Bad Data in Manufacturing: The Cost and the Solution

- Mar 07, 2019-

“85% of industry data points go unused” and “only 20% of the actual potential asset failures are detected”

Everybody knows we’re on our way towards a world where data reigns king. In manufacturing, where Industry 4.0 unlocks immense data crunching capability, the value potential is endless. The plot twist? Only a handful of manufacturing companies are equipped to handle this new volume of manufacturing data – and recognize which data is useful, and which data is not.

In fact, according to a DataRPM report [I], “almost 85% of industry data points go unused”. This results in “only 20% of the actual potential asset failures are detected” and “a massive 80% are categorized as individual one-offs”.

The cost is real, and it is significant. Consequently, data integrity (which is the maintenance of data, and the assurance of its accuracy and consistency over its life-cycle) must become a top priority for every manufacturer.

This paper serves as a quick and actionable intro to manufacturing data integrity.

Commitment to Manufacturing Data Integrity Trickles Down from Leadership

As leaders, manufacturing executives ought to consider themselves as stewards of company-wide excellence. This is why they must personally invest the time required to nurture a data-driven culture. There are many channels and communication tools they can leverage to spread this message: from social media and email, to shop floor visits and training sessions. The overall objective is to identify cultural and knowledge barriers, drop them, and replace them with a sense of commitment to data integrity. Ultimately, the excellence in data concept should be integrated as one of the company’s core values.

In this effort, establishing upskilling programs to educate the workforce on the fundamentals of data collection, analysis, reporting, and retention is key. Similarly, it’s also important to establish well-thought-out data integrity systems, and documentation to ensure the health of data is (reasonably) certain. Finally, in-depth reviews should be standardized and conducted by independent personnel, as a fail-safe mechanism.

Encourage Transparency and Open Report of Problems

What differentiates a company that tries to be data-driven with a company that truly is data-driven, is the latter’s open environment for reporting problems. This insight is essential. All employees should feel completely safe in reporting errors, omissions or any other sort of problem irrespective of rank or role. There are many ways to achieve this, including: Independent, anonymous reporting mechanisms, the establishment of an independent quality group with direct ties to senior management, or even the personal recognition and reward of employees’ who find problems through a bounty system. Similarly, it’s important to make sure employees fully understand the long-term cost of bad data – which is tragically bigger than the short-term problem of reporting it.

If manufacturing companies succeed in data integrity, the potential is immense. From reducing unplanned downtime and scrap production to breaking productivity ceilings and improving equipment efficiency.

SOURCES

[I] DATARPM: ANOMALY DETECTION & PREDICTION DECODED: 6 Industries, Copious Challenges, Extraordinary Impact