1 Introduction

Our society is increasingly influenced by crises, many of which have long-term persistence and impact. Whereas earlier global crises were spanned decades, new crises have begun to emerge nearly every year. For example, we had hardly made it through the COVID-19 pandemic before we were hit with an energy crisis in the wake of Russia’s war against Ukraine. To prepare for future crises, it is essential to advance research on crisis response strategies (Puumalainen et al. 2023).

A crisis is a social construction based on an exogenous shock (Abolafia and Kilduff 1988), and this social construction allows powerful actors to direct attention toward themes related to the crisis. For example, crises generate well-defined market needs that help focus ecosystem innovation (Battaglia et al. 2021). In addition, crises typically mean resource constraints that undermine the viability of existing offerings and require new solutions (Corsini et al. 2021). Such foci of attention then align the efforts of innovation ecosystems (Lingens et al. 2021), meaning crises drive the generation of open, ecosystemic mechanisms for innovation (Brem et al. 2021; Lee and Trimi 2021; Liu et al. 2021). Hence, we define crisis innovation as an ecosystem-level process to meet the needs of—and overcome the resource constraints derived from—an exogenous shock.

Firms and nations struggle to develop new technologies, services, and business models to help overcome or adapt to these situations (Brem et al. 2020). While the crisis captures managerial attention, the ability of a firm to deal with a crisis can be reduced by managerial overconfidence, knowledge gaps, lack of independent thinking, or understanding of complexity (Bouncken et al. 2022). Management literature reflects these struggles in ever shorter socio-technical cycles, and much has been written about innovation in relation to specific crises (Paunov 2012). Indeed, innovation was so central to overcoming the pandemic that it yielded a vast amount of scholarly reflection (e.g., Emami et al. 2022; Di Minin et al. 2021; Barragán-Quintero et al. 2020). Earlier literature on innovation in times of crisis largely explored the impact of the 2008 financial crisis (Brem et al. 2020; Filippetti and Archibugi 2011); the economic downturn of the early 1990s (Di Minin et al. 2010) has also been subject to investigation, as have the climate crisis (Averina et al. 2022) and events in the wake of the fall of the Iron Curtain (Meyer-Krahmer 1992).

While innovation under crisis has become the focus of increasing scholarly inquiry, there is not yet a coherent literature stream on crisis innovation. This is unsurprising since research in this field happens in many diverse disciplines. For instance, earlier research indicates that a good relationship of trust and proximity in buyer–seller networks helps to address unexpected shocks (Kranton and Minehart 2001), but this research is not embedded in the innovation management field since the article was published in an economics journal. The dispersed nature of literature related to innovation under crisis means that contributions refer to specific innovation topics e.g., frugal innovation (Corsini et al. 2021) or financial innovation (Daly et al. 2019). However, theory is not consolidated around the concept of crisis innovation. Therefore, recent research does not appropriately leverage earlier efforts or learning from previous crises that could be applied to future crisis prevention and response (cf. Bruhn et al. 2023; Viardot et al. 2023).

The purpose of this article is therefore to connect the divergent streams of research on crisis innovation so that future work may better draw on these rich foundations. Hence, we conduct a systematic literature review of research to achieve a comprehensive overview of existing work and its interconnections. To this end, we employ topic modeling to identify and map the main strands of research. We then analyze the contributions of each strand and derive a conceptual framework with implications for theory and practice. With this approach, we aim to contribute to a better understanding of crisis events and effective responses to them.

2 Methodology

The study extends topic modelling with an interpretative analysis characteristic of systematic literature reviews (SLRs). The methodology for such an independent literature review needs to be clearly laid out and replicable by future studies (Kraus et al. 2022). We have conducted an SLR using topic modeling to group the findings and results of previous studies on crisis innovation and to provide an overview and classification of the literature in this field. Such algorithmic coding can be more objective than its manual counterpart and also allows for analyzing large sets of articles (Kraus et al. 2020a, b). We thus also extend SLR methodology by including an inductive coding approach (cf, Sauer and Seuring 2023).

