Case Studies & Practical Applications in Service Operations

TAKEAWAYS:

• Case studies bridge theory and practice by illustrating how service operations principles (e.g., queuing theory, quality management) are applied in real businesses like Ncell, Daraz, or Nabil Bank.
• Process analysis in cases reveals inefficiencies (e.g., long customer wait times at eSewa) and solutions (e.g., automated call routing or inventory optimization at Himalayan Java).
• Strategic decisions (e.g., Daraz’s logistics expansion or NTC’s network upgrades) show how companies balance cost, quality, and customer satisfaction using tools like SWOT, PESTEL, or game theory.
• Quality tools (e.g., Six Sigma at Toyota or ISO certification at Chaudhary Group) are demonstrated through case failures/successes, teaching students how to diagnose and improve service delivery.
• Transportation/assignment problems are solved in cases like Pathao’s driver routing or NTC’s fiber-optic cable deployment, linking academic models to operational logistics.
• Exam focus: Expect data-driven analysis of cases (e.g., "How would you reduce wait times at a bank?"), requiring you to apply queuing models, cost-benefit analysis, or quality tools to hypothetical scenarios.

1. Why Case Studies Matter in Service Operations

Service operations management is not just about theories—it’s about solving real problems in dynamic environments. Case studies help you:

• See theory in action: Learn how companies like Ncell (telecom), Daraz (e-commerce), or Nabil Bank (finance) apply concepts like queuing theory, Six Sigma, or transportation models to daily operations.
• Develop critical thinking: Analyze failures (e.g., eSewa’s server crashes during Dashain) and successes (e.g., Pathao’s surge pricing during festivals) to propose improvements.
• Prepare for exams: Past questions often ask you to "analyze a case and suggest operational improvements"—this unit teaches you how to structure such answers.

How Cases Are Structured in Exams

Exam cases typically include:

1. Background: Company profile, industry, and current challenges.
2. Data: Metrics (e.g., customer wait times, defect rates, delivery delays).
3. Problems: Gaps in quality, efficiency, or customer satisfaction.
4. Questions: "How would you improve X using Y concept?"

Example from past exams:

"BIROI Electronics faces long customer call wait times. Analyze using queuing theory and suggest solutions." Your answer must:

• Identify the M/M/1 queuing model (single server, Poisson arrivals).
• Calculate average wait time (Wq) using $L_q=\frac{{\lambda }^2}{\mu (\mu -\lambda )}$.
• Propose fixes (e.g., add servers, prioritize calls, or use IVR).

2. Real-World Cases: Where Theory Meets Practice

Case 1: Ncell’s Network Optimization (Queuing Theory + Transportation)

Company: Ncell (Nepal’s largest telecom provider) Challenge: During festivals (Dashain, Tihar), call drop rates spike due to network congestion (too many users sharing limited towers). Theory Applied:

• Queuing Theory: Customers (arrivals) vs. network capacity (service rate).
• Transportation Problem: Optimizing base station locations to reduce dead zones.

How Ncell Solves It:

1. Dynamic Bandwidth Allocation: Uses M/M/c queuing (multiple servers = multiple frequency bands) to handle peak loads.
2. Predictive Tower Placement: Solves transportation assignment problems to place towers where demand is highest (e.g., Kathmandu Valley vs. rural areas).
3. Customer Prioritization: Uses non-preemptive priority queues (e.g., emergency calls get faster access).

Visual: Ncell’s Network Congestion Model

flowchart TD
    A["Customers (Poisson arrivals: λ)"] -->|"Arrive"| B["Network Queue"]
    B -->|"Service Rate: μ"| C["Active Calls"]
    C -->|"Complete"| D["Exit"]
    B -->|"Blocked if queue full"| E["Call Drops"]
    F["Solution: Add Towers"] -->|"Increases μ"| B
    G["Solution: Prioritize Calls"] -->|"Reduces λ for non-urgent calls"| B

Exam Link:

"If Ncell’s average call arrival rate (λ) is 100 calls/min and service rate (μ) is 80 calls/min, calculate the probability of a customer waiting (Lq). How would you reduce Lq by 50%?" Answer:

• Use $L_q=\frac{{\lambda }^2}{\mu (\mu -\lambda )}=\frac{100^2}{80(80-100)}=31.25$ calls in queue.
• To reduce $L_q$ by 50%, either:
  • Increase μ to 120 calls/min (add towers).
  • Or reduce λ by offering SMS for non-urgent queries.

