Unlock Safety through Data Analytics
Originally published: 08.01.18 by Ali Vahed
How one commercial construction and energy firm uses analytical data to improve safety throughout the company.
McKinstry, like many construction companies, is a family. As a family, our top priority is making sure everyone goes home safe and healthy after each and every day. That’s why we take safety so seriously.
Numbers don’t lie; construction is dangerous. According to the U.S. Occupational Safety and Health Administration (OSHA), one in five workplace fatalities are in construction. Improving construction safety cannot happen by accident. We ust make a conscious, deliberate effort to do better. Big data and business intelligence dashboards offer a platform to drive awareness, accountability and change.
McKinstry started its analytical data safety transformation in 2013. Over the past five years, we’ve developed a safety analysis and reporting program based on deep data tracking, analytics and reporting. Here are the six steps we followed along with some insights gleaned along the way.
The End Comes First
Start with the end in mind. Your desired end state should drive program development from day one. Gather each of your stakeholder groups to identify what goals you hope to achieve, who needs to be involved and how they’ll interact with the platform.
The McKinstry team engaged stakeholders early to make sure we created something employees would actually use. To do that, we needed to understand what they needed and wanted. More on that buy-in process later.
Stakeholders identified the following outcomes for the analytical data platform:
Comprehensive: Safety needed to be relevant for every employee across the company. The team wanted to drive safety accountability from the executive team on down — hitting our jobsites, our fabrication shops, our service fleets, our sales teams and our corporate support team. Data criteria needed to cover incidents across each group — from personal protective equipment to keyboard ergonomics.
Actionable: Gathering and reporting data is fantastic, but it means nothing if teams cannot take action. The platform needed report templates offering flexibility and depth. Teams needed to understand how to use the report templates to drive real behavior change.
Real-Time: The platform needed to report real-time data. Monthly, and even weekly reports create gaps that delay analysis and action. Our teams needed access to data in shorter increments to track and forecast trends. Technology is available to make real-time, or near real-time, reporting easy. Why not use it?
Eliminate the Middle Man: Inserting a safety team member into the process as a required intermediary creates a bottleneck and major point of failure for data and reports.
I know this from experience. It used to be my job.
Creating reports from our legacy system of disparate Excel spreadsheets took 15 hours a week or more. The new platform had to connect users directly to the data without any bottlenecks in the middle.
Open to Everyone: The entire McKinstry team needed access to safety data and reports. This level of transparency is scary, but very necessary. The team wanted to remove any barrier that could hold back improved safety.
What’s more, McKinstry wanted the ability to share data with our partners and clients. Safety is a core value at McKinstry. We walk the talk. Our extended family needed to know that they could count on our ability to deliver a safe jobsite.
Identify What You Have; What You Need
Data analytics is only effective if you collect the right data. McKinstry obsessed over data taxonomies and governance. It’s not a challenge you should tackle alone. That’s where I entered the picture.
McKinstry contacted local universities to find a talent partner. At the time, I was completing my master’s thesis on risk assessment tools for mechanical contractors.
No one could deny the match, so an internship was created. Together, we established a layered taxonomy to track hundreds of data points:
• Description of reported safety incidents
• Recordable incident rate (injuries per 200,000 hours worked)
• Injury costs (tracked across downtime, lost hours, lost productivity, etc.)
• Safety training and certification compliance
• Office safety incidents
• Fleet accidents with details
• Tools in use
• Design elements
• Worker’s compensation details
• Severity details
• Levels of experience
From there, the McKinstry safety team engaged an obvious partner – our insurance provider. Insurance companies have leveraged big data for decades to assess and track risk. They were eager to help, offering suggestions to advance our datasets.
McKinstry also set positive data points to record safety recognition and awards. The team set a goal and tracked an industry best-in-class metric for each data point.
All of this data can be sliced by time, location, business unit, project, team and more. We have the ability to focus on issues with surgical precision, find the underlying causes and implement fixes.
Find the Right Tool
There are hundreds of data analytics and business intelligence platforms available today. We chose Tableau and Microsoft Power BI for pilot testing. Both were more than capable, but Power BI was a better fit for our specific requirements.
Why? Because we started by identifying our desired end state. The platform needed to be open and available to anyone. McKinstry uses the Microsoft 365 platform as a technology standard. Power BI is part of that platform and has better integration.
Also, every member of our team has a license and access to the software by default.
McKinstry designed the safety platform with stakeholder engagement from the beginning. The team crafted an internal communications strategy to expand that engagement companywide.
McKinstry launched a new incident reporting hotline around the same time. The program fed off each other, creating a mechanism for the larger safety transformation.
We wanted to shift employee’s perception of what constitutes a safety incident.
Jobsite injuries of course needed to be reported, but safety expands even further into the enterprise.
Overgrown shrub blocking parking lot visibility; report it. Having ergonomics issues with your chair; report it.
The incident reporting hotline drove engagement, creating new data to populate the database for tracking, analysis and reporting. Surprisingly, most incidents were self-reported.
Early stakeholder engagement demonstrated our commitment to change. It showed employees that we all had to be personally accountable. It created a true safety culture across the company.
Yes, our number of reported incidents skyrocketed. That spike could be normalized during analysis.
Take Action; Create Change
Initial engagement exceeded expectation. The McKinstry safety team knew we needed to take visible action to maintain engagement and change behavior. Training became our early victory.
The high number of reported incidents painted a clear picture early on. The most reported, and most alarming, incidents were highest in apprentices and new employees working within 90 days of the start date.
The team quickly reworked our new employee training programs to address recurring safety issues. We trained team leaders to reinforce that training to apprentices and new employees in the field.
The new training programs worked. Employees noticed. Engagement continued to grow.
The biggest “aha” moment came with leadership engagement. Before the data analytics platform, leadership team questions and report requests focused on “what’s happening.” The new platform provided ongoing, real-time access to that information.
Now, the leadership team conversation focuses on “why” and “what are we going to do about it.” Providing access to data shifted the conversation to action and solutions.
Never Stop Improving
The McKinstry safety analytics platform is now in version 3. We’ve launched three major upgrades based on user feedback and requests. Another is under development.