Data quality for manufacturing

In late 2023 and early 2024, automotive leaders Toyota and Honda initiated recalls affecting nearly two million vehicles. The problem wasn’t a flaw in vehicle assembly but sensor hardware defects that corrupted safety-critical input data.

On December 20, 2023, Toyota recalled approximately 1.12 million vehicles globally because a short circuit in a seat weight sensor, supplied by Aisin Electronics, could cause the passenger airbag to not deploy as designed.

A few months later, on February 1, 2024, Honda recalled over 750,000 vehicles in the U.S. for a related issue with the same Aisin-supplied sensor. A manufacturing defect, traced back to a change in materials at a tier-2 supplier, could cause the passenger airbag to deploy when it should be suppressed for a child, posing a significant safety risk.

This pair of recalls highlights a critical vulnerability woven into the fabric of modern manufacturing. As supply chains become more complex and factories more automated, companies are increasingly reliant on the quality and integrity of data from thousands of third-party components.

A hardware defect at the sensor level can corrupt the entire system’s input data, leading to safety-critical failures and costly public recalls.

The solution is to build a resilient foundation of proactive data quality management.

This means shifting from simply inspecting finished products to rigorously governing the data and hardware quality throughout the multi-tier supply chain.

By treating component data as a core asset, manufacturers can identify risks early, prevent widespread failures, and build the operational trust essential for the modern era.

Key takeaways

• Small data errors can lead to massive failures, as hardware defects at the sensor level corrupt input data and trigger costly recalls.
• Focus on what matters first by identifying your Critical-to-Quality (CTQ) data points – the few variables that most impact your product – instead of trying to fix everything at once.
• Shift from a reactive to a proactive mindset by using trusted data to predict future problems instead of only analyzing past failures.
• Governance is the foundation for scaling your efforts, requiring a central platform like Collibra to ensure lasting trust and reliability in your data.

The core challenges in manufacturing data quality

For many manufacturers, the factory floor is overflowing with data from sensors, machines, and enterprise systems. Yet, turning this flood of information into a competitive advantage remains a major hurdle.

The core challenge isn’t a lack of data, but a lack of trusted, usable data. This distinction is where operational excellence is either won or lost.

From data swamps to actionable insights

The “data-rich, information-poor” paradox is a common reality. Without proper governance, a data lake quickly becomes a data swamp: a vast, disorganized repository of information that’s difficult to navigate and nearly impossible to use for reliable decision-making.

This problem is worsened by data silos. Critical information from the production line (MES), maintenance logs, and quality control systems often exists in isolation, preventing a holistic view of the operations.

Understanding what data quality is, is the first step to fixing these fundamental data quality issues.

To solve this, leading companies are breaking down silos by creating a central hub for trusted, well-documented information.

A great example is implementing a Data Marketplace, which creates a single, reliable “shop” for engineers and analysts to find and use the data they need with confidence.

The high cost of “bad data” in a manufacturer setting

In manufacturing, the principle of garbage in, garbage out has severe financial consequences. Inaccurate or incomplete data fed into planning and operational systems directly leads to waste.

Think of a CNC machine that runs for an entire shift using a slightly incorrect calibration setting due to a single data entry error. The result is an entire batch of high-value parts produced out-of-spec, all of which must be scrapped.

These small errors compound into significant and measurable losses, such as:

• Increased scrap rates from out-of-spec production.
• Costly rework and wasted labor hours.
• Unplanned downtime caused by predictable failures.
• Wasted energy and raw materials from inefficient processes.

To get a handle on the problem, you must be able to measure its impact using key data quality metrics. Without trusted data, a manufacturer is essentially flying blind, making critical decisions based on a distorted picture of reality.

Building a foundation for quality control using data

Confronted with a data swamp, the instinct can be to try and fix everything at once. This approach often leads to paralysis. A more effective strategy is to build a strong foundation by focusing systematically on what truly matters. It begins by identifying the most critical data points and then ensuring they are captured accurately and consistently.

Step 1: Identify your critical-to-quality (CTQ) data points

Not all data is created equal. The first step in any practical data quality improvement plan is to identify your Critical-to-Quality (CTQ) data points.

These are the handful of measurements that have the most direct impact on your final product’s quality and customer satisfaction.

Instead of trying to monitor thousands of variables, work with your process engineers to answer a simple question for each production line: “Which 3-5 measurements will tell us 80% of the story about this process’s health and output quality?”

Focusing on this small but vital subset of data provides immediate clarity and makes the task manageable.

This targeted approach is a core tenet of any effective data quality framework, ensuring that your initial efforts are concentrated where they can deliver the most significant value.

Step 2: Standardize data collection and implement checks

Once you’ve identified your CTQs, the next step is standardization.

You must ensure that this critical data is captured consistently across all shifts, machines, and operators.

This means using the same units (e.g., Celsius, not a mix of C and F), the same formats for timestamps, and the same definitions for status codes. Without standardization, your data is fundamentally unreliable.

