A Multipath Development Framework for Inter-Organizational Relationships: A Metasynthesis of Qualitative Studies

Inter-organizational relationship (IOR) development is suited to the relational view of strategic management (Holm et al., 1999), explaining sustainable competitive advantage through cooperation, coopetition, and networking (Dyer & Singh, 1998). Inter-organizational relationships (IOR) are defined as contractual, long-term, relevant, and usually complex (Czakon, 2009) relationships between or among inter-independent organizations (Baptista 2013) engaged in mutual interactions addressing shared, strategic goals (Holm et al., 1999; Mandják et al., 2015). It is acknowledged that organizations with the capabilities to manage inter-organizational relationship development turn out to be better prepared to sustain relationship-related competitive advantage (Mitręga & Pfajfar, 2015). Importantly, even though there is increasing interest in IOR, this phenomenon still calls for deeper understanding (Alimadadi et al., 2019; Batonda & Perry, 2003; Gelei & Dobos, 2014; Jap & Anderson, 2007; Klimas et al., 2022).

Relationships are not static phenomena. Instead, they evolve dynamically throughout the relationship life cycle (Kusari et al., 2013) and the change in IOR is typical (Palmatier et al., 2013). Nonetheless, as claimed by Ferreira et al. (2017), the question still unanswered is how relationships do develop over time. This would appear to be crucial (Jap & Anderson, 2007), as awareness, evaluation, and handling the relationship development path are important for gaining optimum benefits from relationship maintenance (Davis & Love, 2011; Harmeling & Palmatier, 2019).

The literature still identifies a relevant deficit of knowledge on the course and specificity of the development path of IOR, as well as pointing at the fragmented state of the findings to date (Akrout, 2014; Batonda & Perry, 2003; Ferreira et al., 2017; Hastings et al., 2016). Indeed, although interest in the life cycles of relationships has been observed for decades (Ford, 1980; Dwyer et al., 1987; Ring & Van de Ven, 1994; Wilson, 1995) our knowledge remains mainly conceptual or – at the most – results from explorative and contextually limited empirical investigations. So far, the exploration of long-term business relationships has been given considerable interest, especially in the field of relationship marketing (Batonda & Perry, 2003; Akrout, 2014). Most scholars focus their attention on the buyer-supplier perspective (characterized by exchange: product–money), including the benefits gained from developing relationships with customers and suppliers (Dwyer et al., 1987; Meng, 2010; Lee & Johnsen, 2012). At the same time less attention has been given to the evolution of IOR relationships in dyads, networks, or ecosystems considered from a strategic management perspective, e.g., both cooperative and coopetitive relationships linking heterogenous partners. This constitutes an interesting and important research gap.

In particular, the existing knowledge does not adopt a holistic view (Bell et al., 2005), as the majority of prior research findings (Batonda & Perry, 2003) refers to selectively chosen, single phases of IOR development (Jap & Anderson, 2007) and focuses on single countries (e.g., Australia – Heffernan & Poole, 2004; Davis & Love, 2011; Hastings et al., 2016; China – Duanmu & Fai, 2007; the UK – Meng, 2010; Taiwan – Lee & Johnsen, 2012; Portugal – Baptista, 2013; France – Akrout, 2014; Hungary – Mandják et al., 2015; India – Panda and Dash, 2016; Brazil – de Almeida Moraes et al., 2017). Given the state of the art, there is a need for cumulative development of the comprehensive knowledge about the trajectory of IOR development. What is more, the existing stock of knowledge, and the existing pool of empirical studies in particular, build on seminal concepts which are not necessarily valid and appropriate nowadays. It is worth noting that the vast majority of prior works referring to life cycles of inter-organizational relationships (Batonda & Perry, 2003) adopt models based on stage theory (e.g., Ford, 1980; Dwyer et al., 1987; Wilson, 1995). Those models, however, assume that every relationship development occurs in a sequential, incremental, predictable, and irreversible set of stages. This in our opinion does not describe modern, highly dynamic, and networked reality. Indeed, the surrounding business environment is multidimensionally different from this 1980 picture. It is thus desirable to consider if any prior conceptual assumptions should be sustained, and if so, how to do so. We do agree with Ferreira et al. (2017), who claim that the question is how relationships develop over time in such a renewed dynamic context.

