Data Science for Value Chain Management (2024)

Value Chain Management (VCM) refers to the process of organizing activities that add value to the goods or services to achieve a competitive advantage in the marketplace.

This method helps organizations to effectively respond to market trends and improve efficiency to boost profitability.

As a data sciencist and analytics manager, what impact can you have on your company's value chain?

In this article, we quickly delve into the fundamental components of Value Chain Management.

We will then explore 4 examples of data science applications to support strategic primary activities.

Summary
I. The Pillars of Value Chain Management
 1. Activities to create value
Understanding the fundamental components to create value 
 2. What are the primary activities?
Core functions directly involved in product creation, marketing, and delivery.
 3. What are the support activities?
Essential functions that indirectly contribute to value creation
 4. Data Science to Support Primary Activities
Discussing how data science tools and techniques can be employed to 
optimize primary activities and enhance overall value chain management.
II. Inbound Logistics: Supply of Raw Materials
 1. Supplier Mapping with the Graph Theory
Analyze supplier networks for risk assessment and optimization.
 2. Sustainable Sourcing Network Optimization
Select suppliers based on economic and sustainable criteria
III. Operations: From Raw Materials to Finished Goods
 1. Production Planning Optimization with Wagner-Whitin Algorithm
Optimize production planning, balancing setup costs and inventory management.
 2. How to measure the impact of your solution?
IV. Outbound Logistics: Distribute your Final Products
 1. Automate a Supply Chain Control Tower
Automate monitoring and improve the efficiency of outbound distribution.
 2. How can we improve the performance using these diagnostics?
V. Conclusion

Companies constantly search for methods to gain a competitive edge by enhancing efficiency and maximising profitability.

Customer: Samir, we would like to reduce the logistic costs by 20%. We want you to redesign the whole distribution network.

This is the commonality of most of the projects I’ve conducted as a Supply Chain Engineer or Data Scientist.

Therefore, most of the articles published in this blog focus on using data analytics to optimize supply chain processes with the ultimate objective of reducing costs.

Fashion Retailer: How can we produce and deliver our products at a lowest cost?

Value Chain Management (VCM) is a strategic approach that aims to streamline every stage of the business process to optimize performance.

This covers everything from production to delivery to create maximum customer value while minimizing costs.

Activities to create value

The value chain framework was originally introduced in Michael Porter's book “Competitive Advantage: Creating and Sustaining Superior Performance”.

This revolutionized how businesses perceive their operations by dissecting any business into a series of interconnected activities contributing to the creation and delivery of value to customers.

• The Primary activities are directly involved in creating, selling, maintaining and supporting a product or service.
• The Support activities include infrastructure, technological development, human resources management, and procurement.

Let’s explore their definitions now using the example of a Fashion retailer producing T-shirts in Asia that are sold in Europe.

What are the primary activities?

Primary activities include inbound logistics, operations, outbound logistics, marketing and service.

These are the core functions directly involved in the creation, production, marketing and delivery of a product or service to the end customer.

For our example of the value chain of a T-shirt:

• Inbound logistics includes sourcing cotton from suppliers and transportation to the manufacturing facilities.
• Operations include transforming cotton into fabric patterns, which are then sewn together to create a T-shirt.
• Outbound Logistics involves all the logistic processes to package, store and deliver the t-shirts to their final destination.
• Marketing and Sales focus on promoting to generate sales.
• Service includes after-sales support, customer service, and additional services such as customization.

How do you support and orchestrate these activities?

In addition, support activities play a critical role in ensuring the smooth functioning of these primary activities.

Let’s delve into these essential support functions underpinning the entire value chain.

What are the support activities?

The support activities indirectly contribute to the value of a product or service by enhancing the efficiency and effectiveness of primary activities.

Support activities encompass infrastructure, technological development, human resources management and procurement.

As a data scientist in the digital transformation department, you are part of the support functions.