Case 2: Daraz’s Logistics: The Transportation Problem

Company: Daraz (Nepal’s Amazon) Challenge: Last-mile delivery delays in Kathmandu due to traffic and inefficient routes. Theory Applied: Transportation Model (minimizing cost/time for deliveries).

How Daraz Solves It:

1. Hub-and-Spoke Model:
   • Hub: Central warehouse in Kathmandu.
   • Spokes: Local delivery centers in Thapathali, Gaushala, etc.
2. Vehicle Routing Problem (VRP):
   • Uses algorithms to optimize delivery routes (e.g., a driver in Lalitpur serves orders in Thapathali → Gaushala → Koteshwor).
   • Reduces empty return trips by clustering orders geographically.

Visual: Daraz’s Delivery Network

mindmap
  root((Daraz Logistics))
    Hub["Central Warehouse (Kathmandu)"]
    Spokes["Local Delivery Centers"]
      Thapathali["Zone 1: 500 orders/day"]
      Gaushala["Zone 2: 300 orders/day"]
      Koteshwor["Zone 3: 200 orders/day"]
    Routes["Optimized Paths"]
      Example["Driver Route: Thapathali → Gaushala → Koteshwor → Hub"]
    Cost["Minimized by:"]
      Distance["Shortest path algorithms"]
      Traffic["Avoiding peak hours"]
      Fuel["Reduced idle time"]

Exam Link:

"Daraz has 3 warehouses (A, B, C) supplying 4 zones (1, 2, 3, 4). The cost matrix is given. Find the optimal distribution using the Northwest Corner Rule." Answer: Use the Northwest Corner Method (a transportation algorithm) to allocate shipments:

From\To	Zone 1	Zone 2	Zone 3	Zone 4	Supply
A	5	3	4	2	100
B	1	2	3	4	150
C	4	1	2	3	100
Demand	120	80	60	90	450

Steps:

1. Allocate to A→Zone 1 (100 units) (meets A’s supply).
2. Next, B→Zone 1 (20 units), then B→Zone 2 (80 units) (meets Zone 2’s demand).
3. Continue until all demands/supplies are met. Total Cost: Calculate as $\sum (units\times cost)$.

Case 3: Nabil Bank’s Loan Approval (Decision Making + Game Theory)

Company: Nabil Bank Challenge: Loan default rates due to poor risk assessment. Theory Applied:

• Decision Trees: Weighing risk vs. reward in loan approvals.
• Game Theory: Modeling borrower vs. bank strategies (e.g., will the borrower default?).

How Nabil Bank Solves It:

1. Decision Tree for Loan Approval:

   • Node 1: Check credit score (high/low).
   • Node 2: Assess collateral (yes/no).
   • Outcome: Approve/reject based on expected value (probability of repayment × loan amount).

2. Game Theory: Borrower vs. Bank Payoff Matrix

   • Bank’s Strategies: Approve/Reject.
   • Borrower’s Strategies: Repay/Default.
   • Nash Equilibrium: The bank approves only if repayment probability > default risk.

Visual: Nabil Bank’s Loan Decision Tree

flowchart TD
    A["Start"] --> B["Credit Score High?"]
    B -->|"Yes"| C["Collateral Available?"]
    C -->|"Yes"| D["Approve Loan (90% chance repayment)"]
    C -->|"No"| E["Reject (Low Risk)"]
    B -->|"No"| F["Collateral Available?"]
    F -->|"Yes"| G["Approve with Guarantor (70% repayment)"]
    F -->|"No"| H["Reject (High Risk)"]

Exam Link:

"Nabil Bank faces a borrower with a 60% chance of repayment. The loan is Rs. 500,000. If defaulted, the bank loses Rs. 500,000 + Rs. 50,000 (legal costs). Should the bank approve? Use expected value." Answer:

• Expected Value (EV) of Approval: $EV=(0.6\times 500,000)+(0.4\times -550,000)=300,000-220,000=Rs.80,000$.
• EV of Rejection: Rs. 0 (no loss, but no profit).
• Decision: Approve, since EV > 0.