2.1 Data sources and retrieval strategy

We used the Web of Science (2022) core collection database to extract articles related to crisis innovation. The use of appropriate databases is crucial for conducting a comprehensive and reliable literature review. The Web of Science (WoS) database was chosen as the main source of data for this study because of its relevance and reliability for the research topic. WoS is one of the most comprehensive and widely used citation databases in the academic community that covers over 13,610 journals across all disciplines (Singh et al. 2021; Falagas et al. 2008). While other databases are available, each has its own strengths and limitations. For example, Scopus has a broader coverage than WoS, but it includes more lower impact journals (Chadegani et al. 2013). In contrast, WoS is known for its rich content of leading and high-impact journals (Kulkarni et al. 2009). The WoS impact classification thus guarantees a higher minimum level of journals (Sauer and Seuring 2017). While combining WoS and Scopus has been recommended (Kraus et al. 2020a, b, 2022), this would thus lower the average quality of the sample. Google Scholar has a larger coverage than both WoS and Scopus, but it also includes non-peer-reviewed sources, such as books, dissertations, patents, and websites (Harzing and Alakangas 2016). To collect the most relevant articles in this field of research, we searched crisis innovation* in the documents’ topics (title, abstract, and keywords). Then, we limited the results by language = (“English”), document types = (“article or early access”), and Web of Science categories = (“business or management or economics”). The resulting sample contained 1773 articles from 1970 to 2022. The study aims to identify cross-disciplinary research on crisis innovation, and the sample was therefore not further screened for fit with a specific field or method. Since the search employed a single database, no duplicates were identified.

2.2 Data processing procedure

To find and analyze logical information from the collection of textual data in this field of research, we used the latent Dirichlet allocation (LDA) technique (Blei et al. 2003). Our research methodology predominantly employs Latent Dirichlet Allocation (LDA) for topic modeling, marking a clear divergence from the generative AI approach (cf. Burger et al. 2023). While LDA identifies patterns and provides topic distributions based on the existing literature, generative AI models like ChatGPT autonomously generate new content, summaries, or explanations. Our methodology maintains an accurate boundary from introducing such generated content, aiming instead for a precise analysis and synthesis of the established patterns in our dataset. In natural language processing and machine learning, a topic model is a statistical method for determining the “topics” in a dataset of documents. Topic modeling is an emerging quantitative method to analyze and extract logical information from a collection of textual data. It has recently been introduced in management research in the context of marketing (Mustak et al. 2021), innovation management (Lee and Kang 2018), open innovation (Bagheri et al. 2022), and management (Hannigan et al. 2019). LDA can uncover hidden semantic structures and topics in a large body of unstructured textual data using natural language processing, machine learning, and statistical algorithms (Blei 2012). While there are various toolkits available for extracting and analyzing topics such as the Stanford Topic Modeling Toolbox (Ramage et al. 2009), Gensim, MALLET (McCallum 2022), and R Packages(Roberts et al. 2019; Grün and Hornik 2011), some of these tools are specifically designed for use in systematic literature review processes. For example, MySLR is a web-based platform that utilizes LDA-based topic modeling for management studies (Ammirato et al. 2022). We used a four-step methodological process to identify and analyze the topics, which is explained in the following paragraphs.

Step 1. Text pre-processing: We used Python programming language v. 3.6.5 to build a topic model, using the article abstracts as an input for the text pre-processing, and employed several techniques to create a dictionary for topic modeling: In this step, we split all of the abstracts into sentences and the sentences into tokens (tokenization). Then, we removed words with fewer than three characters (stop words), deleted all punctuation and email addresses, and extended the stop word list to remove structural words from the abstract e.g., aim, purpose, study, framework, method (Blei 2012). Next, we built word bigrams, such as linking the words “COVID” and “pandemic” as covid_pandemic. We also used the lemmatization algorithm to change the words into their root form (e.g., “took” to “take,” “drawing” to “draw”) and used the word stemming algorithm to filter such words and reduce them to their base form (e.g., “strategic”/ “strategy” to “strategi”). Finally, we removed all terms that occurred fewer than five times across all documents or that appeared in more than 70 percent of records. All of these tasks were implemented using gensim v. 3.8.3 (Rehurek and Sojka 2010), NLTK v. 3.7, and spaCy v. 3.0.0. (Honnibal et al. 2020).

