Manjish Naik and Sarah Lukens join us to discuss how to overcome data quality issues with KPIs.

What is a KPI?

A Key Performance Indicator (KPI) measures the performance of an asset in either a process or manufacturing plant. The metrics can be high-level, for instance Overall Equipment Effectiveness (OEE), or granular like corrective or reactive count.

KPI metrics can fall into two categories – leading or lagging. Leading metrics give you a forecast of what’s to come. An example is PM compliance. By meeting your PM schedule, it would allow you to find issues in advance, making you more proactive towards fixing arising issues. Lagging indicators like reactive count, take into account something that’s already happened, to analyze and improve future processes.

Having a varied system of KPIs to work with helps you to better understand your asset performance in the field.

How to get started with data

KPIs have been known to bring about issues such as inconsistent measurements and inconsistent data. So, how do we get past these? It all starts with the quality of the data. These are a few questions you need to answer:

1. What is the importance of good quality data?
2. What is good quality data?

Data quality refers to its fitness to a given purpose. It needs to be complete enough to allow you to measure and understand critical indicators.

How to establish KPIs

KPIs get calculated from the maintenance work done physically. It can also be calculated from your Computerized Maintenance Management System (CMMS), like SAP or Maximo. The process starts with you understanding how many failures happened within a specific period. You could establish this by talking to floor technicians (which isn’t very scalable), or relying on data from CMMS

The data quality from a CMMS is preferred as it helps understand how metrics get calculated, and what insights come out. A CMMS has predefined fields for cost, parts, labor, skill type, activity type, among others. However, these still don’t guarantee quality of the collected data.

Factors hindering the accuracy of data quality

There are three main factors:

1. People – Humans manage the data collected throughout the work order on the CMMS. If they’re unaware of the intended use of the data, they may not give accurate input.
2. Processes – Rarely are there well-defined processes for people to input data into the CMMS. As such, everyone comes with their own processes to complete the task, with some resorting to ways of circumventing the entire process.
3. Technology – A CMMS may not be designed with reliability in mind. It may not take into account the particular analytics of the given plant, missing the necessary fields to calculate certain metrics. Codes relating to work order types may not be clear, or the taxonomy may be ambiguous. 

There are organizations trying to streamline data quality issues. These include:

1. ISO 14224 – They offer information on how to set up a data collection system for maintenance.
2. SMRP – They offer guides on availability in metrics.

However, even with these organizations, we still can’t measure data correctly. That’s because:

1. These organizations offer theoretical solutions without practical recommendations.
2. The CMMS might not be effective in capturing the information that’s necessary for calculating metrics.
3. Without having a way to define the dates of when assets go down or when they got fixed, you can’t have consistency in your availability evaluations.

How to overcome data quality issues

1. Improve historical data already in the system

Rather than storing the CMMS data you need to be able to extract this information in a scalable way. Natural Language Processing (NLP) approaches, responsible for programming computers and algorithms, can be used to extract and refine historical data.

Is it worth it to go back and clean the data, or should the organization move forward collecting new data? 

The answer depends on whether you can afford the cost, and if you’ll get value at the end of the process.Fundamentally, it’s worth it if you can go at least five years or a decade in the past, and no more. So, rather than paying to create and store data, organizations should unlock historical data. It’s necessary to focus on unlocking valuable assets rather than every stored asset to get the most value.

2. Applying industry best practice

Take a holistic approach on aspects like:

1. People – Hold training to show how the data recorded in the CMMS gets used
2. Processes – Have standardized definitions for all the maintenance terms, with practical examples on how to get these into the CMMS
3. Technology – Design CMMS with KPIs in mind to have all the required fields. The picklist should also be clear enough for anyone to use

How can you improve your current data quality?

Start by measuring your data quality to separate the good data from the bad. For the not so good data, find out what makes it substandard. You can also start implementing best practices at your organization. As for your historical data, start using NLP processes to extract and make use of what you already have.

How to become successful with data quality

You need the correct mindset to start getting good quality data. Also, don’t get overwhelmed and don’t procrastinate. This will help you to sustain your improvements through regular monitoring and tracking of the system.

Eruditio Links:

• Eruditio
• HP Reliability
• James Kovacevic's LinkedIn
• Reliability Report
• Eruditio Supports: www.help.eruditio.com

Manjish Naik & Sarah Lukens Links:

• Manjish Naik Linkedin
• Sarah Lukens Linkedin
• PHM Society
• SMRP.org
• SMRP Best Practices Guide

204-Data Quality Issues with Manjish Naik & Sarah LukensJames Kovacevic

share

• Social:
• Link:
• Embed:

Rooted In Reliability podcast is a proud member of Reliability.fm network. We encourage you to please rate and review this podcast on iTunes and Stitcher. It ensures the podcast stays relevant and is easy to find by like-minded professionals. It is only with your ratings and reviews that the Rooted In Reliability podcast can continue to grow. Thank you for providing the small but critical support for the Rooted In Reliability podcast!