In this paper we address the following cognitive gaps: the scarcity of empirical findings on inter-organizational relationships (Jap & Ganesan, 2000), which makes it difficult (although possible) to run comparable studies and develop knowledge; then the deficit of knowledge on the course and specificity of the relationship life cycle (Hastings et al., 2016; Ferreira et al., 2017; Meng, 2010). Further gaps are found around the following: the issue of relationship renewal versus definitive termination (Ferreira et al., 2017), intense focus on only one specific phase of relationship life cycles, a concentration on a single industry or country, the adoption of a static view using only one functional background (Ford, 1980; Dwyer et al., 1987; Ring & Van de Ven, 1994; Wilson, 1995), and the lack of comprehensive studies on cooperative relationships.

Therefore this paper aims at the identification and comparison of conceptual models of IOR development across qualitative studies. To do so it uses the results of systematic (Denyer & Tranfield, 2009), integrative (Elsbach & van Knippenberg, 2020), and qualitative metasynthesis (Erwin et al., 2011) of the literature on inter-organizational relationship life cycles (including also works on their dynamics, development (stages, phase, states, steps), evolution, etc.). As a result this study develops an original, multipath, situational framework for the development of inter-organizational relationships over time. We claim that as the developed framework integrates exploratory and contextually different qualitative outputs (Andriopoulos & Slater, 2013) it allows us to make more general (Durach et al., 2017; Okoli, 2015) and comprehensive conclusions (Finfgeld-Connett, 2010) than contextually limited qualitative studies available in the existing literature. As metasynthesis focuses on evidence from the field research (e.g., here the evidence from different case studies) the conclusions drawn as well as the developed multipath model should be seen as practice-based (Denyer & Tranfield, 2009; Erwin et al., 2011), but still prone to subjectivity bias (Lachal et al., 2017).

Our desk research aims also to present the original, situational, multipath development framework for the inter-organizational relationship. The developed framework covering 4 phases and 11 sub-phases builds on 18 previous qualitative studies on IOR life cycle; thus, it should be seen as integrating dispersed and so far not generalizable qualitative findings into a context-free approach (Erwin et al. 2011; Garip & Kamal, 2019; Melendez-Torres et al., 2015; Sandelowski et al., 1997; Zimmer, 2006). In this paper we point to gaps in the field of knowledge deficit of development paths of long-term IOR: (1) the course and specificity of relationship life cycle; (2) number of phases of business relationships; (3) relationships development over time.

In our reviewing process, we followed a systematic approach to the literature review and used a metasynthesis approach being relatively new in management sciences Zhao et al., 2021).¹ Our review shows that prior qualitative findings focused on IOR sensu largo and sensu stricto are scant and lack consistency, thus calling for summary, synthesis, and integration (Bell et al., 2005). The above shortcomings result from differences in the adopted theoretical backgrounds (i.e., theoretical perspectives, conceptualizations, definitions, and cognitive scope of relationships development), lack of sound methodological approaches (i.e., no works providing verified and valid operationalization or measurement), and finally differences concerning the number and specificity of IOR life cycle phases.

All of that makes it impossible to conduct coherent research, and thus to draw general conclusions, provide (even contextually) generalizable results (Finfgeld-Connett, 2010), and finally to cumulatively develop knowledge (Durach et al., 2017). Therefore, we used metasynthesis, in line with the cumulative approach to knowledge creation (Elsbach & van Knippenberg, 2020). Furthermore, we saw the metasynthesis technique often overlooked when discussing and using rigorous reviewing methods in social science (Post et al., 2020). Finally, we found it suitable, as there are some qualitative findings while quantitative results are deficient.