Step 2. Topic modeling: In this step, we created a dictionary containing all pre-processed words. Then, we built an LDA model using MALLET, a Java-based software package for statistical natural language processing written by McCallum (2022). A significant problem with the topic modeling technique and fitting the model is finding the optimal number of topics. To address this, we calculated the topic coherence score, which provides a convenient measure to judge how good a given topic model is by computing the degree of semantic similarity between high-scoring terms in the topic. Appendix A provides information about technical details of rendering topics with coherence score.

Step 3. Topic exploration: In this step, we worked to uncover the main topics in crisis innovation research. Using the pyLDAvis library v. 2.1.2 to interpret topics, we analyzed the nine topics identified in the previous step (Sievert and Shirley 2014). This package uses Jensen–Shannon divergence (JSD) to compute the centers of topics in the two-dimensional plane and then uses multidimensional scaling to calculate the distances between topics in two dimensions. Hence, JSD is a method for measuring the similarity of two probability distributions. (See Appendix B for technical aspects of the inter-topic distance map).

Additionally, we used the word_cloud library v. 1.8.1 to visualize topics and their words; the size of each word in a specific topic depends on how frequently it is used in that topic.

Step 4: Topic analysis: In contrast to prior literature reviews using topic modeling (Mustak et al. 2021; Bagheri et al. 2022), we add an interpretive analysis of the identified topics. Based on the identified topics, each article in the sample was classified according to its dominant topic. We then reviewed the articles for each topic to identify, summarize, and connect the most relevant contributions. We first read the abstracts of all the articles for each topic in order to select representative works. The represeantativeness of these articles was verified through a comparison with the topic words. The selected articles were then studied in detail and related to the other key contributions for the same topic.

3 Results

3.1 Descriptive statistics

We collected 1,773 total articles from the Web of Science database from 1970 to 2022. Figure 1 illustrates the growth in scientific articles on crisis innovation in the business, economics, and management literature. The analysis of the data highlights a clear trend in the number of articles on crisis innovation over time. From 1992 to 1999, the number of published articles has remained relatively constant, while a substantial increase has occurred from 2008 onwards. This trend has reached its peak in 2022 with a remarkable 328 articles published on this topic, indicating a consistent and growing interest in the area of crisis innovation.

Fig. 1
figure 1

Number of documents per year

Full size image

Table 1 shows the most cited journals in this field of research. With 1,803 citations and 32 articles, Research Policy is the most relevant source in the field of crisis innovation. Technological Forecasting and Social Change ranks second, with 1,296 citations and 46 published papers; the Journal of Business Research ranks third, with 1,108 citations and 49 published papers. The managerial relevance of crisis innovation is demonstrated through the presence of the practitioner-oriented journal Harvard Business Review in this list.

Table 1 Most cited journals in crisis innovation research
Full size table

3.2 Topic modeling and exploration

Using the topic modeling technique, we were able to uncover hidden semantic structures in the unstructured text that we collected. We found that the model with nine topics could demonstrate a broad set of crisis innovation topics and provide a meaningful setting for further analysis. The resulting taxonomy of crisis innovation is reported in Table 2, which lists the most relevant terms for each topic, along with the number of documents and sum of citations. The topics are different types of crises which are responded to by innovation, i.e., societal crisis, funding crisis, financial crisis, economic crisis, digitalization, transformation, political crisis, strategy crisis, and organizational crisis. Digitalization is perhaps not usually considered a crisis, but it is indeed an exogenous shock to the ecosystem which generates new needs and costs that are met by innovation. The taxonomy is illustrated by a word-cloud analysis of the topic modeling results, presented in Fig. 2. In the word clouds, more frequent words are displayed in a larger font, and the illustration thus visualizes the weight of keywords within each topic. Together, these results highlight the variety of topics present in diverse academic disciplines.

Table 2 Taxonomy of crisis innovation
Full size table
Fig. 2
figure 2

Word clouds of each topic in the taxonomy of crisis innovation

Full size image

We also normalized the weight of each topic per year to analyze the annual changes in crisis innovation research. As Fig. 3 shows, the publication proportion for all the topics fluctuates from 1970 to 2004. Afterward, all the topics experience relatively smooth growth. In the early years, each topic has several amplitudes due to the limited number of studies. However, over time, there is a consistent and growing interest in the area of crisis innovation, resulting in more stable pathways for each topic.

