Five Years On: Reflections on the Definition of Risk Project

It’s now about five years since phase two of the National Fire Chiefs Council’s Definition of Risk (DoR) project got underway. At the time, it marked a shift towards a more consistent way of thinking about risk across the sector.

There had already been work through NFCC to define what risk is in principle, centred on the now familiar idea of likelihood and consequence. The next step was to take that definition and turn it into something more practical: developing a set of methodologies that FRSs could actually apply.

Since then, those methodologies have been published, referenced in CRMPs and Assessment of Risk work, and used in different ways across the country. What’s been less visible is any real reflection on what has changed in practice, and how FRSs have taken the work forward.

For context, I led the development of the national risk methodologies (Phase 2 of the project) on behalf of ORH for NFCC, and presented the outputs to FRSs through a series of workshops. This isn’t intended as an inside account of the project itself, but more a reflection based on the published work and what I’ve seen and heard from services since.

What the Project Set Out To Do

At a high level, the aim was to provide a consistent and practical way of applying the definition of risk.

Building on the earlier work to define risk as a combination of likelihood and consequence, the focus of this phase was on developing methodologies that could move services beyond simply mapping where incidents had happened, towards a more structured understanding of why risk sits where it does.

Three main areas were developed:

• Domestic Dwelling Fires

• Road Traffic Collisions

• Other Building Fires

Each followed the same broad framework, but as the work developed it became clear that the way that framework needed to be applied varied quite significantly depending on the type of incident.

What Has Happened in Practice

Looking across published CRMPs and Assessment of Risk documents, there is clear evidence that the work has been used.

Some services have applied the dwelling fire methodology quite directly. Nottinghamshire FRS is a good example, using it to develop LSOA-level risk scores and bandings that align closely with the national approach.

Others have taken a different route. Surrey FRS references the NFCC methodologies but uses them alongside a wider set of local inputs. In London Fire Brigade, the DoR approach is being incorporated into a broader Assessment of Risk framework rather than used in isolation.

These are just examples from what is publicly available, but they give a reasonable sense of how the work has been taken forward.

In practice, the approach hasn’t been applied in a single, uniform way. Services have generally taken elements of the methodologies and adapted them to fit their own context, or used them as a starting point for further development. That feels like a fairly natural outcome given the differences between FRSs and the role that local context still plays.

One Framework, Different Risks

Looking across the three methodologies, one of the clearer messages is that while the overall framework is consistent, the risks themselves are not.

Domestic Dwelling Fire risk is largely driven by people and vulnerability. Road Traffic Collision risk is shaped much more by the characteristics of the road network and how it is used. Other Building Fire risk varies depending on the type and use of buildings.

That has important implications. A consistent framework is useful, but it doesn’t follow that the same analytical approach should be applied in the same way across all hazards. In practice, each methodology adapts to reflect the nature of the risk it is describing.

That flexibility is one of the more useful outcomes of the project, even if it wasn’t expressed in quite those terms at the time.

Domestic Dwelling Fires

The Domestic Dwelling Fire methodology is the most developed of the three and has seen the most use in practice.

A large part of its strength comes from the scale of the data. Working at a national level allows the analysis to draw on far more incidents than any individual service would have available, which is particularly valuable given the long-term reduction in fire numbers.

The modelling itself reflects well-established patterns. A wide range of potential factors were tested, with the strongest relationships linked to deprivation and broader measures of vulnerability. That aligns with prior research, but the value here was in demonstrating it through a consistent national dataset.

In practice, the methodology provides a structured way for FRSs to build an LSOA-level risk profile. For some services, particularly those that had not previously developed this type of analysis, that represents a meaningful step forward in how risk is understood and described.

From the workshops and subsequent conversations, one of the more practical outcomes was that it helped bring a level of consistency to how FRSs approached this type of analysis. Some were already doing detailed work in this space, while others had done relatively little. The methodology provided a way of bringing that more into line, without replacing what stronger analytical teams were already doing.

There are, however, some limitations that were recognised at the time.

The data used in the original work is now dated, and there is a clear case for refreshing it using more recent Census, IMD and incident data. The underlying relationships are unlikely to have changed significantly, but the inputs themselves have moved on. It’s also important to note that LSOA boundaries have changed as part of the 2021 Census, so these should be reflected when using more recent data.

The modelling of consequence is also more difficult, simply because serious outcomes are relatively rare and harder to analyse robustly at smaller geographies.

