This article contributes to the theoretical development of understanding lessons learned from a crisis by integrating pragmatist theory into the conceptualization of lessons learned during crises. While pragmatism has increasingly informed research on health systems resilience (e.g., Ansell, Ansell Sørensen & Torfing, 2021; Ansell & Boin 2019; Ansell & Greyer, 2016; Greenhalgh & Engebretsen, 2022), its core insights have not yet been fully incorporated into how lessons learned are defined, framed, and operationalized within crisis models. In particular, the dynamic, contested, and temporally unstable nature of lessons learned drawn from crises remains under theorized. This article addresses that gap by drawing on pragmatist theory to conceptualize lessons learned as contextually emerging, continuously evolving, and shaped by the lived experiences and interpretations of decision makers. Importantly, the aim is not to analyze the learning process of individuals or a particular group (such as the national-level decision makers interviewed) but rather to examine how interpretations of lessons learned can shift across different stages of a crisis, and how these shifting interpretations can influence the ways in which lessons are ultimately implemented or neglected. The results of this article seek to provide empirical examples and further illustrations of what this means at the level of national decision making during a crisis, while the discussion section seeks to offer insights into the potential consequences for crisis preparedness and the further development of models that more effectively integrate lessons learned into evolving crisis frameworks.

In essence, pragmatist theory, rooted in the work of John Dewey (1896 [1996]; 1916 [1996]; 1938 [1996]), offers a relational and temporally sensitive understanding of lessons learned. Rather than viewing lessons learned as products of a discrete cognitive event, pragmatism allows their conceptualization as a transactional process between individuals and their ever-changing social and material environments. Lessons learned are not merely about acquiring knowledge but about navigating uncertainty, responding to disruptions, and reconfiguring practices considering new experiences. This perspective foregrounds the situatedness of lessons and its dependence on context, continuity, and collective meaning-making.

In the context of crisis governance, this theoretical lens provides an opportunity to move beyond linear or cyclical models of crisis response, which often treat lessons learned as isolated, retrospective insights. Instead, we approach lessons as interpretive constructs that are continuously negotiated, often politicized, and shaped by institutional memory and forgetting. As Greenhalgh and Engebretsen (2022) argue, implementing lessons learned in health systems is not only about technical adaptation but also about navigating conflicting values, interests, and narratives. This resonates with the idea that lessons learned are not neutral or universally agreed upon; they are contested, selective, and embedded in broader struggles over meaning and legitimacy.

Moreover, the pragmatist emphasis on temporal continuity challenges the assumption that lessons learned can be implemented only during or immediately after acute crisis phases. Lessons may emerge gradually, be reinterpreted over time, or fade from institutional memory altogether. The selective nature of remembering—shaped by what is politically salient, institutionally convenient, or socially acceptable—plays a crucial role in determining which lessons are retained and acted upon (Coraiola et al., 2023). In this sense, lessons learned are not only about what is known but also about what is remembered, forgotten, or deliberately sidelined.

Our theoretical position also aligns with scholarship on policy learning, which emphasizes the role of experience, analysis, and social interaction in shaping how knowledge is constructed and used by policy actors (Dunlop & Radaelli, 2018). Their definition of learning as “the updating of beliefs based on lived or witnessed experiences, analysis or social interaction” underscores the interpretive and non-linear nature of learning processes. Importantly, learning can be unintentional and may produce dysfunctional or counterproductive outcomes—what Dunlop and Radaelli (2018) refer to as “unwanted learning.” In our study, we adopt an open-ended approach to lessons learned, recognizing that not all learning leads to improvement and that some lessons may reinforce existing path dependencies or institutional blind spots (Heino et al., 2024).

Finally, our focus on national-level decision makers situates this study at the intersection of crisis governance and policy learning. By analyzing how these actors interpreted lessons at two distinct stages of the COVID-19 pandemic, we explore how the understanding of these lessons is shaped by evolving crisis dynamics, institutional constraints, and political considerations. This theoretical framing justifies our empirical delineation and supports the article’s broader aim: to deepen understanding of how lessons learned are framed during crises, and how these framings evolve over time in ways that influence both policy and practice.

