Using data to drive safety performance: from Excel spreadsheets to real-time dashboards

Most organizations start tracking safety performance with simple figures: frequency rate, severity rate, number of accidents. These indicators are useful, but they often arrive too late. They tell you what has already happened. Yet managing means preventing it from happening. The real challenge of HSE data is therefore to move from a “reporting” logic to a “control” (risk mastery) logic.

The most realistic path is a maturity journey. At the beginning, a solid Excel file can be enough—but it must be well designed: stable definitions, comparable data, and above all indicators that support decision-making. Many spreadsheets fail because they confuse the amount of information with the ability to act. Useful “steering” answers precise questions: where are our major risks? where are we drifting? which actions are overdue? where is exposure increasing? which site is showing weak signals? If your spreadsheet does not answer these questions, it is not steering—it is archiving.

The first step forward is to balance “lagging” indicators with “leading” indicators. Accidents and occupational illnesses remain essential, but you also need to look at what comes before them: reports of hazardous situations, near misses, results of field observations, compliance of critical controls (lockout/tagout, permits, inspections), up-to-date training/authorizations, the quality of incident investigations, and action closure time. This shift changes the culture: you no longer measure only what hurts—you measure what protects.

Then comes the most underestimated topic: data quality. A company can have a beautiful dashboard and poor data—and therefore make bad decisions with high confidence. Here, quality is first and foremost a definitions issue. What is a “near miss”? What is a “safety visit”? What counts as an action “closed”? As long as sites do not speak the same language, the dashboard is an illusion. HSE data governance is a real project: definitions, data-entry rules, consistency checks, contributor training, and arbitration of exceptions.

Moving to dashboards (and even real time) should not be a technology obsession. It must be justified by a managerial benefit. For example, in subcontracting and simultaneous operations, near-real-time monitoring of authorizations, prevention plans, certifications, and deviations can genuinely reduce risk. Likewise, on multi-shift sites, real-time visibility on certain signals (work permits, ongoing lockouts, simultaneous operations) can prevent dangerous situations. But on other topics, “real time” adds no value: a well-run weekly review can be far more effective than a continuous stream of data.

Another key point is integration into the routine. A dashboard should not be “a screen.” It should feed a structured discussion: what is drifting, why, what decision do we take, and how do we verify effectiveness? Mature organizations use data to make prevention a management discipline: fewer opinions, more facts, and above all more decisions.

Conclusion

HSE data is valuable not because it is sophisticated, but because it supports better decisions. Good safety steering is not about “seeing everything.” It is about seeing early, seeing accurately, and quickly turning information into action.