Case 4: eSewa’s Server Crashes (Queuing + Capacity Planning)

Company: eSewa (Nepal’s leading digital payment platform) Challenge: Server overload during Dashain, causing transaction failures. Theory Applied: M/M/c Queuing Model (multiple servers = multiple payment processors).

How eSewa Solves It:

1. Scaling Servers Dynamically:
   • Uses cloud servers to add capacity during peak hours (e.g., 10x more servers on Dashain Day 1).
2. Prioritization:
   • High-priority queues: Government transactions (e.g., license payments).
   • Low-priority queues: Non-urgent payments (e.g., bill splits).
3. Load Balancing:
   • Distributes transactions across servers to avoid bottlenecks.

Visual: eSewa’s Payment Processing Queue

flowchart TD
    A["Users (λ = 10,000 transactions/min)"] --> B["Load Balancer"]
    B --> C["Server 1 (μ = 2,000/min)"]
    B --> D["Server 2 (μ = 2,000/min)"]
    B --> E["Server 3 (μ = 2,000/min)"]
    C --> F["Process Payment"]
    D --> F
    E --> F
    F --> G["Complete/Error"]
    H["Solution: Add Servers"] -->|"Increases μ"| B

Exam Link:

"eSewa’s server can handle 3,000 transactions/min (μ). During Dashain, arrivals (λ) spike to 5,000/min. Calculate Lq and suggest fixes." Answer:

• Lq (average queue length): $L_q=\frac{{\lambda }^2}{\mu (\mu -\lambda )}=\frac{5000^2}{3000(3000-5000)}=-111.11$ → System is unstable (λ > μ).
• Fixes:
  1. Add servers: Increase μ to 6,000/min (2 servers).
  2. Prioritize transactions: Reduce λ for non-urgent payments.
  3. Use caching: Store frequent transactions to reduce load.

Case 5: Himalayan Java’s Inventory Management (Just-in-Time)

Company: Himalayan Java (Nepal’s largest coffee chain) Challenge: Wasted coffee beans due to spoilage (beans last ~30 days). Theory Applied: Just-in-Time (JIT) Inventory + Economic Order Quantity (EOQ).

How Himalayan Java Solves It:

1. JIT Ordering:
   • Orders beans only when stocks hit reorder point (no bulk storage).
   • Uses daily sales data to predict demand (e.g., 500 kg/day in Kathmandu).
2. Supplier Partnerships:
   • Works with local farms for same-day delivery to avoid spoilage.
3. Safety Stock:
   • Keeps 2-day buffer for unexpected demand spikes (e.g., festivals).

Visual: Himalayan Java’s Inventory Flow

flowchart TD
    A["Daily Sales: 500 kg"] --> B["Check Inventory"]
    B -->|"< 1,000 kg"| C["Order 500 kg from Farm"]
    C --> D["Delivery in 24 hrs"]
    D --> E["Update Inventory"]
    F["Festival Demand: +20%"] --> G["Increase Safety Stock"]

Exam Link:

"Himalayan Java sells 500 kg coffee/day. Ordering cost = Rs. 1,000/order, holding cost = Rs. 50/kg/year. Calculate EOQ." Answer:

• EOQ Formula: $EOQ=\sqrt{\frac{2DS}{H}}$ Where:
  • $D=500\times 365=182,500$ kg/year,
  • $S=1,000$ (ordering cost),
  • $H=50$ (holding cost).
• Calculation: $EOQ=\sqrt{\frac{2\times 182,500\times 1,000}{50}}=\sqrt{7,300,000}\approx 2,700$ kg.
• Recommendation: Order 2,700 kg every 14 days (since $2,700/500=5.4$ days).