Fig. 3
figure 3

Research on each crisis innovation topic over time

Full size image

To investigate the structure of topics, we built an inter-topic distance map. This exercise provides an overall view of the topics and how they relate to each other, allowing us to do a deep inspection of the terms highly associated with each individual topic. The topics that are closer on the map have more words in common. Figure 4 illustrates the results of the inter-topic distance plot, showing that the “transformation” and “digitalization” topics overlap. The overlap between these two topics can be explained by the fact that digitalization is often a key driver of transformation during times of crisis. Digital technologies and processes can enable organizations to quickly adapt to changing circumstances, whether that be shifting to remote work or changing business models. Moreover, the “strategy crisis” and “organizational crisis” topics are close to each other, indicating their conceptual closeness. This result implies that these two types of crises may have similar underlying causes and can be tackled through similar solutions. While strategy crisis focuses on developing new strategies to adapt to the crisis context, organizational crisis deals with internal crisis within the organization. Both topics highlight the need for firms to be resilient and adaptable in times of crisis, and to activate dynamic capabilities to improve innovation performance and increase evolutionary fitness. The topics “financial crisis” and “funding crisis” are also very close to each other. While funding crisis is centered on the firm level and relates to a sharp reduction of the funds available for innovation, financial crisis is systemic and refers to the financial aspects of a crisis. Both topics overlap in the sense that financing innovation activities depend on the health of the financial sector. By contrast, “political crisis,” “economic crisis,” and “societal crisis” are completely separate topics with distinct term distribution.

Fig. 4
figure 4

Inter-topic distance map

Full size image

3.3 Topic analysis

Each of the identified topics represents a main research area in the crisis innovation literature and the topics thus constitute a taxonomy that shapes the conceptual structure of this field. We therefore reviewed representative articles for each topic in depth. Below, each topic is first defined based on the keywords and the articles of the sample related to the topic. Then, the main findings are reported.

Societal crisis (topic 1) is related to the organization of innovation during society-wide crises such as the COVID-19 pandemic, and centers on forms of open innovation, innovation ecosystems, and other collaborative networks (Temiz and Broo 2020; Brem et al. 2021). For example, the mechanisms for the emergence of new ecosystems have been studied (Radziwon et al. 2022), as have the transformation and restructuring of existing ecosystems (Cosimato et al. 2022). In this context, the urgency rendered by crisis means that self-organized community-based ecosystems can be more efficient than industry- or government-driven initiatives (Dąbrowska et al. 2021). Enabling technologies, (e.g., synthetic biology) can also provide foundations for efficient ecosystem growth in times of crisis if structural, societal, and ethical barriers are overcome (Nylund et al. 2022). The motivations for innovation have also received attention during the pandemic. On the one hand, innovation was destined to save lives and improve health (Brem et al. 2021), including alleviating suffering i.e., through compassion venturing (Majchrzak and Shepherd 2021). On the other hand, innovation improved quality of life under COVID-imposed resource constraints (Brem et al. 2021). These motivations converge when life-saving solutions are needed under resource scarcity, inviting an extension of theory on frugal innovation beyond emerging markets (Corsini et al. 2021; Giones et al. 2020). The time-constrained nature of crisis innovation has also led to research on the timing and life cycle of such organizational forms. To this extent, researchers are considering the durability of initiatives after the peak of the crisis (Kronblad and Pregmark 2021).

Funding crisis (topic 2) relates to a sharp reduction of the funds available for innovation. In the realm of crisis innovation, this topic builds on Filippetti and Archibugi’s (2011) seminal contribution identifying the structural characteristics that foment innovation in times of crisis for European firms: stronger national systems of innovation with capacitated human resources, high-tech specialization, and a developed financial system. To this extent, triple-helix interactions involving university, industry, and government, are of increasing importance (Kashani and Roshani 2019). For Latin American firms, access to public funding has been identified as the most important factor (Paunov 2012). Funding is a bottleneck for innovation, although crisis does not hit innovative firms harder in this regard (Lee et al. 2015). When funding is scarce, turning previous profitability into slack resources is key for a firm to keep investing in innovation during crisis (Zona 2012). Therefore, firms that were already innovative before crises keep investing more, as do fast-growing new firms (Archibugi et al. 2013). For emerging markets, the gains from such investments are contingent on good management under crisis (Nemlioglu and Mallick 2021).

