Factory Digitalization Ver05. - What I Learned About “Three Essentials” in Manufacturing Digitalization

Shigenori Tanaka
3月14日
読了時間: 2分

Mar 14, 2026

Hello everyone,

Digital transformation in manufacturing is never easy. Production processes are complex, sources of variation are everywhere, and equipment often spans multiple generations.

As a result, many factories struggle with the same issues:

Data that isn't accurate
Systems that don't work as intended
Information that can't be used for decision-making.

Over the past six years, I have worked with sand casting plants - one of the most complex manufacturing environments - and implemented:

Real-time production visualization
Batch-level data integration
Linking production data with inspection results
AI-based defect-rate reduction models
“Prescriptions” tailored to each product pattern

Through these projects, I learned one thing very clearly:

Digital transformation delivers results only when the following three elements come together:

① Reliable Data

② Properly Designed Systems

③ Effective Daily Implementation

If any one of these is missing, even an expensive digital system will never pay off.

When companies invest in digitalization, they typically define a purpose and calculate the expected ROI - whether through increased output or improved gross margin from defect reduction. To achieve that planned ROI, all three elements must be realized.

① Reliable Data

Anyone who has worked with factory data knows this reality: Data is often inaccurate.

Input errors, missing entries, sensor drift, forgotten process records— these issues exist in every factory.

However, visualization allows you to finally see:

Where errors occur
Which processes are drifting
Which data cannot be trusted

Visualization is the first step toward creating reliable data, and it is also essential for improving the accuracy of AI models.

② Properly Designed Systems

System development is the vendor's responsibility. But if the system is not designed based on a correct understanding of the process, data will not connect, analysis will not work, and the improvement cycle will never start.

In my projects, I built systems that connect process data with inspection results, train AI models to detect defect tendencies in advance, and generate actionable “prescriptions” for each product pattern.

This is only possible when the system is built on accurate process understanding.

③ Effective Daily Implementation

And finally—the most important element—is implementation.

No matter how advanced the system is, if managers and supervisors don't understand it, operators will not follow the prescriptions.

The shop floor will never adopt something they don't believe in.

Only when operation takes root does the cycle begin:

Data → Analysis → Prescription → Improvement

Summary

Through these six years of experience, I learned the following:

Reliable data, proper systems, and effective implementation

— only when these three come together can manufacturing digitalization deliver the results that were originally planned.

Contact