Do early warning systems work? It depends...

Every time Lloyds Register Foundation launches a report, it feels a bit like Christmas: tons of data, new perspectives, new angles. The launch of its latest report was no exception. The title might feel a little daunting for non-academics - ‘Effectiveness of disaster early warning systems in reducing the impact of extreme weather- and climate-related events: A systematic review '–but the ambition is clear: provide a synthesis of the collective evidence on early warning systems across the academic literature. 

For those of us working on setting up early warning systems, there is a core assumption that no one really questions: better warnings save lives and protect livelihoods. Much of the momentum behind initiatives like Early Warnings for All, CREWS, SOFF and REAP is built on that conviction. But what happens when we subject that assumption to a rigorous academic test?

The researchers initially identified around 25,000 studies related to early warnings, yet only 111 met the criteria for inclusion in the report. Of these, 83 focused on national meteorological and hydrological services and disaster management agencies, with just 17 focused on communities.

There are many individual nuggets from the different studies reviewed. Many relate to the stakeholders who need to be involved: the value of reaching women through informal and digital networks, or how using trusted community intermediaries and local languages increase early warning system effectiveness. Others relate to the best ways to disseminate warnings, at times questioning the growing reliance on technology: rural loudspeaker warnings for typhoons were significantly associated with fewer injuries and illnesses. The targeting of (impact-based) warning also proves helpful, whether providing sector-specific advice for farmers, fishers and livestock owners; or connecting heat warnings with practical information about cooling and other locally available resources. In other words, a technically excellent warning is not in itself sufficient to achieve an effective warning that leads to action.

But the overall message from the review is surprising: the evidence collected through this exhaustive process is much less conclusive than one might have expected. Livelihood impacts were examined in only one study, without statistical testing. Across studies examining mortality, morbidity, loss of life and livelihoods, the overall picture is far from definitive. This should not be read as evidence that early warning systems do not work. Instead, it is a reminder that we still have surprisingly little academic (or, in the terms of academics, “robust”) evidence about what works, for whom, where and under what conditions.

The study nevertheless identifies conclusions that are repeated across studies. For instance, context matters. There is no universally "best" early warning system. Effectiveness depends less on choosing a particular technology and more on how the system is designed, implemented and connected to the people expected to act on the warning.
 

Reading between the lines, the review also shows how much our understanding of what makes an early warning system effective has evolved. Historically, much of the emphasis was on forecasting: can we predict the hazard accurately, and can we do it early enough? Those questions remain fundamental, but this review reflects a much broader and more contemporary understanding of people-centred, multi-hazard early warning systems. What matters is not simply the quality of the forecast, but how the different parts of the system connect across the entire warning-to-action value chain.

The report clearly identifies that across the literature, preparedness and response capability (i.e., knowing what to do when a warning arrives) emerged as the component most consistently associated with better outcomes. However, as practitioners know, this should not be interpreted as saying that anticipatory action protocols are more important than strong forecasts, but rather that the last mile uptake of an otherwise effective early warning system is what will ultimately save lives. 

Ultimately, people need not only information, but also the resources and capacities that allow them to act on that information before disaster strikes. This broader understanding is increasingly reflected in how the early warning early action community thinks about effectiveness and sustainability. Different expert groups, such as the Early Warnings for All Monitoring and Evaluation Group, have been focusing on the underlying conditions required for sustained investment in early warning systems.

In 2025, for the first time, the Global Status Report on Multi-Hazard Early Warning Systems looked beyond the four traditional early warning systems pillars to consider five cross-cutting enablers: governance, coordination, communication, finance, and monitoring and evaluation. The systematic review shows that this broadening of the definition of what constitutes an “early warning ecosystem” is starting to be reflected in the literature. This is clearly outlined in the review’s Finding 1: “weak governance, coordination, institutional capacity and sustainable resourcing constrained the implementation and long-term sustainability of early warning systems”. 

Finally, I was pleased to see the review’s authors focus on four key components at the heart of effective early warning: the timeliness of alerts, the accuracy of hazard prediction, public awareness and education, and technological advancement. Across the literature reviewed, early warning systems incorporating these components showed a strong positive influence on reducing loss of life during disasters.

All these different pieces need to come together to shape a more comprehensive understanding of the sector. Better forecasting and better technology remain essential, but they are only part of the story. Effective early warning systems depend on connecting those technical capabilities with awareness, communication, institutions, coordination and sustainable resources. The challenge ahead will be to design research approaches that can look at all of these different aspects. With growing political momentum and academic interest in the Global South, it will be interesting to see what progress a similar review might show in a few years’ time.