Financial crisis (topic 3) refers to the financial aspects of crisis. As a topic, financial crisis overlaps with funding crisis; however, whereas funding crisis is centered on the firm level, financial crisis appears to be systemic. Financial innovation to obscure risk was largely responsible for the 2008 financial crisis; this innovation’s riskiness and radicalness increased abnormal returns before and during this crisis, while its complexity decreased these returns (Schöler et al. 2014). The financing of a firm’s innovation activities depends on the financial sector’s health, which was restored after the 2008 financial crisis through bailouts and restructuring. Such efforts need to be coordinated on a global scale, since financial shocks tend to spread from one country to another in times of intense financial innovation (Daly et al. 2019). However, the new structures have been unable to overcome the inherent contradiction of the need for long-term commitment to innovation and the financial sector’s short-term horizon (Mazzucato 2013). Therefore, new ventures and sectors struggle during financial crisis. Then, startup funding focuses on a few core sectors and the most experienced venture capitalists in a market (Conti et al. 2019). Adding to this effect, firms that are the customers of firms relying on the interbank market may find it more difficult to finance innovation (Giebel and Kraft 2019).

Economic crisis (topic 4) involves macroeconomic crises, such as the recession of the early 1990s. From a macroeconomic perspective, the effects of an economic crisis can be partly overcome by investing in innovative enterprises and by economic policies stimulating entrepreneurial activity (Castaño et al. 2016). Whereas such policy impact was formerly spatially mediated and localized, the recent rapid pace of technological change and global economic integration have made policy development geographically interconnected (Crescenzi and Iammarino 2018). In this increasingly complex context, policymaking should aim for crisis prevention rather than palliative measures (Rizzi et al. 2018). However, public policy adapting to a crisis situation may also make industries more resilient and responsive to future crises (Unger 2000). For example, the European energy industry went through a digital transformation in response to the 2008 financial crisis (Midttun and Piccini 2017). The reformed institutional and industrial landscape has meant that energy providers are emerging strong from the current energy crisis—much to the dismay of customers suffering from the price hikes that keeps the industry in good health.

Digitalization (topic 5) considers the incorporation of digital technology into products, services, and processes. Digitalization in itself can constitute a crisis, as it has for newspapers, which were made obsolete by new media (Rothmann and Koch 2014). Other recent crises have also sparked an increase in digitalization generally—and digitalization-driven retail business model innovation in particular—by increasing the range of technologies as well as the quantity and quality of information employed, leading to a reframing of ecosystem partnerships and strategic alliances (Mostaghel et al. 2022). However, such crisis-driven digitalization is only sustainable in the long run if it avoids drawing exclusively on organizational elasticity and is accompanied by structural, processual, and cultural adaptation (Reuschl et al. 2022).

Transformation (topic 6) refers to such structural changes made by firms and institutions in order to adapt to crisis. External events not only alter the motivations and desired outcomes of innovation, but also the innovation dynamics and the technology life cycle. Empirical foundations are found in large-scale events, such as the fall of the Iron Curtain (Meyer-Krahmer 1992; Szántó 1994), as well as in regionally limited critical incidents (McAdam and Mitchell 2010). Ryan et al. (2022) discuss the collapse and subsequent transformation of global value chains and the corresponding restructuring of multinational enterprise governance. The flexibility required by crisis, coupled with the opportunities presented by digitalization, requires a transformation toward network structures and network economy rather than the prior hierarchical organizations (Voronkova et al. 2022). Digitalization and transformation’s overlapping and central position on the inter-topic distance map (Fig. 4) points to the salience of digital transformation as the lasting structural change that may occur as a result of digitalization.