• Firm Infrastructure includes the organizational structure, control systems and administrative tasks that enable the smooth operation of its value chain.
• Technological Development activities relate to the technology and systems that support the value chain.
• Human Resource Management involves recruiting, training and retaining employees who contribute to every value chain stage.
• Procurement involves sourcing and purchasing the inputs needed for the value chain, from raw materials to office equipment.

What impacts can you have as a key actor in the Technological Development activites?

Data Science to Support Primary Activities

As a newly hired data science manager, you would like to propose a roadmap for implementing advanced analytics tools to support Primary Activities.

The objective is to support carefully selected value chain activities and make your team a strategic asset for the company.

This includes all the processes and activities associated with receiving, storing and distributing the inputs internally before production.

In our example, these inputs could be raw materials like cotton, dye and other supplies needed to manufacture a T-shirt.

See Also

What are the 5 V's of Big Data? | Definition & Explanation

💡 Based on your past experience, you propose a set of analytics tools to answer operational challenges and optimize processes.

How can we use descriptive analytics to monitor the sourcing performance?

Supplier Mapping with the Graph Theory

You would like to propose solutions to support suppliers' risk assessment.

💾 Input data: purchase orders, supplier information, factory/warehouse capacity and shipment records in Excel files.

The procurement team needs a tool that provides visibility across the supplier network to perform risk assessments, supplier consolidation and logistic network design.

❔ Problem Statement: You have a set of factories that receive critical components and raw materials from suppliers.
-
How can you estimate the impact of the failure of a specific supplier on your overall manufacturing footprint?

How many suppliers are critical for the factory B-45 ?

🚀 Solution: Graph theory is a mathematical field that studies the relationships between objects represented as vertices and edges in graphs.

In this specific case, the graph theory can be used to

• Visualize all the suppliers involved in the value chain of a specific item
• Visualize all the factories involved in the same value chain: raw material processing plants and assembly parts.

I used this theory to analyze the routing strategies for a retail company.

The objective was to visualize all the stores that were delivered on the same route.

This example can be easily adapted to audit the supplier networks

• If a supplier is delivering the factory B-45, create a link
• Create a link between the raw material processing factory B-45 and the assembly line C-78.

For more information about the Graph Theory,

What kind of analysis can support the Graph Theory?

• Risk Assessment: how many items depend on a specific supplier?
  🎯 Visually estimate the importance of this supplier and assess its risk.
• Network Design: Where are the suppliers for a specific factory located?🚛 To reduce sourcing lead times and dependence on sea freight, you can push the suppliers to relocate their plants close to this factory.

This visualization provides enough visibility to the sourcing teams and can support discussions around risk assessment or supplier rationalization.

Sustainable Sourcing Network Optimization

The sourcing teams requested a tool to select the right suppliers to reduce the environmental footprint of the inbound flows.

💾 Input data: factories demand in (units/month), supplier and factory locations, the environmental impact of each supplier (CO2, water, …) and the production cost of each supplier.

Your colleagues need a tool that selects the right suppliers based on

• Constraints like supply ≥ demand, limited capacities of the suppliers and maximum emissions or water usage per unit produced.
• Specific objectives: minimize cost, water usage or CO2 emissions.

❔ Problem Statement: Which suppliers should we select to minimize CO2 emissions?

🚀 Solution: Using linear programming with Python, you can automatically select the right suppliers based on the objective selected by the user.

In the application I deployed, users can easily simulate several scenarios based on economic or sustainable criteria to facilitate decision-making.

This article provides a detailed presentation of the theory behind this tool.

What is the impact of your solutions?

Deploying these analytics solutions can help your organization sustainably secure the supply of raw materials at the lowest costs.

We can move now to the transformation of these raw materials with the second activity.

This is the stage where the actual production of the T-shirt occurs by transforming inputs from inbound logistics into finished products.

This activity includes processes like fabric cutting, sewing, dyeing and printing.

💡 To absorb the increasing price of raw materials, the manufacturing department requested your support to reduce production costs.

How can we use prescriptive analytics to optimize production processes?