Our analytical attention is paid to understanding and comparing the development paths of long-term IOR by identification, interpretive analysis, and comparison of various multi-phase life cycle models. Both the synthesis of prior qualitative evidence, as well as the development of an integrative framework for the non-definitive development path of IOR, contribute to the existing literature since our approach meets the requirements of both cumulative knowledge creation (Bell et al., 2005; Elsbach & van Knippenberg, 2020), and advancement of theory (Post et al., 2020).

Our literature review follows a systematic (Okoli, 2015), integrative (Snyder, 2019), and metasynthesis-based (Finfgeld-Connett, 2018) approach. The application of metasynthesis was considered reasonable as prior studies in the field, even though focused on similar topics and compatible research questions, were usually conducted on selectively chosen phases of the IOR development path, and used relationship life cycle (RLC) phases as a context and not as target phenomena, run in different industry and country contexts. Furthermore, all of them were based on case study analyses, being important in terms of the methodological compatibility requirement. Furthermore, metasynthesis helped us to identify knowledge gaps, reveal limitations, and allowed us to deal with knowledge redundancy (Erwin et al., 2011).

In this paper, using typical qualitative concept-centric synthesis (Kraus et al., 2020), we sum up the existing knowledge on the development path of inter-organizational relationships through the determination of the characteristics of RLC models and identification of their typical phases, breaking events, and the conditions underlying their trajectories. Through the above, this paper aims to demonstrate the multipath framework for the development of inter-organizational relationships. As we used metasynthesis, we integrate prior qualitative findings (Erwin et al., 2011; Finfgeld-Connett, 2018) in a way that results in the cumulative (Melendez-Torres et al., 2015) and context-free, thus more generalizable (Finfgeld-Connett, 2010) original framework.

Basically, in our paper we focus on answering two specific research questions:

To find relevant answers (i.e., meeting criteria of replicability, exclusivity, aggregability, and algorithmic character – Denyer & Tranfield, 2009), we carried out SLR using a predetermined protocol consisting of 5 typical phases (Okoli, 2015): (1) setting up the research questions, (2) literature collection, (3) literature screening and selection, (4) content analysis aimed at literature integration, synthesis, and interpretation, and (5) reporting (see Table 1).

The reviewing process started with posing research questions. Those questions resulted from a brainstorming session of our research team and underlined the literature searching process.

When gathering literature, we used 20 pairs of key search words linking inter-organizational relationships (IOR) with life cycle (RLC) concepts. The search process was conducted using academic databases (i.e., Ebsco, Scopus, and JSTOR), and a set of inclusion and exclusion criteria. As a result, an initial database was created (baseline sample – Durach et al., 2017) covering 547 publications.

By application of eligibility criteria (i.e., screening criteria – Okoli, 2015) we limited the search results to 101 works. This set of literature has been supplemented by 7 seminal papers from the explored research field. That is how we obtained the screened database covering 108 papers either (mainly) conceptual or empirical. In these articles, only 18 papers presented findings from qualitative studies. As metasynthesis was used as a reviewing technique these were extracted from the screened database to the final database and analyzed in detail. Following methodological requirements and practical guidelines of metasynthesis (Erwin et al., 2011; Melendez-Torres et al., 2015) the analysis of collected qualitative works was focused on integration (i.e., summary, multidimensional comparisons) and interpretation of different multi-phase life cycle models. The collected papers were analyzed using qualitative, thematic, and concept-centric synthesis (Kraus et al., 2020) remaining in line with the qualitative approach to metasynthesis (Finfgeld-Connett, 2018).

The analysis was conducted by considering: the number and iterative character of the phases covered by the life cycle model, the linearity and dynamics of the relationship development path, the subordination of relationship development to passing time (i.e., time dependence), and the approach to breaking events triggering changes of phases. The analysis of prior works was carried out using a qualitative approach with the general aim of deeper understanding of IOR development (Zimmer, 2006). Finally, it meets the criteria of methodological rigor imposed on reviews using systematic approach (Erwin et al., 2011; Garip & Kamal, 2019) and metasynthesis technique (Sandelowski et al., 1997; Melendez-Torres et al., 2015).²

According to the principles of metasynthesis (Finfgeld-Connett, 2018) the selected 18 works published between 2003 and 2019 in 13 different journals were comparatively analyzed (see Table A presented in Appendix I).