3. Quality Management in Cases: Toyota vs. Chaudhary Group

Case 6: Toyota’s Six Sigma (Defect Reduction)

Company: Toyota (Global Automotive) Challenge: Defects in car assembly leading to recalls. Theory Applied: Six Sigma (DMAIC):

1. Define: Reduce defects to <3.4 per million.
2. Measure: Track defect rates per car.
3. Analyze: Use fishbone diagrams to find root causes (e.g., poor training).
4. Improve: Implement Poka-Yoke (error-proofing, e.g., colored bolts).
5. Control: Monitor with statistical process control (SPC) charts.

Visual: Toyota’s DMAIC Process

mindmap
  root((Six Sigma: DMAIC))
    Define["Goal: <3.4 defects/million"]
    Measure["Track: Defects per 100 cars"]
    Analyze["Root Cause: Poor Training"]
    Improve["Poka-Yoke: Colored Bolts"]
    Control["SPC Charts: Monitor Trends"]

Exam Link:

"Toyota’s assembly line has 5 defects per 1,000 cars. Calculate Z-score and suggest Six Sigma steps." Answer:

• Z-score: $Z=\frac{5-0}{3.4}=1.47$ (not yet Six Sigma compliant).
• Steps:
  1. Measure: Use control charts to track defects.
  2. Analyze: Use 5 Whys to find root cause (e.g., "Why are bolts loose?" → "Workers not trained").
  3. Improve: Train workers + use Poka-Yoke (e.g., bolts that only fit one way).

Case 7: Chaudhary Group’s ISO Certification (Quality Standards)

Company: Chaudhary Group (Nepal’s largest conglomerate) Challenge: Inconsistent quality across factories (e.g., cement, textiles). Theory Applied: ISO 9001:2015 (Quality Management Systems). How Chaudhary Group Solves It:

1. Document Processes:
   • Creates Standard Operating Procedures (SOPs) for every product (e.g., cement mixing ratios).
2. Audit Regularly:
   • Internal audits every 6 months to check compliance.
3. Customer Feedback:
   • Uses Net Promoter Score (NPS) to measure satisfaction.
4. Continuous Improvement:
   • Kaizen: Small, incremental improvements (e.g., reducing cement curing time).

Visual: Chaudhary Group’s ISO Compliance Flow

flowchart TD
    A["ISO 9001 Requirements"] --> B["Document SOPs"]
    B --> C["Train Employees"]
    C --> D["Internal Audit"]
    D -->|"Non-Compliance"| E["Corrective Action"]
    D -->|"Compliant"| F["Customer Feedback"]
    F --> G["Kaizen: Improve"]

Exam Link:

"Chaudhary Group’s cement plant has 10% defects. How would you use ISO 9001 to reduce this?" Answer:

1. Define SOP: Standardize mixing/curing process.
2. Train Workers: Reduce human error.
3. Control Charts: Monitor defect rates daily.
4. Kaizen: Adjust mixing time to reduce cracks.

4. Transportation and Assignment Problems in Cases

Case 8: NTC’s Fiber-Optic Cable Routing (Assignment Problem)

Company: Nepal Telecom (NTC) Challenge: Minimizing cable costs while connecting cities (Kathmandu, Pokhara, Biratnagar). Theory Applied: Hungarian Algorithm (for assignment problems).

Example Problem: NTC needs to connect 3 cities (A, B, C) to 3 data centers (1, 2, 3) with the following costs (in Rs. lakhs):

Solution:

1. Subtract row minima:
   • Row A: min=3 → [2, 0, 3]
   • Row B: min=2 → [0, 2, 1]
   • Row C: min=4 → [2, 1, 0]
2. Subtract column minima:
   • Column 1: min=0 → [2, 0, 2]
   • Column 2: min=0 → [0, 2, 1]
   • Column 3: min=0 → [3, 1, 0]
3. Find optimal assignment:
   • Assign B→1, A→2, C→3 (lowest cost path).

Total Cost: $2+3+4=Rs.9$ lakhs.

Exam Link:

"NTC must connect 4 cities to 4 towers. Use the Hungarian Algorithm to find the minimal cost assignment." Answer: Follow the steps above, showing row/column reductions and optimal path.