The first round of analysis was conducted by author 2 and focused on identifying lessons learned, as interpreted by the participants at the two different stages of the crisis, that is, the two earlier defined time periods in which the data of this study was collected. In this phase, we took emerging tensions as a starting point for the analyses based on the pragmatist view that they constituted potential turning points in which old routines broke down and that the need arose for an evaluation or change in habitual activities (Elkjaer, 2008; Miettinen, 2000). Such turning points can be interpreted to potentially produce lessons learned because they may redirect courses of action (Ibid.). This analytical starting point and focus on tensions as potential markers for lessons learned is typical and characteristic for operationalization in many studies applying pragmatism (Ibid.), and it can be said to influence how lessons learned is approached, defined, and operationalized for the purpose of coding of qualitative interview data. An understanding of lessons learned as something that potentially breaks down old routines and as responses to needs to change habitual activities further allows a shift from seeing lessons learned not only as something that take place at the recovery stage of a crisis (e.g., Thomas et al., 2020), but as evolving outcomes of continuous processes and adaptive transactions with changing circumstances (e.g., Ansell & Boin, 2016; 2019; Greenhalgh and Engebretsen, 2022).

In accordance, leaning on our pragmatist theoretical background, tensions were defined as instances that bring about the need to question prevalent ways of thinking and/or acting. Based on these theoretical underpinnings and conceptual definitions, the analysis began as a process of open coding, where the core aim was to identify key code categories around how the lessons learned were interpreted, with an acknowledgement of similarities and differences between the two interview datasets collected in 2021 and 2022–2023. Notably, instead of an analysis of the nature of different kinds of learning processes during the COVID-19 crisis, we directed our analytical focus to the learning outcomes (i.e., lessons learned) that were reached after the emergence and based on the tensions. In this phase, then, lessons learned were interpreted and traced across the data as solutions to tensions as defined above; solutions referring here to novel forms of understanding; decisions, and/or conclusions that informed reorganization of action, or transformed structures of the health system. In sum, this phase produced knowledge of what the lessons learned were, as interpreted by the interviewees (see Figure 1).

In the second phase, author 1 focused on analyzing the way in which the implementation of the identified lessons and long-term development needs were discussed by the interviewees in the two datasets. The aim was to support analyses conducted in phase 1 and to produce complementary knowledge of how and the extent to which the implementation of the learning outcomes (i.e., lessons learned) was described by the participants and how the understanding of implementation was interpreted at the two time points (in 2021 and 2022–2023). In addition, the long-term development needs discussed by the participants were coded, with particular attention paid to their level of specificity (e.g., what, how, when, and who should act in relation to the referenced development needs). In sum, this phase of the analyses produced complementary knowledge of the meanings given to the lessons learned by the interviewees (see Figure 1).

Finally, in the third phase, a thematical synthesis of the open coding from phases 1 and 2 was conducted collaboratively by authors 1 and 2. This aimed to deepen our interpretations of the evolving nature of the lessons learned through comparisons of similarities and differences between the two datasets (Ryan & Bernard, 2003). Based on this final integrative phase, the following three interpretive themes were constructed: 1) lessons learned as contested and politicized, 2) lessons learned as temporally impractical and easily forgotten, and 3) lessons learned through a dichotomy of crisis and normalcy. These three themes summarize the central ways in which the lessons learned were interpreted across the two different points in time of the analyzed data. In sum, the three interpretive themes produced knowledge of how the lessons learned evolved over time (see Figure 1). The following results section is structured according to these three interpretive themes.

The three interpretive themes produced in the analyses are summarized in Table 1, providing an overview of the key results of this study and illustrating how lessons learned from a pandemic such as COVID-19 cannot be taken as a self-evident or simple process. The interpretation and framing of lessons learned does not take place only in a technical or rational manner but is also a social and political process. Accordingly, national-level decision makers’ interpretations of lessons learned demonstrate reconciliation of competing viewpoints, the formation (and perishableness) of collective memory as a product of historical reality and current politics, and the balancing act between points of normalcy and crisis.

In what follows, each of these three interpretive themes are presented in more detail with extracts from data to provide empirical illustrations.