Political crisis (topic 7) can be a motive for innovation, but innovation can also cause political crisis: for example, a sociopolitical legitimacy crisis is developing due to the regulatory voids generated by platform business model innovation. This means the firms need to shape business opportunity and the business model through a liminal movement in interaction with regulators and other ecosystem participants (Garud et al. 2022). During the pandemic, governments implemented draconian measures limiting the free movement of people and goods, which have led to a reevaluation of proximity and of globalization (Dosi and Soete 2022). In times of crisis, mechanisms of globalization meet with many obstacles—but also become more important for innovation diffusion, such as through the emergence of dominant designs (Brem et al. 2020). Regulatory interventions are however often identified as perpetuators of crisis rather than solutions. While the crisis of the US healthcare system could be remedied by disruptive innovations, for example, transformation of this highly regulated industry is hindered by the efforts of regulators, physicians, and pharmaceutical companies to preserve the existing systems (Christensen et al. 2000). At worst, regulatory responses to crisis can lay the groundwork for future adversity, as in the case of bailouts that allow for even larger financial markets to generate financial crises with greater impact on society (Crotty 2009).

Strategy crisis (topic 8) indicates that prior strategies are no longer adapted to the crisis context and that strategy needs to be developed in accordance with the new situation. Such new strategies can either focus on innovation, particularly business model innovation, perseverance through operational management, or retrenchment through controlled shutdowns (Kraus et al. 2020a, b). For example, open innovation strategies can protect firms’ innovation capabilities during resource cuts (Di Minin et al. 2010). Radical and exploratory innovation strategies are generally more effective during crisis and generate better firm performance and variability than exploratory strategies do (Osiyevskyy et al. 2020). However, firms can identify opportunities for exploitative innovation through the analysis of, for example, patent data (Guderian et al. 2021). Firms can endure crises while simultaneously taking advantage of exploratory efficiency by adopting a temporal separation logic and increasing risk-taking in times of crisis (Leppäaho and Ritala 2022). When firms want to generate breakthrough innovations in the absence of external crises, they can also induce critical incidents. Firms can then purposefully construct internal crises to intensify efforts toward organizational learning and innovation (Kim 1998).

Organizational crisis (topic 9) refers to an internal crisis of the firm or organization. While an organizational crisis can lead to firm demise, it can also yield entrepreneurial action that results in new products, processes, or business models, particularly in family firms (Soluk 2022). This action is supported by new alliances, digital platforms, and improved adaptive capacity (Soluk et al. 2021). To generate such a resilient response, firms generally need to activate the dynamic capabilities of sensing, seizing, and transforming on both the firm and ecosystem levels (Khurana et al. 2022). Then, dynamic capabilities lead to better innovation performance, which in turn increases evolutionary fitness (Makkonen et al. 2014; Newey and Zahra 2009). For example, committing to vigilance around threats and opportunities allows firms to act and adapt more quickly to crisis (Schoemaker and Day 2021).

4 Discussion

This discussion is organized as follows: First, we provide a summary of key insights for research. Based on this overview, we then situate the findings within the context of earlier literature and develop a conceptual framework. This serves as the basis for our discussion of managerial implications as well as for the limitations and recommendations for future research.

4.1 Implications for research

For innovation management, this article contributes to our collective understanding of innovation in times of crisis and adds to the literature on the emergence of innovation ecosystems (Giones et al. 2019; Thomas et al. 2022; Nylund et al. 2022). The definition of crisis innovation as an ecosystem-level process may aid both those studying innovation ecosystems under crisis (Brem et al. 2021; Lee and Trimi 2021; Liu et al. 2021), as well as those studying specific underlying processes that need to be placed in a larger context (Voronkova et al. 2022; Garud et al. 2022). The framing of the definition in terms of needs and resource constraints encompass varying motivations generated by a common underlying event and can thus help define different types of initiatives.

We also find that innovation as a crisis response differs according to the type of crisis. The taxonomy of crisis innovation presented in Table 2, delineates innovation as a response to societal crisis, funding crisis, financial crisis, economic crisis, digitalization, transformation, political crisis, strategy crisis, and organizational crisis. This division into different types of crisis innovation helps distinguishing between different strands of research, and also supports cross-disciplinary research since it suggests related lines of theoretical development. Specifically, the inter-topic distance map of Fig. 4 shows which topics are most closely related in the literature. Researchers can venture into an adjacent topic for richness of their analysis, and can also explore more distant topics to explore more complementary perspectives. Conscientious efforts can also be made to bring topics together that may be conceptually related, but that today have distant research approaches, e.g., societal and economic crisis. However, the taxonomy goes beyond a mere classification since it analyzes how the contributions on each topic have advanced the field of innovation management.

