Fatality rates across UK industry are telling a story of two trajectories. In some sectors, particularly manufacturing and agriculture, steady investment in technology and process discipline has cut workplace deaths dramatically. In others, including construction, logistics and parts of the energy and service ecosystem, fatality numbers remain flat or have begun to climb again — despite unprecedented access to data, automation and monitoring systems that should, in theory, have made work safer.
New figures from Astutis show that workplace fatalities in public administration, defence, health and education have risen by 67 percent in the decade since 2014–15. Deaths in arts, entertainment and recreation are up 25 percent, and those in wholesale, retail and hospitality have climbed by around 20 percent.
In contrast, the traditionally hazardous agriculture, forestry and fishing industries recorded a 28 percent reduction, while manufacturing cut fatalities by almost 39 percent. Construction stands apart as the stubborn outlier: 35 deaths this year — exactly the same as a decade ago. The data paints a clear picture of divergence. Some sectors have embedded safety improvement as a continuous data-driven process where others have relied on compliance and incident response rather than prevention — a distinction that now shows up starkly in the statistics.
According to Rick Standish, Executive Industry Consultant at Hexagon Asset Lifecycle Intelligence, the divide is less about the tools used than the mindset behind them. “We see safety outcomes diverge across industrial sectors based on how each sector thinks about risk, more than the tools they use,” he explains. “In industries we serve, such as energy production and process, there’s a deeply ingrained safety mindset: Safety is treated as a closed-loop, data-driven, continuous-improvement process. Some sectors, in contrast, still manage safety through reactive compliance.”
That difference in maturity is visible across much of UK industry. In the energy and process sectors, digital twins, automation and condition-based monitoring are integrated into daily operations, creating systems that remove people from high-risk areas and enable early intervention when hazards emerge. Incident rates fall where data is trusted and acted upon in real time. They rise where information remains fragmented, delayed, or lost in paperwork.
Moving from reactive investigation to predictive prevention, Standish says, “means connecting engineering, operations and maintenance information into one digital thread.”
This thread is, increasingly, what separates those industries reducing fatalities from those standing still. Manufacturing and process plants lend themselves naturally to structured systems that allow safety technology to embed effectively. Construction and logistics, by contrast, struggle with fluid work environments, transient labour forces and fragmented oversight chains.
The problem is not lack of access to innovation, but the absence of a uniform framework to apply it. “Sectors such as construction, logistics and warehousing face a different set of challenges,” notes Rayan Cherri, Chief Marketing Officer at Ipsotek.
“Their environments are more varied, with dispersed or transient workforces and constantly shifting conditions. The difference is less in access to technology than how effectively it can be adopted, using data-derived insights to reinforce safe practices, inform training and guide operational decisions. Where processes are standardised and supported by live-generated insights, safety performance improves; where they are fragmented, progress is naturally slower”
A culture of safety
Technological capability, in other words, is no longer the limiting factor — organisational culture is. Safety technology has become pervasive, from wearables and sensors to computer vision and edge analytics. Yet technology can only highlight risk; it cannot ensure that leadership acts on it.
“Computer vision can interpret live CCTV footage to detect risks such as missing PPE or unsafe proximity to machinery,” says Cherri, “Yet identifying risk is only the first step. The real impact comes when this data informs training, process reviews and leadership decision-making.”
The result, seen across many mid-risk sectors, is a widening gulf between detection and correction — a gap that can prove fatal.
For industries where operations are complex, distributed or high-intensity, connectivity itself is becoming a frontline safety tool. Stephane Daeuble, Head of Value Proposition and Business Development for Enterprise Campus Edge at Nokia, argues that the next major leap will come from integrating automation, robotics and real-time monitoring over private wireless and edge computing networks.
“Automation, robotics, and intelligent monitoring systems have already saved countless lives by removing workers from hazardous environments and detecting dangers before they escalate,” he says. “Once a Private Wireless and Edge computing foundation is in place, industrial enterprises can deploy ‘over-the-top’ safety applications that extend pervasive coverage, connect workers, and embed safety into every layer of operations.”
Those applications include geo-fencing and positional tracking to enforce restricted zones automatically; biometric and environmental sensing to detect fatigue, heat stress or exposure; and AI-driven vision systems to monitor PPE compliance and unsafe behaviour with on-premises processing. By fusing data from IoT devices, cameras and sensors, Daeuble says that industries move “beyond situational awareness towards contextual awareness — understanding not just what is happening, but why, and what might happen next.” The ultimate goal is predictive safety: eliminating exposure before an incident can occur.
Such systems are transforming how risk is understood, but they do not replace the human element. Leadership commitment and workforce engagement remain the decisive factors in whether data translates to prevention. As Daeuble notes, “Technology can drive a quantum leap in safety performance, but it cannot replace leadership accountability or cultural commitment. When AI, automation, and connectivity are coupled with a culture that prioritises safety and integrity, fatalities can become not just rare — but preventable.”
Across industry, those sectors with ingrained safety governance and strong data cultures are trending towards zero harm. Those that continue to treat safety as a compliance box, rather than a live operational discipline, are stagnating. This results in an uneven industrial landscape in which the difference between progress and regression lies less in innovation budgets and more in trust, leadership and follow-through.
Industrial safety has entered a split decade: one half connected, predictive and improving; the other reactive, fragmented and still losing lives. The challenge for the years ahead is not to invent new tools, but to ensure that existing ones are embedded in the cultures, decisions and daily practices that determine whether risk is managed, or merely measured.