Production Planning Optimization with Wagner-Whitin Algorithm

You would like to propose a solution to optimize production planning.

💾 Input data: Customers send purchase orders with specific quantities to be delivered at a certain time.

In the example above, the customers shared the quantities to be delivered for the next 12 months.

❔ Problem Statement: How do you organize the production batches to minimize the total cost of production per unit?

For this exercise, you need to find the balance between

• Setup Costs: fixed costs you have each time you set up a production line
  If you produce only the quantity requested per month, your inventory will be low, but setup costs will explode.
• Holding Costs: cost of storage per unit per time
  If you produce the total quantity in the first month, your setup costs will be minimized, but you will accumulate too much inventory.

🚀 Solution: Production planning minimizes the total cost of production by finding a balance between minimizing inventory and maximizing the quantity produced per setup.

Wagner and Whitin developed an algorithm that finds the optimal planning by dynamic programming that balances setup and inventory costs.

If you want to know more about the theory, this article provides a detailed presentation.

How to measure the impact of your solution?

I would try with a prototype supporting a specific factory in its planning

1. Start by assessing the current production planning: inventory, number of setups for a specific period and costs.
2. Run the tool for the exact same period.
3. Validate the results with the production planners and calculate the potential savings.

You can try now with this prototype that I deployed a few months ago

We can now move to the outbound logistics to deliver these finished goods to customers.

In distribution centres storing finished products, logistic operational teams manage processes to prepare and deliver orders to the customer.

For our fashion retailer, this includes storage, order fulfilment, transportation and store delivery across Europe.

💡 Because store managers complain about delivery lead times, the operation director requested your support to improve the distribution process.

How can we can automatically monitor the distribution network with diagnostic analytics?

Automate a Supply Chain Control Tower

You are in contact with the Distribution Planning Manager; her team monitors all the orders for store replenishments.

For each item sold in stores, the inventory level (in units) is monitored in the ERP by distribution planners.

When the inventory level reaches the minimum level set by the planners

1. The system automatically creates replenishment order with item quantities and requested delivery dates
2. Warehouse operational teams prepare the orders for shipment
3. Transportation teams organize the delivery to the stores

Which metrics can you use to monitor this complex process?

The different systems involved in the distribution chain record timestamps at each key step.

From the order creation to the store delivery

• Timestamps are recorded when the process is finished ;
• Expected times of completion are calculated based on the service level agreements ;

To support her root cause analysis, she would like you to implement a system to automatically flag delays in the intermediate steps.

For instance, the example at the bottom missed the “Shipping Time” target, which delayed the delivery.

With these rules, you can automatically create late delivery reason codes to support diagnostics.

For more details on how to implement this solution,

How can we improve the performance using these diagnostics?

These insights can be used by the Distribution Planning Manager to push operational teams:

• Reporting the number of delays by week with reason codes;
• Challenge the operational teams with root cause analyses supported by the reason code mapping;
• Measure the impact for the stores with an overall KPI measuring the percentage of orders delivered On Time In Full (OTIF);

As a data analytics manager, you contributed to the lead time reduction by providing enough insights to support continuous improvement initiatives.

Value Chain Management is an approach that enables businesses to gain a deeper understanding of their operations, streamline processes and create superior value for their customers.

As a Data Analytics Manager, you have a key role to play.

Integrating data science into value chain management offers businesses opportunities to enhance operational efficiency, reduce costs, and ultimately drive profitability.

By exploring examples ranging from inbound logistics to outbound distribution, we’ve illustrated the transformative impact of data science on strategic processes.

This can become an enabler for any major transformation impacting the entire value chain.

As organizations continue to embrace digital transformation, you have the tools to showcase your team's potential to become a strategic asset of the company.

Let’s connect on Linkedin and Twitter. I am a Supply Chain Engineer using data analytics to improve logistics operations and reduce costs.

If you are interested in data analytics and supply chain, please visit my website.

• “Competitive Advantage: Creating and Sustaining Superior Performance”, Michael Porter