The contested and politicized nature of the lessons learned was illustrated in the data through tensions emerging between knowledge production and decision-making, the interviewees’ tendency to focus on the successfulness of the measures taken, their experiences of the severity of the pandemic having been undermined after the acute stage of the crisis, and overall critical perspectives to broad-scale external evaluations.

The tensions between knowledge production and decision-making were evident especially in the interviews during the acute stage, which denotes a time of heightened uncertainty. Some interviewees (e.g., 2, 16, 29) had experienced the diversity of expert voices or the reconciling of a variety of scientific interpretations with political decision-making as challenging, while others stated that their experience and frustration over decision-making had become politicized, with expert views having been bypassed during the pandemic (e.g., 2, 3_2, 7, 8, 11, 28_2). In particular, the interviewees from the central government (23, 13) emphasized the importance of political decision-making in managing crises affecting the whole society. Others problematized political influence, for example, because of the lobbying power of industry in politics, which can result in resolutions in which some voices are marginalized:

Notably, the interviewees’ considerations of political influence in terms of lessons learned were not as widespread as in the follow-up interviews (2022–2023) as they were in the interviews conducted during the acute stage of the pandemic (2021). In contrast, the follow-up interviews emphasized a stronger tendency to focus on the successfulness of crisis measures:

This discrepancy between the two interview sets can be interpreted as a desire to conclude discussions of the crisis or an unwillingness to question decisions made during the acute stage, as other issues were already in focus at the time of the follow-up interviews. This was also illustrated as decreased questioning regarding political steering, which could also be connected to the fact that major restrictive measures leading to broad societal discussions had already been dismantled by the time of the follow-up interviews. Furthermore, the data included examples where the interviewees wanted to underline that the decisions made during the pandemic should always be interpreted situationally—that is, based on an understanding of the contextual and temporal situation in which the decisions were made—and that any criticism of these decisions should be made through the vocabulary of development needs rather than highlighting potential failures:

Paired with the imprecise nature of many of the development ideas, this, too, can be interpreted as enforcing the tendency to focus on the success stories of pandemic management. As referred, some of the participants (e.g., 18_2) maintained that the severity of the pandemic had been understated and downplayed in discussions after the acute stage. The follow-up interview data also revealed that the political time frame for analyzing the lessons learned was ending, and some of the interviewees were of the mindset that the most important lessons had already been learned and that it was now time to move on to other more pressing issues:

The contested and political nature of the lessons learned was further illustrated in how the interviewees deliberated on the evaluative measures. By the time of the follow-up interviews, many organizations had already carried out internal evaluations or launched development projects based on lessons learned. While some interviewees highlighted the importance of broad-scale external evaluations because of the tendency of civil servants to always explain things in a positive light (21_2), others spoke of discomfort about external evaluations of measures taken during the crisis and stated that this would not be necessary because of existing internal evaluations (3_2). The interviewees also revealed that evaluations of measures taken during the pandemic had already become highly politicized, making them challenging to execute (11_2):

When an external evaluation was seen as important and worthwhile, emphasis was often placed on the consequences of restrictive measures, as this was seen as beneficial for preparing for future crises (2, 9_2, 18_2, 23_2, 28_2).

The temporally impractical and easily forgotten tendencies of lessons learned were illustrated in the data through the notion that there never seemed to be “a right time” for their implementation. Many interviewees (8_2, 16_2, 26, 26_2) further highlighted that both the collective experiences of a crisis and lessons learned from it were easily erased from memory. Attempts at managing continuity in terms of preparedness were seen as a counterforce to these tendencies, while scant resources and the impracticality of preparedness plans were highlighted as hindrances to implementing lessons learned in practice.

At the follow-up interviews (2022–2023), several participants (13_2, 14_2, 21_2, 26_2) estimated that recovery from COVID-19 pandemic had not yet begun; that organizations and people working at different levels of the health system were experiencing ongoing stress because of other crises and challenges that had followed COVID-19 pandemic; and that an eventual recovery would take years. These issues are noteworthy, since many crisis models (e.g., Thomas et al., 2020) see the collection and implementation of lessons learned at the acute stage of a crisis as challenging and, consequently, define the recovery from a crisis as particularly important from the perspective of the analysis and implementation of lessons. Against this backdrop, the data represented lessons learned as temporally impractical: Their analysis and implementation were still undetermined, even elusive. Alternatively, from the perspective of the above-exemplified political reasons, it was probably already over.

