Supplier Selection

Supplier Selection Framework

Selection Process Stages

1. Need Identification

• Define requirements and specifications
• Estimate demand volumes
• Determine sourcing strategy
• Build business case

2. Market Research

• Identify potential suppliers
• Conduct market analysis
• Assess supply market dynamics
• Benchmark pricing and terms

3. Supplier Pre-Qualification

• Screen for basic requirements
• Verify financial stability
• Check certifications and compliance
• Assess capacity and capability

4. RFx Development & Issuance

• Create RFP/RFQ/RFI documents
• Define evaluation criteria
• Issue to qualified suppliers
• Manage Q&A process

5. Proposal Evaluation

• Score supplier responses
• Conduct technical evaluations
• Perform cost analysis
• Site visits and audits

6. Negotiation

• Discuss terms and conditions
• Negotiate pricing and volumes
• Finalize service levels
• Address contingencies

7. Final Selection

• Make recommendation
• Obtain approvals
• Award contract
• Transition planning

Supplier Evaluation Methods

Weighted Scoring Model

Most Common Approach:

• Define evaluation criteria
• Assign weights based on importance
• Score each supplier on each criterion
• Calculate weighted total score

import pandas as pd
import numpy as np

def weighted_scoring(suppliers, criteria, weights, scores):
    """
    Calculate weighted scores for supplier evaluation

    Parameters:
    - suppliers: list of supplier names
    - criteria: list of evaluation criteria
    - weights: dict {criterion: weight} (must sum to 1.0)
    - scores: dict {(supplier, criterion): score} (0-10 scale)

    Returns:
    - DataFrame with scores and rankings
    """

    # Validate weights sum to 1.0
    total_weight = sum(weights.values())
    if not np.isclose(total_weight, 1.0):
        print(f"Warning: Weights sum to {total_weight}, normalizing...")
        weights = {k: v/total_weight for k, v in weights.items()}

    results = []

    for supplier in suppliers:
        weighted_score = 0
        criterion_scores = {}

        for criterion in criteria:
            score = scores.get((supplier, criterion), 0)
            weight = weights.get(criterion, 0)
            weighted_value = score * weight

            criterion_scores[criterion] = score
            weighted_score += weighted_value

        results.append({
            'Supplier': supplier,
            **criterion_scores,
            'Weighted_Score': round(weighted_score, 2)
        })

    # Create DataFrame and rank
    df = pd.DataFrame(results)
    df['Rank'] = df['Weighted_Score'].rank(ascending=False, method='min')
    df = df.sort_values('Weighted_Score', ascending=False)

    return df


# Example usage
suppliers = ['Supplier_A', 'Supplier_B', 'Supplier_C']

criteria = ['Price', 'Quality', 'Delivery', 'Service', 'Innovation']

weights = {
    'Price': 0.30,
    'Quality': 0.25,
    'Delivery': 0.20,
    'Service': 0.15,
    'Innovation': 0.10
}

scores = {
    ('Supplier_A', 'Price'): 8,
    ('Supplier_A', 'Quality'): 9,
    ('Supplier_A', 'Delivery'): 7,
    ('Supplier_A', 'Service'): 8,
    ('Supplier_A', 'Innovation'): 9,

    ('Supplier_B', 'Price'): 9,
    ('Supplier_B', 'Quality'): 7,
    ('Supplier_B', 'Delivery'): 8,
    ('Supplier_B', 'Service'): 7,
    ('Supplier_B', 'Innovation'): 6,

    ('Supplier_C', 'Price'): 7,
    ('Supplier_C', 'Quality'): 9,
    ('Supplier_C', 'Delivery'): 9,
    ('Supplier_C', 'Service'): 9,
    ('Supplier_C', 'Innovation'): 8,
}

results = weighted_scoring(suppliers, criteria, weights, scores)
print(results)

Total Cost of Ownership (TCO) Analysis

Beyond Price:

• Purchase price
• Transportation and logistics
• Quality costs (defects, returns)
• Inventory carrying costs
• Administrative costs
• Risk and disruption costs

class TCOCalculator:
    """Total Cost of Ownership calculator for supplier comparison"""

    def __init__(self, supplier_name):
        self.supplier_name = supplier_name
        self.costs = {}

    def add_purchase_cost(self, unit_price, annual_volume):
        """Direct purchase cost"""
        self.costs['purchase'] = unit_price * annual_volume
        return self

    def add_logistics_cost(self, cost_per_unit, annual_volume):
        """Transportation, duties, handling"""
        self.costs['logistics'] = cost_per_unit * annual_volume
        return self

    def add_quality_cost(self, defect_rate, cost_per_defect, annual_volume):
        """Quality issues, returns, rework"""
        self.costs['quality'] = defect_rate * cost_per_defect * annual_volume
        return self

    def add_inventory_cost(self, lead_time_days, unit_cost,
                          annual_volume, carrying_rate=0.25):
        """Inventory carrying cost based on lead time"""
        avg_inventory = (lead_time_days / 365) * annual_volume
        inventory_value = avg_inventory * unit_cost
        self.costs['inventory'] = inventory_value * carrying_rate
        return self

    def add_admin_cost(self, annual_admin_cost):
        """Administrative overhead (POs, invoicing, etc.)"""
        self.costs['admin'] = annual_admin_cost
        return self

    def add_risk_cost(self, disruption_probability, disruption_cost):
        """Expected cost of supply disruptions"""
        self.costs['risk'] = disruption_probability * disruption_cost
        return self

    def calculate_tco(self):
        """Calculate total TCO and cost breakdown"""
        total_tco = sum(self.costs.values())

        return {
            'supplier': self.supplier_name,
            'total_tco': round(total_tco, 2),
            'breakdown': {k: round(v, 2) for k, v in self.costs.items()},
            'breakdown_pct': {
                k: round(v/total_tco*100, 1)
                for k, v in self.costs.items()
            }
        }