The topics of digitalization and transformation have a central position in the inter-topic distance map and influence research on several other topics. Resource scarcity, coupled with new needs, means crises drive digitalization (Midttun and Piccini 2017; Giones et al. 2020). While crises can prompt new resource allocation behaviors, digital tools enable the firm to convert resource preservation, recombination, or sharing into digital process innovation, product innovation, and business model innovation (Soluk 2022). For example, sharing of intellectual property prompted by crisis can enable digital fabrication that speeds up innovation and side-steps problematic supply chains (Corsini et al. 2021). Open innovation with suppliers is particularly to profit from innovation in such automated contexts (Nylund et al. 2020).

Digital resource sharing includes not only existing resources but also the co-creation of new knowledge and solutions (Cosimato et al. 2022). This co-creation is possible through ecosystemic sensing, seizing, and transforming (Khurana et al. 2022). Here, frugality on an individual and organizational level can contribute to long-term organizational resilience (Giones et al. 2020).

Digitalization-driven retail business model innovation has gained traction to mitigate the negative effects of the COVID-19 pandemic. The rise of digital services and retail platforms has enabled value co-creation with customers and platform owners, leading to models that are much more ecosystemic (Mostaghel et al. 2022). Platform-based business models (e.g., that of Uber) can deepen a sociopolitical legitimacy crisis due to discrepancies between the business model and regulatory frameworks (Garud et al. 2022). Thus, for platform models to lead to successful digital transformation, orchestrators and other actors need to actively engage in building ecosystem legitimacy (Thomas and Ritala 2022).

Kronblad and Pregmark (2021) found that fully digital business models (i.e., digital natives) have thrived in retail and service businesses before, during, and after crisis. Other firms have employed temporary business model innovation to adapt to crisis requirements (Clauss et al. 2022; Kraus et al. 2020a, b). However, firms that were forced toward digital business models under crisis (e.g., educators, restaurants, and retail businesses during COVID-19) have at least partly reverted to analog models (Kronblad and Pregmark 2021) because the new models stretched organizational elasticity without adapting structures, processes, and culture (Reuschl et al. 2022). Thus, much of the digitalization driven by COVID-19 has not been consolidated in the aftermath of the pandemic. However, in some cases (e.g., newspapers), digitalization eliminates the foundation for the previous business model and requires firms to engage in digital business model innovation or yield to digital natives (Rothmann and Koch 2014).

While digital transformation increases industry resilience and helps economies overcome crises, initiatives need to consider those hardest hits by crises. Digital illiteracy and the unaffordability of digital devices are obstacles to widespread innovation diffusion (Misati et al. 2022), as are concerns about data integrity and weakened democracy in light of the corporate power over digital platforms (Boersma et al. 2022; Wu 2018). On the other hand, the impact of compassion gendered by adverse events can be magnified by crowd-based digital innovations (Majchrzak and Shepherd 2021).

We thus contribute to the literature on innovation management in the context of digital transformation (de Paula et al. 2023; Danuso et al. 2022). Our findings are particularly relevant to the literature examining digital transformation across multiple levels, including that of ecosystems (Lanzolla et al. 2020, 2021; Dąbrowska et al. 2022; Furr et al. 2022). Table 3 summarizes the findings for innovation management and digital transformation for each topic of the taxonomy.

Table 3 Summary of findings for innovation management and digital transformation for each topic of the taxonomy of crisis innovation
Full size table