While the general importance of analyzing and implementing lessons learned was widely acknowledged among the participants (e.g., 17), the inability to see the permanence of their effects in the now was also highlighted and as referred, their nature to be easily forgotten after a crisis:

Overall, the development ideas represented in the data were typically not stated very concretely or clearly (i.e., the specific ways in which the referenced issues should be changed; who should oversee the change process; or how the development work should be funded, implemented, and assessed). There were also some exceptions to this. For example, telehealth solutions (e.g., remote consultation) representing a concrete change process that was regarded as having been expedited by COVID-19, with possible long-reaching consequences (8_2). Generally, limited resources and outsourcing of professionals were often experienced by the interviewees as connected to the lack of implementation of lessons learned. The tendency to forget was paired with scant resources by some of the interviewees:

Complementarily, scant resources were seen as a significant threat and a future crisis scenario for the health system (23).

In the data, the implementation of lessons learned was also deliberated from the perspective of managing continuity (13). This was discussed, for example, through developing preparedness plans and the need for more frequent and hands-on practice (14) as well as active monitoring and evaluation of plans (23). The notion of inadequate preparedness plans was also illustrated through descriptions of the plans being of little use in practice during crisis management (14). Still, others saw a carefully and more holistically executed crisis management plan as important in terms of directing resources (11_2). The courage to analyze not only well-functioning models but also unpleasant crisis events was regarded as a potential tool for supporting continuity and looking at crises and shocks in a broader timeframe (14_2).

As the third and final interpretive theme resulting based on analyses, lessons learned were often interpreted in the data through a crisis–normalcy dichotomy; a crisis operation mode appearing at times as something to be restored back to a normal state as fast as possible. Furthermore, many of the lessons learned and their implementation could be characterized in light of the data as short-term and temporary, while only a few themes could be interpreted as potentially including long-term effects. Crises were also repeatedly described as inevitably unpredictable, hence, impossible to fully prepare for in advance.

At the time of the follow-up interviews (December 2022 to January 2023), most of the interviewees expressed that COVID-19 was no longer the most pressing issue on their work agenda. By that time, COVID-19 pandemic had been followed by and become entangled with other crises, such as the Russian war in Ukraine, challenges concerning the economy and sustainability, and the extensive health and social sector reform in Finland. Nevertheless, many of the interviewees still experienced COVID-19 as a crisis that could not yet be said to have ended. Therefore, the transition from “crisis mode” to “normalcy” could not be interpreted in the data as having fully taken place, nor could it be described as a clearcut or swift process. Rather, it seemed to be taking place gradually and during a vaguely identifiable time frame. Although interpretations of “crisis” and “normalcy” could be said to be illustrated in the data as overlapping and intertwined, the said dichotomy was often the starting point of the interviewees’ reflections of the lessons learned.

From the perspective of a “crisis mode”, crisis action measures were repeatedly referred to as issues that were impossible to decide on in advance and, therefore, inevitably something that must at least partly be dealt with reactively and in the moment, particularly in the interviews during the acute stage. In these deliberations, the interviewees (e.g., 12, 18, 25) often highlighted the importance of basic preparedness during normal times and the impossibility of preparing for every crisis beforehand based on preparedness plans or prescribed crisis operation models. As illustrated below, this highlighted the importance of leadership (12), which in the broader context of the data can relate to such themes as clear mandates, responsibilities, jurisdiction, and role division:

Notably, lessons learned interpreted through the crisis–normalcy dichotomy were not seen by the participants in identical ways. To exemplify, the participants reflected in varying ways on how crisis leadership should be organized compared to leadership during normal times: Some participants believed crisis leadership should be built based on models of normal times (3_2, 22_2, 8), while others stated that crisis responses can effectively lean on distinct leadership models. Decentralized governance in Finland represented another example of a tension between normal times and crisis responses; this was seen as a positive issue during normal times but as a challenge from the perspective of crisis leadership (13). Furthermore, scant resources emerged in the data as a challenge also across the normal–crisis continuum: A highly resource-efficient system was seen as incapable of offering adequate flexibility and a buffer effect in times of crisis (22_2). Concerns regarding scant resources did not appear to have faded at the time of the follow-up interviews but constituted points of reflection in the follow-up data through examples such as intensive care unit capacity and lack of human resources throughout the health system. Nevertheless, the interviewees generally did not provide very concrete deliberations as to what should be done to solve the challenge of limited resources.

