The AI revolution in supply chains is being hobbled by a fundamental issue: poor data quality. The old adage “Garbage-In, Garbage-Out” (GIGO) rings true here – flawed data leads to flawed insights. Just as a house can’t be built on shaky ground, AI can’t deliver value with bad data. The problem is, many supply chain organizations still prioritize software over data, treating data as a byproduct, not a strategic asset. This application-centric approach is holding back AI efforts. To unlock AI’s true potential, we need to shift to a data-centric approach where data is unified and integrated across our systems. Without a doubt, a data-centric supply chain is the key to driving real AI innovation in our organizations.

In this article, I’ll share my insights on how to make this data-centric shift happen. First, I’ll explain why supply chains have an unique challenge overcoming the chaotic nature of its data spread over large geographic areas. Also, I’ll look at why current AI solutions are having limited success in our supply chains. Lastly, I’ll show you how supply chains can adopt a data-centric approach to drive business insights and leverage technology to fully unlock AI’s potential.

1. Supply Chains Struggle with AI Due to Chaotic Data.

Most supply chains struggle to leverage AI effectively due to the chaotic and disjointed nature of their data. This is because supply chain data is scattered across multiple systems, stakeholders, and geographies. As a result, seamless data exchange and integration is not the norm. Hence,  most supply chains consist of data silos, information chaos, and analytics that are not aligned with decision-makers, ultimately hindering the adoption of AI. To better understand the current state of supply chain data and its on-going challenges with adopting AI, I’ll detail current deficiencies with supply chain data. In particular, I’ll highlight the disjointed nature of shipping data that affects supply chain analytics.

“… most supply chains consist of data silos, information chaos, and analytics that are not aligned with decision-makers …”

a. The State of Supply Chain Data: 12 Key Deficiencies.

The bottom line – supply chain data is not yet ready to unlock AI’s full potential. This is because of its fragmented and chaotic nature. The root of the problem lies in the industry’s reliance on transactional enterprise systems like ERP, TMS, and WMS. These systems were designed to process transactions rather than support data analysis or AI. As a result, supply chain data, spread across these many systems and geographic areas, is challenging to both integrate and analyze. To better understand the extent of these data deficiencies and their impact on AI adoption, below is a list of top deficiencies with supply chain data.

 Top 12 Deficiencies of Supply Chain Data

For more detailed examples on this topic, see my article, The Data Interoperability Challenge For Supply Chains: 12 Reasons For It And Why Tech Will Never Overcome It Alone.

The Data Interoperability Challenge For Supply Chains: 12 Reasons For It And Why Tech Will Never Overcome It Alone

b. Disjointed Shipping Data: 4 Ways it Compromises AI and Analytics in Supply Chains.

In particular, supply chains face significant challenges in collecting and deriving insights from shipping data than any other types of supply chain data. This is because shipping data is scattered across many different systems, each containing partial information about a shipment. For example, order fulfillment systems may hold details about shipment contents and addresses. Then the carrier systems only provide updates on shipment status, while other systems contain breakdowns of shipping costs. Moreover in complex global shipping operations, many more systems are involved, leading to data that is even more fragmented, duplicated, and inaccurate. As a result, shipping data by its nature is disjointed, severely impairing the effectiveness of AI and analytics in supply chains. Below are the specific problems.

“… shipping data is scattered across many different systems, each containing partial information about a shipment.”

Problems Caused by Messy Shipping Data

Poor Shipping Data – Here Are The 4 Reasons Impeding High Tech Visibility And Actionable Analytics

• Shipment Visibility Limited Due to Data Silos. It is almost impossible to maintain a Single Source of Truth (SSOT) with shipping data. This results in both managers and systems reacting blindly to unplanned events because each department is oblivious to the consequences of their actions. Moreover, AI fueled by this bad data will not be of much help. 
• Fragmented, Out-of-Date Data Limits Insights. Because of bad data, planners and management, much less AI, are able to confidently move forward with effective analysis.
• Complicated Shipping Data Structures Limits Automation. Shipping data is not unified. Partial data is tied to a purchase order, or multiple tracking numbers, or invoices, and or other reference numbers. As a result, it is difficult to automate or apply AI.
• Financial and Operational Data Not Correlated to Enable Optimization Efforts. For instance, the finance department and shipping operations make independent decisions based on the data they have, resulting in unintended consequences.

For more on the current state of shipping data, see my article, Poor Shipping Data – Here Are The 4 Reasons Impeding High Tech Visibility And Actionable Analytics.

“Visibility emerges when data is connected.”

Grant Sernick

2. Why Current AI Solutions Fall Short in Supply Chains.

From my perspective, many supply chain AI solutions fall short because they are too focused on the software algorithm and not on the data. In other cases, AI solutions are just “bolted on” to existing enterprise systems as an afterthought at best. In other instances, the AI vendor has a great solution, but their customers’ data is fragmented and messy. Additionally, many supply chain organizations are overwhelmed with data, not knowing what to do with it. Hence, data problems are the primary reason why most AI solutions fall short within the supply chain industry. So, let’s look more closely at the reasons why AI solutions fall short in supply chains. 

