Critical to the success of any data analytics initiative is focusing on data quality with the goal of having clean data. This means the data must be accurately labeled, free of duplicate records, and blended to generate the correct results. Cleaning up the business’s data quality issues is critical to becoming a truly data-driven enterprise and building a data-driven culture.
Clean data is especially important when manufacturing processes are controlled, monitored, and measured by automation. We are in the Fourth Industrial Revolution. Data exchange technology and the rise of automated systems in manufacturing industries are spreading to factory floors in ways that parallel the growth of information technology.
The acronym that describes this process is—IioT, the Industrial Internet of things. Technopedia describes the IioT as “a term for all of the various sets of hardware pieces that work together through connectivity to help enhance manufacturing and industrial processes.
Since the emergence of the Fourth Industrial Revolution, significant advances in machine learning, artificial intelligence, and the internet of things have been pushed and pursued in the industry sector. However, many manufacturers still need to work on incorporating advanced analytics into their business and manufacturing processes.
This article will address the causes and costs of poor-quality data—and how to make your data strategy part of your organization’s continuous improvement program.
The rapidly increasing volume and variety of data from business applications, sensors, third-party sources, and e-commerce transactions add to the challenge. Historically, disparate data from multiple applications and data sources have created data quality issues and, ultimately, bad data. To compensate, data analysts and scientists must apply data cleansing to the data before incorporating it into their analytics dashboards and models.
And most importantly, leaders are left with bad data that drives poor decision-making, with the effects summarized in a 2021 Experian Global Management Research report. According to Experian, the following highest to lowest impacts result from poor-quality data:
In short, data quality management is needed. The problem is highlighted by Gartner’s findings that about 20% of data circulating in the ethernet is bad data.
According to Oracle, data management is “the practice of collecting, keeping, and using data securely, efficiently, and cost-effectively.” When done within the bounds of the organization’s data governance policy and controlling regulations, effective data management results in the best business decisions and processes (in manufacturing, for example) that benefit the organization.
TechTarget defines data quality as “a measure of the condition of data based on factors such as accuracy, completeness, consistency, reliability, and whether it is up to date.” So, what are the widely accepted metrics and measurements of data quality?
1. Accuracy - the percentage of errors discovered in data records and how that percentage tracks with data integrity rules
2. Completeness - the number of records with missing or incomplete data
3. Consistency - how well your data is merged from different sources and combined into a Single Source of Truth.
4. Timeliness - how real-time records are tracked, used and archived according to time stamping to prevent aging of the data set
5. Uniqueness - how the input system copes with and tracks duplicate data
6. Validity - how the data conforms to standards of formatting so that it can be incorporated into the business rules and operations within the bounds of data governance policies
Say that your manufacturing company wants to create a quality dashboard as part of its Continuous Improvement Initiatives. The first step is to consider the amount of data needed to get a complete and accurate view of the company's critical quality metrics (data points) and requirements of data profiling.
However, most companies create and collect large amounts of process data but typically use them for tracking purposes only. So, the analytics dashboard needs to include scrap data, warranty claim data, inbound inspection data, rework data, and more to get a complete and accurate picture of the “quality health” of the company.
Conventional thinking has been that businesses can glean the data for a quality business intelligence dashboard from an Enterprise Resource Planning (ERP) system. However, complete reliance on ERP won’t help much in fusing data with manufacturing.
That is because much of the detailed data from the shop floor is not captured in the ERP system or stored in the enterprise data warehouse. Instead, this data is processed and stored in multiple, fragmented systems. The key is to break down those enterprise data “satellite silos” and capture the data for the (analytics mentioned above) dashboard.
In turn, breaking down those silos that rely on many data points can result in a replicable, proactive, and consistent approach to data accuracy and myriad related issues. That consistency builds organizational trust in the information and leads to increased adoption across the organization.
In Manufacturing 4.0, production, quality control, and demand planning are among the many functions that manufacturers can enhance through good data quality. Having trusted and complete data improves visibility into manufacturing processes and reduces or eliminates engineering flaws, manufacturing over-and under-runs, product defects, and other quality-related problems.
The opportunity in the previous scenario is its investment in data quality tools and the right skill set, which will allow the business to prep, blend, and cleanse its existing process data with the data in its ERP and EDW systems. As a result, the data can quickly analyze and spot patterns to draw actionable insights and the best business intelligence from the information.
Discover how you can take the first steps to establish a data strategy that works. Download our eBook "The Executive's Guide to Building a Data Strategy That Leads to Business Growth & Innovation."