A separate but related point is the question of scale. The LSOA-level approach works well for strategic planning and response considerations, but it is not necessarily the right level for all use cases. For prevention activity, there is often a need to work at a more granular level, closer to individual properties and the people within them.

That isn’t a limitation of the methodology so much as a reminder that the way risk is modelled needs to reflect what it is being used for. In practice, many FRSs are already navigating that by combining area-based approaches with more targeted, property-level work.

Road Traffic Collisions

The approach to RTC risk is quite different, reflecting the nature of the problem.

Rather than focusing on people and demographics, the methodology looks at the road network itself. Roads are classified using factors such as class, type, speed and whether they are urban or rural, and then analysed to understand how likelihood and consequence vary across those combinations.

From an analytical perspective, it works well. The patterns that emerge are intuitive, and the outputs (particularly the mapping) provide a clear and consistent way of describing risk across an area.

In practice, many FRSs still rely heavily on historical incident data when describing RTC risk. The DoR methodology offers a way of adding structure and context to that, rather than replacing it. The fact that there is often a strong alignment between the modelled outputs and historical patterns can also provide a useful sense-check.

What is less visible is the extent to which the methodology has been adapted locally. Unlike Domestic Dwelling Fires, there is less evidence in the public domain of FRSs building on the national approach or tailoring it to their own context.

There are a few possible reasons for that. Responsibility for prevention sits largely with partner agencies, which may reduce the incentive to develop this type of analysis further. It may also reflect the relative complexity of the approach, or simply that it has been a lower priority compared to other areas.

While the methodology provides a solid framework, its role in practice seems to be more about supporting and structuring existing analysis rather than fundamentally changing it.

Other Building Fires

Other Building Fires are the most complex of the three areas, and that came through clearly in the work.

It is worth noting that this work was taken to draft stage for consultation rather than being finalised in the same way as the other methodologies. As a result, much of the reflection here is based on that draft work and on what has been seen in practice elsewhere.

This is not a single, consistent category of risk. It covers a wide range of building types, each with very different characteristics and potential consequences. A fire in a care home presents a very different risk to one in a warehouse or retail unit.

The methodology explored ways of breaking this down into more meaningful categories, which is clearly the right direction. However, defining those categories in a way that works both analytically and operationally is not straightforward. It is also complicated by data limitations, particularly around building stock and how it aligns with incident records.

As categories become more specific, the number of incidents in each group reduces, which affects the robustness of the analysis. At the same time, consequence becomes a more prominent factor, with impacts extending beyond the building itself to include wider community and economic effects.

In practice, this is an area where national analysis can provide a starting point, but local knowledge and context are likely to play a much larger role in how risk is ultimately assessed.

What Could Be Improved

Looking back, many of the areas for development were already recognised at the time, and remain relevant.

Refreshing the underlying data is the most obvious step, particularly for the Domestic Dwelling Fire methodology. More recent Census, IMD and incident data would strengthen the analysis, even if the overall relationships remain broadly consistent.

There is also more that could be done around validation. The methodologies are good at explaining past patterns, but there is value in understanding how well they perform when applied prospectively.

For RTCs, incorporating measures of exposure (such as traffic flow) would add another layer of insight, although doing that consistently across the country is not straightforward.

More broadly, the way the methodologies are applied will continue to evolve. The core framework of likelihood and consequence remains sound, but the way those elements are defined and used will vary depending on the hazard and the purpose of the analysis. In that sense, the flexibility that has emerged in practice is probably a strength.

Where This Leaves Things

Five years on, the DoR project has moved the sector forward, but not by creating a single, standardised approach.

Instead, it has provided a common framework and a set of methodologies that FRSs can draw on, adapt and build around.

That feels like a more realistic outcome. The work has helped to move the conversation beyond simple incident mapping towards a more structured understanding of risk, while still allowing for differences in local context and priorities.

Final Thoughts

One of the more interesting things in putting this together is that while the Definition of Risk work is widely referenced, there has been less discussion about how it is actually being used in reality, and what FRSs are trying to achieve with it.

Building a risk model is only part of the picture. The more important question is how that understanding of risk is used to inform decisions about prevention, protection and response, and how it fits into a service’s wider planning process.

From the conversations I have had more recently, many FRSs are working through those questions – not just how to apply a methodology, but how to use it in a way that genuinely supports decision-making.

That is where most of the value sits. In shaping the approach, sense-checking how the analysis is being used, and helping translate it into something that supports clear, confident decisions.

If you are working through similar questions, I’d be interested to hear how you have approached it, and where it has been most useful within your service.