Don’t Procrastinate, Innovate: ‘Data’ Should Not Be A Four-Letter Word

As W. Edwards Deming once admonished us, “In God we trust, all others must bring data!” Today, we have plenty of ways to deliver such data.

Thanks to advances in technology, we’ve seen a proliferation of inexpensive, data-hungry smart communication devices (i.e, phones, tablets and cameras) with the ability to directly access asset-management software data. Most recently, we have witnessed the advent of smart handheld instruments (i.e., thermographic-, vibration-, alignment- and electrical-measurement devices) and smart machine interfaces (i.e., Fault code directories, Building Automation Systems [BAS] and Supervisory Control and Data Acquisition [SCADA] systems), many of which are creating their own trending databases in “the cloud.” Those cloud environments, in turn, make for easy access by maintainers, contracted resources and end-use clients.

In many cases, today’s smart handheld instruments and machine interfaces are interoperable, meaning they can communicate with one another and compound their collective data. This capability allows the average maintenance department to offer universal access to all of its related maintenance data and implement a data-informed decision-making approach to asset management. Yet, how many maintenance organizations have done so?

Overcoming the barriers

Deming, of course, was a statistician. He was accustomed to manipulating and making sense of large amounts of raw data. Universal access to raw data, however, is fairly new to most maintenance departments. Unfortunately, unless such data is meaningful, structured and managed carefully, the average user can be quickly overwhelmed. Thus, he/she might, in fact, reject the very data that could be used to focus and simplify a maintenance process.

In 2009, the U.S. Department of Education’s Office of Planning, Evaluation and Policy Development published “Implementing Data-Informed Decision Making in Schools—Teacher Access, Supports and Use.” The report reviewed how teachers— particularly those in districts identified as high data-users—understood and used their data to make informed decisions. Although the school districts had amassed an incredible amount of student-related data, around which it had provided teachers basic training and universal access, the study came back with some surprises, specifically:

• Among teachers with access to data, only 33% felt capable of forming queries and accessing pertinent data.
• Only 42% of teachers agreed that their school had been improved through the use of data.

The study found three main barriers to implementing a usable data-informed decision-making approach to teaching:

1. Lack of training in how to use the data system or to derive instructional implications from system data (ability to query and make sense of the data).
2. Lack of time to engage in data exploration and reflection (time to download and analyze, and prepare data for use).
3. Weakness of the available data (not all data collected was useful or complete).

Despite those findings, the study also identified a positive and innovative approach to understanding and working with data that seemed worthy of modeling: a medium-sized school district that conducts annual “data digs” during which school-leadership teams collaboratively review, discuss and make sense of their student data. The teams use “Data-Driven Dialogue” to analyze and interpret their data. The process is adapted from Bruce Wellman and Laura Lipton’s 2004 paper, “Data-Driven Dialogue: A Facilitators Guide to Collaborative Enquiry”—and it’s easily adaptable for use with maintenance and asset-management data. The process is broken into four steps:

1. Predict—Activate and engage interest in the data, access prior learning, name frames of reference, and establish common ground for dialogue.
2. Explore—Interact with the data, look for patterns and trends, identify data facts and surprises, make observations without inference, identify questions raised by the data, and develop problem statements.
3. Explain—Generate theories of causation, stay open to multiple possibilities, deepen thinking and identify “root causes” rather than symptoms, make inferences about data, and identify the additional data that will validate the theories of causation.
4. Take Action—Move from problems to solutions based on validated theories of causation, identify goals and specific related action steps, and identify data to be monitored to determine whether action steps lead to the solution.

Making sense of your data

It’s clear from the Dept. of Education study noted above that others have encountered many of the same problems we in the maintenance world are beginning to face as plant data becomes more accessible. Still, it’s important for us to remember what Albert Einstein said: “Not everything that can be counted counts, and not everything that counts can be counted.”

In other words, we must beware of storing Meaningless Unrelated Data (MUD) in our databases—and be judicious in the data we collect to ensure it is meaningful for our reporting needs. A good starting point is to select a number of meaningful Key Performance Indicators (KPIs) required to run our business and determine what data is required to be sorted and reported on through the CMMS/EAM system or external database(s) to calculate the values of those indicators.

If maintenance personnel are to leverage and challenge data, they must first comprehend it and understand how it can be used to tell a story. Data is typically displayed in tables, graphs or printouts and can be complex. The maintainer or data-user needs to comprehend what’s in front of him or her. Performing “Data Digs” can be an excellent training and evaluation process that can be managed more easily by regular meetings of groups based on work discipline (mechanical, electrical, etc.).

Data analysis (also known as analytics) is used to apply a meaningful and relative visual appearance to data to communicate direction and insight. Accomplished through the use of graphs and charts, the displayed data should be interpretable immediately in a currency or language understood by the person(s)/department(s) receiving it. Consider the following example:

An improvement in an asset’s availability would be presented to a maintenance audience as a percentage of the maximum availability. The graph might show an increase from 92% availability to 97% by means of a bar or X-Y graph over a specific time period. Depicting that increase of availability for manufacturing or production staff, though, could be more meaningful and understandable when presented as an ability to increase the number of manufactured parts or throughput gain. Conversely, if this gain were to be communicated to the financial department, it would be best expressed in terms of additional $ capacity based on throughput capability. Translation: same data, three audiences, three currencies, which is in line with Einstein’s observation that “if you can’t explain it simply, you don’t understand it enough.”

The bottom line for readers of this magazine is that accurate data analysis and interpretation, particularly at the department-management level, may call for additional training on basic statistical concepts. When set up correctly, data provides the “Digital Ability To Analyze” a plant’s current state of maintenance and turn those revelations into quality data-informed decision-making and insight into the future. When complied, good data should be “Dependable, Actionable, Tangible and Available.” It’s not just a four-letter word. Good luck!