Data Point	Non-Standardized (Problem)	Standardized (Solution)
Temperature	Machine 1: 200 F, Machine 2: 93 C	All Machines: 93 C
Timestamp	25-08-2025 10:30 AM	2025-08-25T10:30:00Z (ISO 8601)
Status Code	Error, Failed, Code 9	E-STOP, MAINT, OP-RUN (Defined list)

This is where you implement systematic data quality checks at the point of collection to enforce these rules automatically.

A crucial part of this is understanding your data’s journey, or its lineage.

For instance, ensuring you have a clear, automated view of how data flows from your SAP system to your analytics tools is a foundational check to guarantee data integrity from its source.

How to improve your operations with proactive analytics

With a foundation of trusted data established, you can move beyond basic monitoring and unlock the transformative power of analytics.

This means shifting your quality control mindset from being reactive to proactive. Instead of analyzing why a batch failed yesterday, you can begin to predict and prevent the failures of tomorrow.

Moving from reactive to predictive quality analytics

Traditionally, quality control has been a historical exercise. You analyze scrap parts to find the root cause of a defect after it has already occurred.

Proactive analytics, fueled by high-quality data, flips this script entirely. By analyzing real-time data streams from your CTQ points, you can identify subtle deviations from the norm that signal a future problem.

This is the essence of effective data quality assurance: creating a system that prevents errors, not just one that reports on them.

This operational foresight is often visualized on a data quality dashboard or tracked on a data quality scorecard.

For example, a slow, steady increase in a machine’s vibration, though still within its normal operating limits, can predict a bearing failure weeks in advance.

This allows you to schedule maintenance proactively, avoiding costly unplanned downtime and scrap production.

The power of an agile approach to data projects

The prospect of implementing predictive analytics can seem daunting, often invoking images of massive, multi-year digital transformation projects. However, the most successful manufacturers adopt an agile methodology.

The goal is not to build a perfect, factory-wide system from day one, but to deliver value quickly and iteratively.

Start with a single production line or even one critical machine. Define a clear, measurable goal, such as “reduce scrap on Line 3 by 10% in the next quarter by predicting overheating events.”

By focusing on a small, high-impact project, you can demonstrate a clear return on investment in a matter of months.

This “start small, scale fast” approach builds momentum, secures buy-in from stakeholders, and allows you to learn and adapt as you expand the initiative across the entire factory.

Taking control with the right tools and expertise

Building a foundation of trusted data and leveraging it for analytics requires a combination of the right strategy, tools, and expertise. While the agile, step-by-step approach is crucial for getting started, scaling these efforts across a manufacturing enterprise depends on having a robust technological backbone and, often, an experienced partner to guide the implementation.

Choosing your platform

A successful data quality initiative relies on the right technology to automate, monitor, and govern your data at scale.

Manually checking spreadsheets is not a sustainable solution for a modern factory. When you begin to research your options, it’s important to understand how to choose a data quality platform that fits your specific operational needs, from real-time monitoring on the shop floor to enterprise-level reporting.

While there are many excellent point solutions among the best data quality tools, the most effective approach is to select a platform that integrates data quality into a broader data governance strategy.

This ensures that quality isn’t just a one-time cleanup project but a continuous, embedded process that supports every aspect of your business, from production efficiency to regulatory compliance.

How Murdio and Collibra empower manufacturers

A comprehensive data quality assessment is the ideal first step to understanding your challenges, but implementing a lasting solution requires deep expertise.

A platform like Collibra provides the central governance backbone to manage your data as a critical enterprise asset. At Murdio, we specialize in implementing Collibra’s data governance solutions for complex industrial and commercial environments.

We’ve helped organizations, from international retail chains managing vast, intricate supply chains to energy giants overseeing critical operational assets, build a foundation of trust in their data.

We help translate high-level strategy into practical, real-world results on the factory floor. If you’re ready to build a foundation of trust in your data, let’s talk.

Frequently Asked Questions

What is data quality in manufacturing?

Data quality in manufacturing refers to the accuracy, completeness, consistency, and reliability of data collected from the factory floor and across the supply chain. High-quality data ensures that decisions about production, maintenance, and quality control are based on a true picture of reality, preventing costly errors.

Why is data quality so important for modern factories?

Modern factories, often called “smart factories,” rely on data to automate processes, predict machine failures, and optimize production. If the input data is flawed (“garbage in”), the outputs and decisions will also be flawed (“garbage out”), leading to increased scrap, unplanned downtime, and even safety-critical failures like product recalls.

What’s the first step to improving data quality?

The best first step is to identify your Critical-to-Quality (CTQ) data points. Instead of trying to fix all your data at once, focus on the 3-5 key measurements per production line that have the biggest impact on your final product’s quality. This makes the task manageable and delivers the fastest return on effort.

How does a Data Governance platform like Collibra help?

A platform like Collibra acts as a central “system of record” for your data. It helps by:

• Defining what good data looks like.
• Tracking data lineage (where it comes from and how it’s used).
• Making trusted data easy for engineers and analysts to find. It turns data quality from a one-time project into a continuous, managed process.