Interestingly, comparisons between the two datasets showcased the deconstruction of some of the lessons learned and implemented. In other words, many of the key measures taken at the acute stage appeared in the data as temporary, as they had already been reversed or dismantled by the time of the follow-up interviews. The most frequently appearing example of this was the dismantling of structures concerning collaboration and communication (14_2):

Notably, many of the interviewees reported having experienced challenges in reference to collaboration and interaction between different levels of the health system but saw these challenges as potentially solvable through the extensive social and health system reform implemented from the beginning of 2023 onward (8_2). However, the realization of these expectations was not yet fully predictable or easy to evaluate at the time of the second data collection round.

Finally, although many of the crisis measures based on lessons learned were both implemented and deconstructed during the timespan of this study, there were also examples that could be interpreted as potentially having more long-term effects. In other words, they could have had impacts that would traverse the transition from “crisis” to “normal” mode. Some of these examples included the development of legislation and preparedness plans, with some interviewees pointing out that they had experienced that the legislation and responsibilities of different actors had not been followed due to a complicated steering system, despite being clear to the interviewees themselves (11).

Previous research has highlighted how power and politics shape outcomes and processes across all levels of health systems (e.g., Kihlström et al., 2023a; see also Bozorgmehr, Zick & Hecker, 2022). In line with this, our first interpretive theme (“Lessons learned as contested and politicized”) reveals how interpretations of lessons learned were deeply intertwined with competing perspectives, political decision-making, and questions of authority and influence.

While earlier studies have emphasized the importance of leaders taking responsibility for crisis outcomes (e.g., Greenhalgh & Engebretsen, 2022; Kapucu & Ustun, 2017; Christensen et al., 2016), our interviewees predominantly focused on the perceived success of measures implemented during the acute phase of the pandemic. This emphasis was accompanied by reflections on how the severity of the pandemic was downplayed in the post-acute phase, and by tensions between knowledge production and decision-making under conditions of uncertainty. Interviewees also expressed critical views toward broad-scale external evaluations, which were often seen as redundant due to existing internal self-evaluation mechanisms.

However, it is not self-evident that internal and external evaluations address the same concerns. This distinction underscores the inherently political nature of evaluation processes and their influence on how lessons are interpreted, remembered, or dismissed. National reports in Finland (e.g., Stenvall et al., 2022) have similarly pointed to the limited extent to which failures and accountability have been addressed, which is an omission that poses a significant barrier to identifying meaningful lessons and fostering development.

Moreover, a strong focus on successful responses can reinforce the perceived adequacy of existing preparedness systems, thereby reducing the perceived need for critical review or innovation. These dynamic risks entrenching vulnerabilities and obstructing adaptive responses to future threats (Heino & Huotari, 2022). In alignment with Perghel and Psychogios (2013), we argue that despite their challenges, broad-ranging external evaluations are essential for ensuring that diverse voices are heard—including those that were marginalized or silenced during the acute phase of the crisis.

Finally, improving how information is processed during crises, and clarifying who holds responsibility for communication and decision-making, are vital steps toward addressing power imbalances in crisis governance. These measures are crucial not only for fostering transparency and trust but also for enabling more inclusive and reflective learning processes.