“… supply chain AI solutions fall short because they are too focused on the software algorithm and not on the data.”

a. The Pitfalls of a Data-Driven Approach: Overload and Silos.

First, being “data-driven” is not a guarantee of success with AI and advanced analytics. While AI requires high-quality data to deliver its full potential, many businesses are merely collecting data. What they are not doing is the next step of deriving actionable insights from the data. As a result, this data-driven approach leads to data overload and the creation of digital silos. To truly benefit from AI and analytics, businesses must go beyond data collection and focus on extracting valuable insights that drive informed decision-making..

For example, a company might gather extensive customer data from multiple sources but fail to integrate and share it effectively across departments. Consequently, each department develops its own fragmented view of the customer, leading to multiple “versions of the truth.” This, in turn, results in inconsistent insights and decision-making. For more on the dangers of being data-driven and just collecting data, click here.

b. Enterprise Systems Built for Transactional Processing with “Bolt-On” AI and Data Analytics as an Afterthought.

For decades, supply chains have heavily relied on enterprise systems such as ERPs, WMS, and TMS to process transactions and automate business processes. Moreover, most software vendors have designed these enterprise systems for transactional processing, not for data analysis or AI. Despite these limitations, many supply chain professionals remain overly focused on their enterprise software thinking it can satisfy all their information needs.

The real problem is that these enterprise systems, as designed, treat data as a byproduct, making it challenging to leverage for both data analytics and AI. In an attempt to address this shortcoming, businesses have added data marts and business intelligence (BI) reporting tools, but this is more of a stop-gap solution to the problem. Indeed, the core problem remains: relevant data is locked in these transactional systems of records, limiting data access and hindering innovation. For more on this application-centric mindset, click here.

You Need To Think Data Centric To Be A Successful Business: Stop Being Data Driven, Application Centric

“… enterprise systems, as designed, treat data as a byproduct, making it challenging to leverage for both data analytics and AI.”

c. Beyond Data Integration: The Need for Standards, Compliance, Security, Lower Costs, and Clarity.

Also when it comes to data connectivity, data integration is just the first step to the successful exchange of actionable data between systems. The true value comes from ensuring that the data sent meets tech standards, complies with regulations, is secure, cost-effective, and understandable. For instance, let’s take a healthcare provider integrating patient data from various sources. In this case, it is critical that they not only transmit their data correctly but also that it adheres to HIPAA regulations, stored securely, and presented in a clear and actionable manner for clinicians. To list, below are data interoperability deficiencies that we need to address when exchanging data.

Data Interoperability Shortcomings

The Data Interoperability Challenge: It’s The Need For Tech Standards, Compliance, Security, Massive Resources, And Be Understandable

• Adhering to Tech Standards: The lack of standards frustrates data interoperability.
• Complying with Regulations: Obeying data protection laws and corporate policies.
• Securing Access to Data: The puzzling challenge to verify, authorize, and authenticate.
• Massive Tech Costs: The high costs of IT data interoperability and integration projects.
• Make the Data Sent Understood: The absence of shared business glossaries across systems and organizations to correctly interpret data.

For more on this topic, see my article, The Data Interoperability Challenge: It’s The Need For Tech Standards, Compliance, Security, Massive Resources, And Be Understandable.

d. Rethinking Data Management Priorities: Control, Producing Insights, and Making AI-Ready.

Traditionally, data management has focused on controlling data—ensuring it’s accurate, secure, and compliant. However, this approach often overlooks the need to provide a foundation for insights and AI. For example, a financial institution might have robust data controls in place but struggles to make that data accessible and usable for AI-driven risk assessment models. To list, below are the top obstacles to making data usable for analytics and AI.

Data Management Challenges in the Age of AI

• Enterprise Data Management Today Focuses More on Controlling Data, Not Making it More Valuable.
• Organizations’ Data Strategies Are Too Focused on Collecting Data Versus Making Data Understandable and Accessible.
• Enterprise Data Management Challenged to Secure Data When It is Fragmented, In Many Silos, Duplicated, and Ambiguous.

For more information on this topic, see my article, Traditional Enterprise Data Management Is Floundering To Make Business Data More Valuable, Accessible, And Secure.

Traditional Enterprise Data Management Is Floundering To Make Business Data More Valuable, Accessible, And Secure

3. Unleashing AI Success with a Data-Centric Supply Chain.

Its time for supply chains to start putting data first so that they can truly realize AI’s full potential. This starts with both leadership buying-in and championing a data-centric business strategy. Without this corporate commitment, data will continue to be fragmented across functional silos. Moreover with enterprise-wide leadership, supply chains can leverage both existing and new technologies to advance AI initiatives, using a data-centric approach. See below for how organizations can unlock AI success by becoming a data-centric supply chain.

“It is time for supply chains to start putting data first so that they can truly realize AI’s full potential.”

a. Leadership Buy-In: Championing a Data-Centric Business Strategy.

