share
In nearly every supply chain transformation or planning system implementation, one challenge consistently surfaces—"data quality."
It’s often cited as a top concern yet rarely examined deeply. Often, we gloss over it, label the data as "bad," and move on. But is that the right diagnosis?
Is the data genuinely incorrect—or is our modeling, understanding, or interpretation flawed? In this article, I want to challenge this broad-brush narrative and offer a more nuanced classification of data issues within the supply chain context.
Data Misunderstood, Not Incorrect
Let’s take sales data as an example. In an Indian business context—especially one that operates ethically and adheres to tax and compliance regulations—sales data is fundamental and tightly governed. It's the basis for GST filings, revenue reporting, and external audits. The chances of it being factually incorrect are quite low.
So, when someone says, "the sales data is wrong," it often reveals a misunderstanding. What they likely mean is:
The sales data does not match their expectations within the planning model or they don’t understand how to extract and map it correctly from the ERP to the planning system.
This is not a data error, it’s a modeling or transformation error. The data is correct, but it’s either being misread, mismapped, or misrepresented.
Redefining Correctness: Bill of Materials (BOM)
Now, let’s consider the Bill of Materials—a frequent topic of debate in planning implementations. In industrial manufacturing, a BOM is usually deterministic. If a product needs 6 screws and 2 panels, it's straightforward—and if it's wrong, it can be fixed. But in industries like food services, the equation changes. Can a static BOM truly capture what a chef adjusts on the fly? Not quite.
Take Paneer Butter Masala, for instance. If the tomatoes are more acidic than usual, a chef might reduce the tomato content, increase the cream, and adjust the sugar—yet the end product remains consistent. A rigid planning model would flag this as a variance, but in reality, it’s just good judgment at work.
(Incidentally, this happens to be my wife’s favorite dish—and she’s been the quiet force behind my journey in building this supply chain consulting firm. Her support, patience, and trust have been foundational. So yes, consistency matters—not just in systems, but in life.)
This example reminds us: the issue isn’t bad data—it’s about designing planning models that respect real-world variability.
Poor Data vs. Poor Modeling
What we often call poor data quality is sometimes a symptom of:
- A rigid planning model applied to a dynamic reality
- Misalignment between data structures in ERP vs. planning systems
- Inadequate data transformation logic
- Lack of master data governance for conditional rules or variants
The real challenge is to ask:
- Is the data wrong?
- Or have we failed to represent or interpret it correctly for the planning purpose?
Final Thoughts: From Blame to Better Design
It's easy to label data as bad. It’s much harder—but far more valuable—to dig into why it doesn’t work in your model. Let’s move away from generic terms like "bad data" and toward more specific diagnostics:
- Is it structurally incompatible?
- Is the logic missing or outdated?
- Is the variability unmodeled?
- Is the data misunderstood?
As supply chain professionals, we owe it to ourselves and our stakeholders to move the conversation forward—from blaming data to designing better models