How predictive analytics inform workplace safety
The business world has more access to detailed data than ever before, and companies are using what’s come to be known as “big data” to be more competitive, more efficient, and more profitable. By applying the science of analytics to their data, managers can more accurately predict the outcomes of decisions they make. For example, a fast-food chain can determine exactly how minor pricing changes will impact total sales.
Safety professionals also have access to a significant amount of data. Beyond the information that’s reported to OSHA and other agencies, safety pros can track details of inspections, monitoring, employee absences, and a wide variety of other data sets. By using systems to analyze all that data, they can learn more about the safety of their workplaces and opportunities for improvements in processes and procedures.
It’s important to note that all this data and the techniques we’ll describe here cannot replace the judgment, experience, and intuition of a safety professional. A computer cannot walk into a jobsite and react to a condition it sees. However, access to more data and analytical tools can inform the work safety professionals perform, helping them become more accurate and efficient and making better use of their time.
In simple terms, data that’s captured can provide two types of insight. Some data provides insight after the fact, in that it explains what happened after it happens. That data provides what’s known as lagging indicators. Your worksite may have had 26 lost-time injuries this year, 18 of which required medical attention, and 21 of which involved injuries to workers’ hands. That’s lagging data, because it’s historical in nature and you can’t go back and change it.
The other kind of data analysis uses predictive data. It’s a kind of data that says if something happens, something else is likely to result. By subjecting large numbers of observations and incidents to data analysis, it becomes possible to predict the likelihood of safety incidents with surprising accuracy. As safety professionals, we share a belief that it’s always better to prevent an incident than to have to investigate an accident. Using predictive analysis effectively can pinpoint the situations in which accidents are most likely to occur and uncover factors you may not have considered.
Simply achieving a very low incident rate or a perfect safety record for the past decade doesn’t ensure that your company won’t suffer some type of accident tomorrow. We’ve all seen cases in which a momentary lapse on the part of a worker or the failure of some type of device either caused or contributed to an incident. That’s where adding a predictive component to your safety program can help. Thoughtful analysis of the data may help you see patterns in behavior that haven’t been obvious.
We know from experience that the more often a worksite is reviewed by a safety professional, the fewer the number of safety incidents that will occur. There can be any number of reasons, including the tendency of workers to pay more attention to safe procedures when they’re being observed, and that observations can uncover unsafe practices of which the workers were unaware.
Most companies believe that establishing a goal of having all inspections turn up no violations or problems is as good as it can get. But does complete compliance equate to complete safety? As tasks change, as new employees transition to the workplace, as new types of equipment are being used, data based on previous incidents doesn’t apply. Any of those new elements could be bringing new and unfamiliar risks to the worksite, and the people doing the inspections may not be aware of those risks.
Even when inspections discover frequent violations, the problem may not rest with the workers who are performing tasks incorrectly. Instead, the real issue may be with the company’s response. If no action is taken to fix the unsafe situation, or workers are not held accountable for their behaviors, the violations are likely to continue.
In both of these situations, a predictive analysis component that’s based on readily available data may be more effective at protecting workers and the company because it takes a broader view and allows safety professionals to study the workplace from a different and more objective perspective.
As safety professionals become increasingly aware of the benefits of predictive analysis, companies are creating systems to automate the process. When considering such a system, it’s important to look beyond hype and promises to the fundamentals of how the system works and what it considers. One of the most important is the number of factors the system tracks and analyzes. A system that’s limited to a small number of factors is inherently less powerful than one examining a long list of workplace indicators. In addition to data from the company’s own experiences, that should include information about the environment within which workers perform tasks, as well as behavioral and other factors.
Having a broad range of factors isn’t enough if the source of that information is subjective, such as limiting data to the observations made by safety professions during inspections. Ideally, a system should also incorporate plenty of objective data, whether that’s from regulators or industry sources. That doesn’t suggest a lack of faith in the safety professionals’ ability to identify dangerous situations; instead, it reflects the reality that we’re often unaware of certain hazards until something happens. Once again, objective data broadens our examinations.
An effective predictive analysis system should do more than provide a high-level view of the potential for accidents. It should allow the users to drill down to specific task areas and individual workers to identify the areas of the greatest risk. It’s somewhat helpful for the system to say that workers in one area of your facility appear to be at higher risk; it’s far more useful for it to say that a worker in that area is more likely to be injured when performing a particular task in specific conditions.
Most safety professionals take pride in using a proactive approach to protect workers, and it’s true that most safety approaches are proactive to a degree. Expanding efforts to include a predictive analysis of data makes their efforts even more proactive and effective, creating the potential to dramatically reduce the number and cost of incidents.