We are all familiar with the typical Safety Observation Reports that our systems produce today. They present data collected in the past week or month such as the number of observations, a chart of observations by safety category, and, for the more sophisticated systems, a list of the observation action items that remain open.
But when I was a project manager, I often wondered how much all this data from the past would actually help my team identify and prevent the hazards that will affect their crews tomorrow or the next day. It seemed as though I was always chasing my tail. It’s also what led me to build a better way to use safety data.
Classical or traditional observation reports are summaries and aggregations of observations, along with additional information such as how many observations were conducted and who made them. If you notice that one site has PPE compliance issues or another that has a lot of housekeeping problems, you can take action to address these specific issues, but they don’t provide hard, forward-looking indicators of risk. We need to identify those metrics that strongly indicate what will happen in the near future. A classic example of this type of metric is the “death cross” from economics.
The death cross is one of the most well-known economic indicators, and it occurs when the price of an asset falls below its moving average. If a major stock index or the price of treasury bills, for instance, experience a death cross, that’s a strong sign that a recession is looming.
So while the information in a traditional safety observations report can be used to spot historical trends, we have no guarantee those trends will continue. Metrics that we know are correlated with increased safety risks are far more valuable.
Instead of relying on report values (values that directly reflect the underlying data), the industry needs to look for predictive metrics that provide real insight. At Smartvid.io, we look for them by entering into a predictive engagement with our customers to layer their data and examine more than just observations. We look at observations within the broader context of additional data, such as incidents, photos and other information that is collected every day on projects.
Of course, safety observations are only useful if they’re done properly. No matter how predictive a particular metric may be, if you have bad data, your metrics will be useless. So it’s critical to ensure that people are gathering data in the right way. You can read about how to make a proper safety observation here.
Once we identify the metrics that act like a “death cross” for safety in construction, safety observation reports can act as an early warning system, letting us know when projects are in danger of an incident. From our modeling, we know that when the rate of observations drops below a certain value, craft worker engagement dips and the risk of an incident increases. Knowing this, management can keep a close eye on that particular measure and take preventative actions when it reaches a critical point.
We have also seen a strong correlation between high observation risk ratings and safety incidents. In our system, each observer assigns scores for frequency and for severity on a one-to-five scale whenever he or she makes an observation. So if you see a high-risk rating on your report, our modeling has shown that site to be more likely to experience a safety incident in the following week.
As our analysis of industry data improves, additional “death cross” metrics for construction will emerge, which is good news for everyone. Because by employing predictive metrics, we give our project teams superpowers that enable them to see into the future, so they know precisely where they need to focus their attention to prevent an incident before it occurs and stop chasing their tails!
Interested in learning more about predictive analytics and how your company can use it to prevent incidents before they happen? Sign up for a complimentary predictive analytics assessment today!