The very first conclusion was that the majority of scholars (14 articles) have not investigated the life cycle of inter-organizational relationships per se, but different issues shaping its trajectory, the key features of relationships across relationship development, the changes among those features, factors leading to those changes, success factors, etc. Indeed, in most cases, the development path has been used only as a research context in studies on the changeability, dynamics, importance, and intensity of different relational attributes. In addition, most of the analyzed works (15 out of 18) use just some components from the conceptual ideas presented by Ford (1980), Dwyer et al. (1987), or Ring & Van de Ven (1994). Thus the picture of the IOR development path lacks comprehensiveness and completeness.

Next, the adopted models of the RLC are perceived through either the stages or states theory. Followers of the stages theory (Ford, 1980; Dwyer et al., 1987; Wilson, 1995) note that relationships develop over time in a structured, linear, and predictable manner. On the contrary, supporters of the states theory (Batonda & Perry, 2003; Hastings et al., 2016; Plewa et al., 2013) assume that relationship development is an evolution of unpredictable states as the development path is complex, highly dynamic, iterative, and nonlinear. Furthermore, as relationships develop at qualitatively different speeds and patterns (Plewa et al., 2013), the IOR life cycle is not sequential or definitive (Ferreira et al., 2017). There is a lack of strong (if any) argumentation relying on particular models (not necessarily verified or suitable to present reality) available in the existing literature. Researchers are not consistent in adopting one commonly acknowledged approach, nor in using appropriate terminology remaining in line with the approach adopted.

What phases of the pattern(s) make up the IOR?

In the literature we can find the models covering a different number of phases ranging from 2 to 7, or under specific conditions even to 10 (Appendix I). Regardless of the number of the considered phases and the detail in which they are treated, three approaches can be identified: (1) focus on the general evolution of IOR over time (one work: Baptista, 2013), (2) exploration of one intentionally selected phase of RLC (two works: Mandják et al., 2015, and Hastings et al., 2016 – note that both papers focus on relationship birth acknowledged as a preliminary phase of RLC), (3) adoption of a multiphase approach with a stronger or weaker focus on the heterogeneity or changeability of phases (15 works).

Following metasynthesis assumptions, the collected papers were analyzed in terms of general similarities (Appendix I) in IOR development trajectory. Scholars do agree on one point that IOR develops over time across subsequent phases. Any other issues (e.g., the number, the scope, the range, and the changeability those phases) are all perceived differently.

Some of the analyzed models start with a pre-relationship phase (e.g., Lau & Goh, 2005; Heffernan & Poole, 2004; Abosag & Lee, 2013; de Almeida Moraes et al., 2017) also labeled as the pre-linkage phase (Plewa et al., 2013) or searching phase (Batonda & Perry, 2003). The pre-relationship phase is considered as a preliminary and immature phase appearing before relationship establishment, which refers to a relationship that formally does not exist. In this phase, firms search for and test potential partners. There are no bonds and commitment, thus the evaluation of potential partners is based mainly on economic aspects (Batonda & Perry, 2003; Hastings et al., 2016). Nevertheless, as claimed by Abosag and Lee (2013), partners are trying to look for signs and sources of individual trustworthiness, as interpersonal and social aspects do play a leading role here. All in all, this is the phase of the very first meetings and preliminary negotiations, whereas as shown by de Almeida Moraes et al. (2017) there is no formally established relationship. The future relationship (as yet non-formalized) can be dormant until some “turning point” appears (also “breaking event” – Mandják et al., 2015), when the pre-relationship finishes, a formal agreement is signed, the relationship appears, and RLC begins.