As data is the fuel for good AI solutions and at the same time this data is spread across every part of the supply chain, corporate leadership needs to make unifying data a priority. Without a doubt, senior executives need to steer their organizations away from an application-centric mindset and towards a data-centric way of thinking. This is how organizations will start thinking of data as a strategic asset, not software byproducts stuck in one of many data silos. Indeed with senior leadership involved, supply chains can simplify data management, enable seamless information flows, and foster organizational agility. 

So, how can executives transform their organization into a data-centric business? What supply chain leaders need to do is implement a data-centric business strategy to assure that their data and organization is AI-ready. For information on how to implement a Data-Centric Business Strategy, see my article, A Data-Centric Business Strategy Checklist: The Way To Energize A Digital Enterprise To Be More Agile, Bold, And Simplified.

“… start thinking of data as a strategic asset, not software byproducts stuck in one of many data silos.”

b. Unify Shipping Data for Planning, Execution, and Financials Using Transport Load IDs.

Unifying Shipping Data Using Transport Load IDs: Here Are 10 Ways It Can Unlock Analytics And Empower Logistics Tech

A potential solution for fragmented shipping data is for supply chains to start using a Transport Load ID to unify shipping data. Without a doubt, a shipper-generated Load ID can bring together all data related to the shipper’s load. This includes planning, operational, and financial shipping data across the supply chain and its many systems. 

For instance, a logistics company might use Transport Load IDs to track shipments from origin to destination, enabling real-time updates on shipment status across carriers, estimated & actual total landed costs, and delivery times. Moreover, this shipping data is linked to purchase order and other product information. Thus, supply chains with this unified data can better optimize routes, manage inventory, improve financial forecasting, and enable all stakeholders to have full visibility to all aspects of the moment of goods. 

For more information on Transport Load IDs, see my article, Unifying Shipping Data Using Transport Load IDs: Here Are 10 Ways It Can Unlock Analytics And Empower Logistics Tech.

c. Five Ways to Transform Supply Chain Technology with a Data-Centric Approach.

It is time for a change in the supply chain tech industry. As data analytics and AI advance, more experts are embracing a data-centric mindset, challenging the traditional focus on enterprise software and automation. Indeed for years, business tech solutions have prioritized complex Software-as-a-Service (SaaS) implementations and workflow automation. While these efforts have improved efficiency, they’ve also led to rigid systems and a narrow focus on automation. Without a doubt, a data-centric approach can transform supply chain technology in several ways. Here are five key areas:

A Data-Centric Approach for Supply Chain Technology

Data-Centric Business Tech: New And Better Ways To Shatter Our Obsession With Software And Automation

• Data Management. Here, businesses can focus more on making data more valuable instead of just controlling it. For instance, use metadata to tag, link, and categorize supply chain data, making it easily searchable and accessible across different departments.
• Enterprise Architecture. In this case, a business can shift its focus from application-specific data silos to a unified data approach that integrates information from various sources.
• Artificial Intelligence (AI) App Development. In this instance, AI developers can use a data-first approach to train models. For example, a tech vendor can develop a predictive maintenance model that is trained on high-quality sensor data from equipment, rather than relying on a complex model with limited data.
• Software Coding. In this case, developers can simplify codebases by focusing on data structures and flows rather than application-specific logic. For instance, a company can refactor its inventory management code to use a standardized data model, reducing software bugs and improving maintainability.
• Semantic Knowledge Graphs. This tech can enable the digital representation of relationships between different entities and concepts in the supply chain. For example, a company can build a knowledge graph that integrates data from various sources, such as supplier information, product catalogs, and logistics networks, to provide a comprehensive view of its supply chain.

For a more detailed discussion on data-centric tech, see my article, Data-Centric Business Tech: New And Better Ways To Shatter Our Obsession With Software And Automation

Final Thoughts.

As I said previously, the AI revolution in supply chains is being obstructed by a fundamental issue: poor data quality. The old adage “Garbage-In, Garbage-Out” (GIGO) rings true here – flawed data leads to flawed insights. To unlock AI’s true potential, we need to shift to a data-centric approach where data is unified and integrated across our systems. So in this article, I shared my insights on how to make this data-centric shift happen. Indeed, it is time for change as current solutions are not working well. Without a doubt, it is time for supply chains to adopt a data-centric approach that will drive business insights and leverage technology to fully unlock AI’s potential.

“… the AI revolution in supply chains is being obstructed by a fundamental issue: poor data quality. The old adage “Garbage-In, Garbage-Out” (GIGO) rings true here – flawed data leads to flawed insights.”

A Data-Centric Business Strategy Checklist: The Way To Energize A Digital Enterprise To Be More Agile, Bold, And Simplified

For more from SC Tech Insights, see the latest articles on Data Analytics, AI, and Supply Chains.

Lastly, if you are in the supply chain industry and need help to implement a data-centric strategy, please contact me to discuss next steps. I have implemented 100s of tech pilot projects and innovative solutions across the supply chain as well as all transportation modes. I specialize in proof-of-concepts (POC) for emerging technologies and data-centric software development methods. To reach me, click here to access my contact form or you can find me on LinkedIn.