What is socio-technical safety?

Why this topic matters

When something goes wrong in a safety-critical setting, it is tempting to ask who made the mistake. Sometimes individual actions matter. But in complex systems, that question is rarely enough.

Healthcare, aviation, rail, emergency response and AI-enabled operations are not just collections of people and equipment or technology. They are networks of tasks, tools, information, procedures, organisational priorities, regulation and local conditions. A medication process, an aircraft approach, a tram operation or an AI-enabled runway inspection all depend on many parts of the system working together.

Socio-technical safety starts from this point. It asks how safety is supported, weakened and recovered through the way the system is designed and operated. It helps us ask better questions: What information was available? What pressures shaped the decision? What feedback was missing? Which assumptions were built into the procedure or technology? How did different parts of the system influence one another?

These questions do not remove responsibility. They make learning more useful. They help improve the conditions in which people work, rather than simply asking people to “be more careful”.

What I mean by a socio-technical system

A socio-technical system is a system where social and technical elements are closely connected. The social side includes people, teams, communication, expertise, culture, leadership, organisational routines and regulation. The technical side includes equipment, interfaces, software, data, infrastructure and the physical environment. Procedures often sit between the two: they are formal artefacts, but their effect depends on how people interpret and use them.

In practice, these elements cannot be separated neatly. A procedure changes how people coordinate. A display changes what people notice. Staffing levels affect what work can realistically be prioritised. A regulation changes what organisations design, monitor and report. A new technology may solve one problem while creating new forms of coordination, workload or uncertainty.

Socio-technical safety focuses on these relationships. It treats safety not only as the absence of accidents, but also as something produced through design, coordination, monitoring, adaptation and learning.

A simple way to think about it

A useful question is: what had to go well for the work to succeed?

In everyday operations, people often adapt to variation. They manage interruptions, incomplete information, time pressure, changing priorities and imperfect tools. Most of the time, these adjustments help the system succeed. Occasionally, the same kinds of adjustments can combine with other conditions and create risk.

This is why socio-technical safety is interested in normal work as well as unwanted events. Understanding how work usually succeeds can reveal where the system is resilient, where it is fragile, and where people are quietly compensating for design or organisational weaknesses.

What this helps you notice

A socio-technical view can reveal issues that are often missed in routine safety analysis. It can show where procedures assume more certainty than the work allows, where decision-makers do not receive the feedback they need, where technology leads to workarounds, or where organisational pressures make some trade-offs more likely.

• In healthcare, risk assessment is not only about the risk form or the risk matrix. It is also shaped by how risk is defined, who takes part, what guidance is available, how uncertainty is handled and how the findings are used.
• In aviation, approach stability is not only about aircraft parameters. It is also influenced by monitoring, briefing, communication, air traffic control interaction, weather, timing, workload and operational expectations.
• In rail and tram operations, risk can emerge from the interaction between drivers, pedestrians, vehicles, signals, procedures, control rooms and the urban environment.
• In AI-enabled systems, safety cannot be assessed only by asking whether the algorithm performs well. It also depends on human oversight, feedback, responsibility, degraded modes, training and assurance.

How to use this lens in practice

A socio-technical lens is most useful when the purpose and boundary of the analysis are clear. A very narrow boundary can hide important relationships. A very broad boundary can make the analysis difficult to use.

Start with the activity or event you want to understand. Then ask who and what is involved, what information flows between them, what decisions are made, what constraints shape those decisions, and what feedback is available, delayed, missing or misleading.

It is also important to distinguish description from judgement. Mapping interactions should not become a way to blame more people across more levels of the system. The value lies in understanding how actions and decisions became understandable in context, and what could be changed to support safer performance.

Choosing a method

Socio-technical safety is a perspective, not a single method. Different methods help answer different questions.

STPA can be helpful when the focus is on hazards, control actions, safety constraints and feedback. FRAM can be helpful when the focus is on everyday work, performance variability and how functions depend on one another. CAST can be helpful when analysing an event through control structures, responsibilities, feedback and systemic constraints.

The best method is not necessarily the most complex one. It is the one that helps people understand the system well enough to improve it.

Limitations and cautions

Socio-technical analysis can become vague if it only says that “everything is connected”. Its value comes from being specific: which interactions matter, what evidence supports that, and what can be improved?

It is also not a replacement for technical assurance, compliance, professional judgement or quantitative analysis. Component reliability still matters. Standards still matter. Procedures still matter. The point is that they are not enough on their own when safety depends on interactions across people, technology, organisations and environments.

Used carefully, socio-technical safety analysis supports better design, better assurance, better learning and more realistic risk management.

Related publication(s)

• Kaya, G.K. A system safety approach to assessing risks in the sepsis treatment process. Applied Ergonomics. 94, 103408. DOI: 10.1016/j.apergo.2021.103408.
• Kaya, G.K., Ozturk, F. and Sariguzel, E.E. (2021). System-based risk analysis in a tram operating system: integrating Monte Carlo simulation with the functional resonance analysis method. Reliability Engineering & System Safety. 215, 107835. DOI: 10.1016/j.ress.2021.107835
• Kaya, G.K., Stallard, R., St-Laurent, M., Li, W.-C. and Sujan, M. (2026). Exploring unstable approaches in aviation: utilising functional resonance analysis method. The Aeronautical Journal. 130(1345), 917–943. DOI: 10.1017/aer.2025.10108
• Losi, E., Kaya, G.K., Camelia, F., Chatzimichailidou, M., Slater, D.H., Patriarca, R. and Sujan, M. (under revision). Systemic safety analysis of complex socio-technical events: insights from applying CAST and FRAM. Reliability Engineering & System Safety. Publication details forthcoming.

Selected references

• Carayon, P. (2006). Human factors of complex sociotechnical systems. Applied Ergonomics, 37(4), 525–535.  DOI: 10.1016/j.apergo.2006.04.011.
• Cherns, A. (1976). The principles of sociotechnical design. Human Relations, 29(8), 783–792. DOI: 10.1177/001872677602900806
• Cherns, A. (1987). Principles of sociotechnical design revisited. Human Relations, 40(3), 153–161. DOI:10.1177/001872678704000303
• Hollnagel, E. (2012). FRAM: The Functional Resonance Analysis Method. Ashgate.
• Leveson, N.G. (2011). Engineering a Safer World: Systems Thinking Applied to Safety. MIT Press.
• Malatji, M., Von Solms, S. and Marnewick, A. (2019). Socio-technical systems cybersecurity framework. Information & Computer Security, 27(2), 233–272. DOI: 10.1108/ICS-03-2018-0031
• Rasmussen, J. (1997). Risk management in a dynamic society: a modelling problem. Safety Science, 27(2–3), 183–213. DOI: 10.1016/S0925-7535(97)00052-0
• Underwood, P. and Waterson, P. (2014). Systems thinking, the Swiss Cheese Model and accident analysis:  A comparative systemic analysis of the Grayrigg train derailment using the ATSB, AcciMap and STAMP models. Accident Analysis & Prevention, 68, 75–94. DOI: 10.1016/j.aap.2013.07.027

Back to all resources