5. Game Theory in Competitive Markets

Case 9: Pathao vs. Taxi Service (Pricing Game)

Companies: Pathao (ride-hailing) vs. Traditional Taxi Challenge: Price wars leading to losses. Theory Applied: Prisoner’s Dilemma (Nash Equilibrium).

Payoff Matrix:

Analysis:

• Nash Equilibrium: Both choose Low Price (since deviating leads to worse outcomes).
• Solution: Collusion (illegal) or differentiation (e.g., Pathao offers surge pricing during festivals).

Exam Link:

"Pathao and Taxi Service face the above payoffs. What’s the Nash Equilibrium? How can they escape it?" Answer:

• NE: Both choose Low Price.
• Escape: Pathao can offer premium services (e.g., airport transfers) to justify high prices.

6. Statistical Process Control (SPC) in Manufacturing

Case 10: Himalayan Batteries’ Defect Control

Company: Himalayan Batteries (Nepal) Challenge: 10% defective batteries due to inconsistent production. Theory Applied: Control Charts (X-bar and R-charts).

Steps:

1. Collect Data: Sample 5 batteries/hour, record voltage (target: 1.5V).
2. Calculate Control Limits:
   • UCL = $\bar{X}+3\times \frac{\sigma }{\sqrt{n}}$
   • LCL = $\bar{X}-3\times \frac{\sigma }{\sqrt{n}}$
3. Plot Data: If points fall outside limits, adjust machine settings.

Visual: Control Chart for Battery Voltage

flowchart TD
    A["Sample 5 Batteries"] --> B["Measure Voltage"]
    B --> C["Calculate Mean (X-bar)"]
    C --> D["Plot on Control Chart"]
    D -->|"Point Outside Limits"| E["Stop Production & Adjust"]
    D -->|"Points Within Limits"| F["Continue"]

Exam Link:

"Himalayan Batteries’ voltage data shows 3 points above UCL. What does this indicate?" Answer:

• Out of Control: Process is not stable (likely machine malfunction).
• Action: Recalibrate voltage regulators.

In the Real World

Exam Tip

How to Score Full Marks in Case Study Questions

1. Structure Your Answer Like This:

   [Problem Identification]
   → [Relevant Theory]
   → [Data Analysis/Calculations]
   → [Solutions with Justification]
   → [Implementation Steps]
   

2. Common Exam Questions & How to Answer:

   Question Type	Your Approach
   "Analyze the case and suggest improvements."	Use SWOT/PESTEL → Identify gaps → Apply queuing, EOQ, or Six Sigma.
   "Calculate wait times using queuing theory."	Use M/M/1 or M/M/c formulas and show step-by-step math.
   "How would you reduce costs in this scenario?"	Apply transportation models or linear programming.
   "What quality tools would you use?"	Mention Fishbone, Control Charts, or DMAIC with examples.

3. Avoid These Mistakes:

   • ❌ No calculations: Always show formulas and numbers (even if approximate).
   • ❌ Vague solutions: Say "train employees" → Instead, say "Implement Six Sigma training for 3 months."
   • ❌ Ignoring data: If the case gives wait times = 20 mins, your solution must reduce this number.

4. Real-World Tie-In:

   • Always relate your answer to a Nepali company (e.g., "Like Ncell, this company should use dynamic scaling during peak hours.").

Final Worked Example: NTC’s Network Upgrade

Case: NTC wants to reduce call drops in rural areas. Your Answer:

1. Problem: High λ (calls/min) vs. low μ (server capacity) in rural towers.
2. Theory: M/M/1 Queuing Model.
3. Data:
   • λ = 150 calls/min, μ = 100 calls/min.
   • $L_q=\frac{{\lambda }^2}{\mu (\mu -\lambda )}=\frac{150^2}{100(100-150)}=-56.25$ → Unstable.
4. Solution:
   • Short-term: Add 1 more server (μ = 200) → $L_q=11.25$ calls in queue.
   • Long-term: Use smaller, distributed towers (like 5G microcells).
5. Implementation:
   • Partner with private telecom firms for shared towers.
   • Use predictive analytics to place towers where λ is highest.

Why This Works:

• Shows mathematical reasoning (queuing formulas).
• Links to real NTC strategies (5G, partnerships).
• Provides actionable steps.