The second interpretive theme (“Lessons learned as temporally impractical and easily forgotten”) illustrated how the appropriate timing of implementing lessons learned was elusive, contributing to their tendency to fade from memory. Interviewees described how limited resources and the impracticality of preparedness plans, particularly the low-level nature of hands-on rehearsals and the lack of concrete implementation measures, hindered the translation of lessons into practice. This finding resonates with previous research suggesting that rigid hierarchies and strictly defined roles, when paired with full compliance to preparedness plans, may inhibit the reinterpretation of lessons in novel ways (Dunlop & Radaelli, 2018). As Heino and Huotari (2022) argue, preparedness should be approached as a reflective practice, in other words, a process that allows space for doubt and the exploration of unimagined possibilities. This perspective aligns with Dunlop and Radaelli’s (2018) call to acknowledge not only successful outcomes but also unwanted or dysfunctional ones, thereby avoiding the pitfalls of crisis narratives that focus exclusively on success while neglecting critical analysis of failures.

Our findings also support prior research on collective memory, particularly the work of Coraiola et al. (2023), which emphasizes the selective nature of remembering and the contested terrain of what is retained or forgotten. In line with their argument, we propose that future research should more seriously engage with how memory is constructed and used, especially in relation to the politics of recognition and the creation of memorable experiences.

Additionally, our results underscore the importance of basic preparedness considering the inherent unpredictability of crises (see also Karreinen et al., 2023b). Since crises often involve unknown characteristics that defy full anticipation, we advocate for a complementary focus in preparedness training—particularly for leaders—on how to act, build trust, and collaborate in the face of uncertainty. As the COVID-19 pandemic has reshaped understandings of what constitutes a crisis (e.g., Boin, Ekengren & Rhinard, 2020), we argue that definitions and practices related to collecting, analyzing, and implementing lessons learned must also evolve. Specifically, crisis models and preparedness plans should acknowledge the prolonged nature of recovery following long-lasting crises and recognize that the recovery phase is not the sole moment for learning. Instead, lessons may emerge and require implementation across multiple stages of a crisis, reinforcing the need for temporally sensitive and adaptive learning frameworks.

The third interpretive theme (“Lessons learned through a dichotomy of crisis and normalcy”) highlighted how the framing of lessons learned was shaped by a perceived divide between crisis and normalcy. When lessons are (forgotten or) implemented in fragmented or gradual ways, preparedness becomes increasingly complex and difficult to manage (Heino & Huotari, 2022). Interviewees emphasized the importance of managing continuity, defined as the ongoing cycle of planning, implementation, and practical rehearsal, as a counterforce to the tendency for lessons to fade over time.

As noted also by Kihlström et al. (2023b), our findings underscore the significance of context-dependent and emergent leadership during crises, as well as the role of collaborative networks in effective crisis management. While collaborative structures were widely acknowledged as essential, most of these networks had been dismantled by the time of the follow-up interviews. This raises important questions about whether such networks should remain temporary constructs or be institutionalized within health systems to support long-term resilience.

In particular, the continuity of cross-sector collaboration demands reflexivity and ongoing adaptation to maintain the relevance and functionality of established partnerships (e.g., Högberg, 2024). Our data also revealed divergent views on appropriate leadership models across the crisis–normalcy continuum, as well as mixed assessments of the benefits and limitations of decentralized systems in this context. These findings suggest that comparative analyses of leadership practices during crises and in times of normalcy could offer valuable insights for future research. Finally, the generally unspecified nature of development ideas expressed by interviewees, combined with the persistent issue of limited resources, underlines the need to translate lessons learned into clearly targeted action plans. These plans should include well-defined responsibilities and mechanisms for accountability to ensure that lessons are not only identified but also effectively implemented.

This study has specific limitations. While having high-level and/or elite health system leaders as informants can be considered a strength, it may also limit pluralism, as those with less power and influence remain absent from the data. Additionally, our data were constrained by recall bias. We acknowledge that many ideas and lessons learned may have already faded in the midst of the acute and post-acute stages of the pandemic, which could have been better covered with a study design using more frequent follow-up data collection points. Finally, it is important to note that data in this study consist of interviews collected at two separate timepoints during the COVID-19 pandemic, thus excluding opportunities to analyze participants’ on-site activities based on data like observational or video materials (e.g., Bechky, 2003). This is a central limitation and an important aspect to be considered in further research. Nevertheless, despite these limitations, our results shed light on the contested nature of lessons learned. This is particularly important during a time when COVID-19 is no longer considered an urgent issue on the political agenda.