Figure 5 organizes the implications of the findings summarized in Table 3 in a conceptual model. Overall, crises create new needs on the level of the entire ecosystem (Battaglia et al. 2021). These needs must be satisfied under conditions of resource scarcity, especially in terms of time and funding as identified in topics 1, 2, and 8 (e.g., Corsini et al. 2021; Zona 2012; Di Minin et al. 2010). This yields crisis innovation, which due to the shared motivations for innovation in crisis, is also organized in sharing ecosystems with the need for co-creation as formulated in topics 1 and 9 (Cosimato et al. 2022; Khurana et al. 2022). Such sharing behaviors, in turn, drive the digitalization of topic 5 (Mostaghel et al. 2022). For digitalization to turn into digital transformation, there must be changes in policy, structure, processes, culture, business models, and legitimacy as exposed in topics 4, 6 and 7 (Unger 2000; Voronkova et al. 2022; Garud et al. 2022). However, the contradiction between long-term structure and short-term urgency is only temporarily overcome by crisis as mentioned in topic 3 (Mazzucato 2013). Due to this ephemeral increase in organizational elasticity, the digitalization that happens during crisis does not always prompt the structural changes required for digital transformation as expressed by in topic 5 (Reuschl et al. 2022). The conceptual model thus highlights specific and sequential mechanisms of crisis innovation. It allows researchers to situate specific contributions within a wider theoretical context, and therefore furthers our understanding of crisis innovation as a whole.

Fig. 5
figure 5

Conceptual model of crisis innovation

Full size image

4.2 Implications for practice

This conceptual framing offers also several starting points to explore managerial implications. A crisis event inevitably creates short-term changes to corporate management. New needs must be addressed in the short term, which calls for constraint-based solutions and flexible organizational answers. Additionally, as opposed to normal circumstances, crisis is a time when managers might need to rely even more on their gut feeling (Scheiner et al. 2015). In this regard, concepts like VUCA can help managers develop responses to the volatility, uncertainty, complexity, and ambiguity appearing in their industry. Since some sectors (e.g., energy) have been through such crisis-initiated events, learning can also be drawn from these sectors without reinventing the wheel (Giones et al. 2019). This also touches on the organizational culture, process, and structure, as well as the business model and overall legitimacy of corporate management.

Managers can also prepare for future crises by increasing the degree to which they engage in sharing behaviors, ecosystems, and co-creation. More flexible structures and business models will help the organization adapt more rapidly to crisis and facilitate the structural adaptation necessary for digitalization to yield sustained digital transformation. Thus, crisis innovation calls for active management of the tension between stability and evolvability (Tilson et al. 2010; Wareham et al. 2014). This requires rapid evolution and thus flexible prior structures. However, the perdurance of crisis innovation requires the generation of structures that bring stability to the ecosystem. Resilience is hence aided by diversification strategies e.g., geographical diversification, that generate structures allowing multiple responses to crisis (Caloghirou et al. 2022).

Additionally, it is important to distinguish between company size and industry. Large firms have many more resources but must follow long internal coordination processes, implying time constraints. By contrast, small and medium-sized companies can react much more quickly than large organizations can, but they typically face resource constraints. Entrepreneurial orientation let firms pivot to cope with such constraints (Puumalainen et al. 2023). For both small/medium and large enterprises, digitalization can help address these constraints to allow for the more rapid responses that are necessary in crisis situations.

4.3 Limitations and directions for future research

The presented research is not free from limitations, which creates potential avenues of exploration for future researchers. For example, our research has uncovered a relationship between crisis innovation and digital transformation. Future research can extend this by considering the impact of the next generation of enabling technologies, such as artificial intelligence (Ferras-Hernandez et al. 2022), or related managerial concepts, such as frugality (Giones et al. 2020). Another aspect that future research could consider is the role of tools that could be applied in crisis situations; earlier research suggests several, including hackathons (Yokoi 2023) and TRIZ (Münzberg et al. 2016). Moreover, future studies could collect and analyze quantitative data to identify patterns in crisis situations, which could help develop even better managerial responses to crises.

The methodology herein presented also has limitations. We read the abstracts of the 1,772 articles in the sample, and the capacity for this manual process sets the limits for the applicability of the methodology to larger literature streams. Future applications of artificial intelligence may handle also this step of the process to enable even larger literature reviews. Additionally, as with all literature reviews, this research may suffer from selection biases: Since a crisis is a social construction, we rely on authors to identify a crisis as such (Abolafia and Kilduff 1988). For instance, relatively little has been written on innovation in the context of climate crisis, but using wording such as “climate change” or “global warming” might uncover relevant work. Innovation-related findings may also be known by “R&D” or “technological change,” and while we have verified that such literature is represented in the sample, future research could consider additional related areas. However, while this article could have been expanded to related literature streams, the current delineation corresponds to our aim of representing and consolidating crisis innovation research.