The first phase of the RLC can be labeled in many different ways, including initiating (Duanmu & Fai, 2007), matching (Ferreira et al., 2017), assessment (Davis & Love, 2011), exploratory (Lee & Johnsen, 2012), exploration (Akrout, 2014), and early development (Lau & Goh, 2005).

It starts with the formal foundation of the relationship. This usually involves testing and ensuring the compatibility of partners in order to ensure that partners will be able to jointly create value (Ferreira et al., 2017). Therefore, in this phase mutual goals are determined, communication processes are developed, and team-building processes are designed and initiated (Hastings et al., 2016). As the organizations are usually unknown to each other, this phase is characterized by high uncertainty and distance but with rapid adaptation (Lau & Goh, 2005). Given intensive intra- and inter-organizational team working processes, this phase is highly personalized, emotionally intensive (Ferreira et al., 2017), and full of fast trust-building processes (Meng, 2010).

The next phase is variably called developing/development (Lee & Johnsen, 2012; Batonda & Perry, 2003), intensifying (Duanmu & Fai, 2007), advancement (Plewa et al., 2013), commitment (Davis & Love, 2011), or growth (Abosag & Lee, 2013), as it refers to relatively systematic leveraging of the cooperation intensity, thus intensifying reciprocal communication, information and knowledge sharing, mutual commitment and investments. Therefore, it is also named a sharing phase as it focuses on intensive integration and consolidation, but also on sharing and bidirectional transfer of a wide range of resources in co-development activities (Ferreira et al., 2017). Importantly, as pointed out by Davis and Love (2011), it is during this phase that norms, standards, and procedures of cooperation are adopted and start to be monitored. Nonetheless, here monitoring mechanisms seem to be important as well, as the engaged parties still focus mainly on their own interests (Meng, 2010). In the development phase, however, we can observe continuous, mutual adaptation in operations and sub-processes (Abosag & Lee, 2013), increasing commitment and capital investments (Lau & Goh, 2005), and finally strong trust as well (Meng, 2010).

After development there is enduring (Davis & Love, 2011), stable/stability (Lee & Johnsen, 2012), maintenance (Akrout, 2014), or the long-term phase (Abosag & Lee, 2013). In practice, it can take a long time after the development phase for a relationship to be stable and mature (Ming-Huei & Wen-Chiung, 2011; Restuccia & Legoux, 2019). In the maintenance phase, partners become crucial to each other, their relationship is highly reciprocally valued, thus their interaction (prepared and negotiated before) becomes stable, providing continuous learning and growth-related mutual benefits (Lee & Johnsen, 2012). The long-term phase is characterized by incremental and systematic in-depth development, exclusive communications arrangements, account managers (Lau & Goh, 2005), and multidimensional bonds between engaged individuals (Abosag & Lee, 2013). Cooperation appears at a high, stable, and long-term level in which one can observe strong adjustment and integration of partners (Batonda & Perry, 2003). Now relationships become stronger, more aware, and truly focused on mutual goals.

Next, there is a decline (Meng, 2010; Restuccia & Legoux, 2019), dissolution (Abosag & Lee, 2013), termination (Batonda & Perry, 2003), or relationship end phase (Heffernan & Poole, 2004). This phase is seen as a process of gradual disengagement of partners (Duanmu & Fai, 2007) as the mutual commitment becomes asymmetric and decreases over time until finally it no longer exists – then a relationship dissolves. The end of a relationship may be predetermined, in which case, dissolution is a mutually agreed upon natural part of the relationship (e.g., the contract has expired – Batonda & Perry, 2003). In that case, the relationship end may be viewed as a natural phase in the RLC. Nevertheless, some findings show the end as unintended, caused by dissatisfaction with the relationship (Duanmu & Fai, 2007; Heffernan & Poole, 2004).

Finally, some of the most recent findings (Restuccia & Legoux, 2019) show that advanced relationships, communication, and exchange between actors continues even when formal interactions and agreement finish. These models assume there is no general end to a relationship as some (even incidental) interactions appear and some types of individual bonds remain unbroken. Indeed, as suggested by Hastings et al. (2016) and Ferreira et al. (2017), the life cycle of a relationship does not necessarily end. The majority of analyzed models (11 works) ends with phases referring to ongoing maintenance (e.g., Plewa et al., 2013) or long-term exploitation of relationships (e.g., Akrout, 2014; Baptista, 2013; Lee & Johnsen, 2012).