# Example: Compare two suppliers
annual_volume = 100000  # units

# Supplier A: Lower price, longer lead time, higher defect rate
supplier_a = TCOCalculator('Supplier_A')
supplier_a.add_purchase_cost(unit_price=10.00, annual_volume=annual_volume)
supplier_a.add_logistics_cost(cost_per_unit=1.50, annual_volume=annual_volume)
supplier_a.add_quality_cost(defect_rate=0.02, cost_per_defect=50, annual_volume=annual_volume)
supplier_a.add_inventory_cost(lead_time_days=45, unit_cost=10.00, annual_volume=annual_volume)
supplier_a.add_admin_cost(annual_admin_cost=25000)
supplier_a.add_risk_cost(disruption_probability=0.10, disruption_cost=200000)

tco_a = supplier_a.calculate_tco()

# Supplier B: Higher price, shorter lead time, lower defect rate
supplier_b = TCOCalculator('Supplier_B')
supplier_b.add_purchase_cost(unit_price=10.50, annual_volume=annual_volume)
supplier_b.add_logistics_cost(cost_per_unit=1.00, annual_volume=annual_volume)
supplier_b.add_quality_cost(defect_rate=0.005, cost_per_defect=50, annual_volume=annual_volume)
supplier_b.add_inventory_cost(lead_time_days=21, unit_cost=10.50, annual_volume=annual_volume)
supplier_b.add_admin_cost(annual_admin_cost=20000)
supplier_b.add_risk_cost(disruption_probability=0.03, disruption_cost=200000)

tco_b = supplier_b.calculate_tco()

# Compare
print(f"\n{tco_a['supplier']} TCO: ${tco_a['total_tco']:,.0f}")
print(f"{tco_b['supplier']} TCO: ${tco_b['total_tco']:,.0f}")
print(f"\nDifference: ${abs(tco_a['total_tco'] - tco_b['total_tco']):,.0f}")

Analytical Hierarchy Process (AHP)

For Complex Decisions:

• Pairwise comparison of criteria
• Consistency checking
• Hierarchical decision structure
• Handles both quantitative and qualitative factors

import numpy as np
from numpy.linalg import eig

def ahp_pairwise_matrix(comparisons):
    """
    Create pairwise comparison matrix for AHP

    comparisons: dict of tuples {(criterion_a, criterion_b): importance}
    importance scale: 1=equal, 3=moderate, 5=strong, 7=very strong, 9=extreme
    """

    # Extract unique criteria
    criteria = set()
    for (a, b) in comparisons.keys():
        criteria.add(a)
        criteria.add(b)
    criteria = sorted(list(criteria))
    n = len(criteria)

    # Build matrix
    matrix = np.ones((n, n))

    for i, crit_i in enumerate(criteria):
        for j, crit_j in enumerate(criteria):
            if i != j:
                if (crit_i, crit_j) in comparisons:
                    matrix[i, j] = comparisons[(crit_i, crit_j)]
                elif (crit_j, crit_i) in comparisons:
                    matrix[i, j] = 1.0 / comparisons[(crit_j, crit_i)]

    return matrix, criteria


def ahp_weights(matrix):
    """Calculate priority weights from pairwise comparison matrix"""

    # Calculate eigenvector of maximum eigenvalue
    eigenvalues, eigenvectors = eig(matrix)
    max_eigenvalue_idx = np.argmax(eigenvalues.real)
    principal_eigenvector = eigenvectors[:, max_eigenvalue_idx].real

    # Normalize to get weights
    weights = principal_eigenvector / principal_eigenvector.sum()

    # Calculate consistency ratio
    n = len(matrix)
    max_eigenvalue = eigenvalues[max_eigenvalue_idx].real
    ci = (max_eigenvalue - n) / (n - 1)

    # Random index values
    ri_values = {1: 0, 2: 0, 3: 0.58, 4: 0.90, 5: 1.12,
                 6: 1.24, 7: 1.32, 8: 1.41, 9: 1.45, 10: 1.49}
    ri = ri_values.get(n, 1.49)

    cr = ci / ri if ri > 0 else 0

    return weights, cr


# Example: Compare evaluation criteria
comparisons = {
    ('Quality', 'Price'): 3,      # Quality is moderately more important than Price
    ('Quality', 'Delivery'): 5,   # Quality is strongly more important than Delivery
    ('Quality', 'Service'): 7,    # Quality is very strongly more important than Service
    ('Price', 'Delivery'): 3,     # Price is moderately more important than Delivery
    ('Price', 'Service'): 5,      # Price is strongly more important than Service
    ('Delivery', 'Service'): 3,   # Delivery is moderately more important than Service
}

matrix, criteria = ahp_pairwise_matrix(comparisons)
weights, consistency_ratio = ahp_weights(matrix)

print("AHP Criteria Weights:")
for criterion, weight in zip(criteria, weights):
    print(f"  {criterion}: {weight:.3f} ({weight*100:.1f}%)")

print(f"\nConsistency Ratio: {consistency_ratio:.3f}")
if consistency_ratio < 0.10:
    print("  ✓ Acceptable consistency (CR < 0.10)")