. Short vs. Long Answers:

   • Short (3 marks): Define one technique (e.g., "Explain PESTEL analysis").
   • Long (10 marks): Compare two techniques + one Nepali case study + advantages/disadvantages.

Common Mistakes to Avoid

• Ignoring the "real-world" link: Examiners want Nepali examples. Never use generic cases like "Coca-Cola."
• Overlooking limitations: Every tool has flaws (e.g., SWOT is subjective). Always mention one.
• Poor structure: Use headings like:
  • Step 1: Problem Identification
  • Step 2: Data Collection
  • Step 3: Analysis (e.g., SWOT)
  • Step 4: Decision & Implementation

Final Checklist Before Submitting: ✅ Used at least 3 Nepali company examples (Ncell, Daraz, Nabil Bank, etc.). ✅ Included 2 visuals (e.g., decision tree + SWOT mindmap). ✅ Compared two techniques in a table. ✅ Linked ethics/social issues to BI. ✅ Answered how, not just what (e.g., "How did Pathao use real-time data?" not "Pathao uses data.").