All in all, relationships are shown as never-ending but replicable (Hastings et al., 2016) and renewable (Batonda & Perry, 2003) as they can switch from a latent or passive form to being actively exploited again (Abosag & Lee, 2013). Interestingly, all of the non-linear models (see Table 2) consider the possibility of relationship renewal (e.g., Ferreira et al., 2017; Hastings et al., 2016). Moreover, some researchers not only assert the importance of a relationship remaining but also consider relationship renewal after a certain period in an inactive state. More specifically, these authors indicate the existence of an additional phase, namely the dormant (Batonda & Perry, 2003) or latent (Plewa et al., 2013) phase, which may lead to re-activation of a relationship (then re-activation, re-born phase can be considered as well) that has remained inactive for some time. A relationship in a dormant state does not mean that the relationship is terminated (finished). Moreover, depending on the particular circumstances, the latent phase can also occur after all of the other phases, including the pre-relationship one (Plewa et al., 2013).

Table 2.

Thus, we claim RLC can develop through ongoing, dynamic, even overlapping or dormant patterns over time (see Table 2). Furthermore, when considering claims about the endless nature of inter-organizational relationships, it is important to point to studies providing evidence that besides continuously maintained non-formal inter-organizational and individual interlinks, formal bonds and exchanges do also occur. In such a vein de Almeida Moraes et al. (2017) explore a post-relationship development path that leads to internationalization, whereas Restuccia and Legoux (2019) investigate post-deployment in relationships aimed at the preservation of warranty promises and ongoing delivery of help-desk services.

In sum, there is no one single, objectively determined, pervasive relationship life cycle model with a commonly acknowledged and comprehensively defined number of phases. Although it should not be surprising given the pluralistic nature of management sciences, the lack of a comprehensive and internally consistent – but not necessarily absolute – view may obstruct relational decision-making processes and thus limit relational capability (Mitręga & Pfajfar, 2015) and the relational rent (Dyer & Singh, 1998). This points to the need for a holistic and coherent view, which may be situational, on IOR development, one which covers a wide range of possible scenarios rather than one universal solution. Indeed, the existing stock of differentiated qualitative findings proves that the relational development path can go in many different directions.

We see the inter-organizational relationship as developing across paths including four distinguishable although not necessarily subsequent and constitutive phases, namely³: (1) initiation and initial development, (2) development, (3) maintenance, and (4) dissolution.

The distinct phases may be consecutive, but there may also be recurrences of previous phases. Furthermore, those phases usually appear on the relationship development path, whereas a particular relationship may bypass certain phases in its development. As actors’ engagement may vary in terms of general interest in a relationship, the individual phases can take three different forms: active (i.e., the typical form of a particular phase in which actors behave actively), passive (i.e., typical activities that are dormant/frozen at least by one actor), or re-activated (i.e., typical activities that are renewed after previously being dormant/frozen) (Table 2).

In its simplest form, the path of relationship development takes the form of a long-term relationship maintenance process (Davis & Love, 2011). It starts with formal relationship establishment (Heffernan & Poole, 2004) and covers longitudinal (or even endless) relationship extension and expansion (Ming-Huei & Wen-Chiung, 2011). Unlike the vast majority of RLC models, the framework of the relationship development path is seen as non-linear, “not necessarily sequential and definitive” (Ferreira et al., 2017). This is due to the development of every relationship remaining under the influence of dyadic, and therefore not fully controlled, factors as well as internal issues considered only by the actors engaged in the relationship. We claim the non-linear character of relationship evolution to be indisputable. Relationship path development is not only cyclical, where the steps repeat (Ring & Van de Ven, 1994). It is ongoing, dynamic, changing, and strengthening over time. Nevertheless, it can end at any phase of the relationship development path (Akrout, 2014), and starts together with a contract or can be re-started after a latent phase (Plewa et al., 2013), which depends on relational experience (Duanmu & Fai, 2007). In addition, each phase has its own evolution pattern (Batonda & Perry, 2003) and its duration is contextually conditioned. Thus, some bonds can be ongoing even if the goals have been achieved (Baptista, 2013; Ming-Huei & Wen-Chiung, 2011; Restuccia & Legoux, 2019), or can be replicable (Hastings et al., 2016) and renewable (Batonda & Perry, 2003) after the latent phase (Abosag & Lee, 2013; Plewa et al., 2013). At the same time, the relationship development path can end, as the final natural phase in a life cycle caused by the ending of a contract (Batonda & Perry, 2003) or by dissatisfaction with the relationship (Duanmu & Fai, 2007; Heffernan & Poole, 2004).

The aim of this paper was to identify IOR development path conceptual models across qualitative studies and to develop – using those models – an integrated and comprehensive framework for the development of inter-organizational relationships. The outcomes of our review directly targeted the identified gaps. Our study confirms prior claims about no common approach to the development of IOR (Jap & Ganesan, 2000; Hastings et al., 2016; Ferreira et al., 2017; Meng, 2010) as well as knowledge deficit on the course and specificity of the relationship life cycle (Hastings et al., 2016; Ferreira et al., 2017; Meng, 2010) especially on dynamics of IOR collaboration in temporary contexts (van Marrewijk et al., 2016) and the questionable issue of relationship renewal (Ferreira et al., 2017). Our study supports prior claims that the relational development path can go in many different directions. Every single relationship development path seems to be unique; however, some minor patterns or regularities are visible (as shown in Table 2). Our review shows that relationships – explored so far in different cognitive, industry, and country contexts as shown in Table A – seldom go through a linear, sequential, and definite step-by-step development process. Indeed, the development of every IOR is framed by a specific context (Baptista, 2013; de Almeida Moraes et al., 2017) therefore there is no one generic, pervasive model of RLC that always displays identifiable stages or states (Batonda & Perry, 2003). Nonetheless, even though there is no absolute model, the existing qualitative studies when integrated show IOR as developing across a path including four distinguishable, although not necessarily subsequent and constitutive, phases (initiation and initial development, development, maintenance, dissolution). Those phases are not constitutive, as there is always a threat of IOR termination as well as the opportunity for IOR renewal (as suggested for instance by Ferreira et al., 2017; Hastings et al., 2016). The metasynthesis of the existing, qualitatively developed knowledge has shown the development path of IORs as highly contextual, complex, and not self-determined, thus rather emerging than pre-determined, possibly rather non-replicable than repetitive. Indeed, our findings show the life-cycle rather as a metaphor that was useful in the early days of IOR research (the 1980s and 1990s), but which did not hold up later when confronted with the results of qualitative empirical research. This may have been due to the dramatically increasing turbulence of the environment, or perhaps because of the emergence of new evidence in the field, or triggered by a significant development of the relational approach. Whatever the reasons, today the lifecycle appears to be inadequate to IOR dynamics.

Given the terminological perspective, we claim to use the term development path (as done, e.g., by Baptista, 2013; Batonda & Perry, 2003; Duanmu & Fai, 2007; Meng, 2010) instead of the “life cycle” or “process” (as done, e.g., by Davis & Love, 2011; Heffernan & Poole, 2004; Lau & Goh, 2005; Mandják et al., 2015) as this terminology seems to be more pertinent. On the one hand, as suggested by Ferreira et al. (2017), by using the term development path and the possible phases that can be considered along this path we show that the pattern of development is non-linear, non-sequential, and non-definitive (Harmeling & Palmatier, 2019; Plewa et al., 2013). On the other hand, in the development path approach, we support prior claims that there is no one generic, pervasive model of RLC that always displays identifiable stages or states (Batonda & Perry, 2003). This is because every relationship development is embedded in specific contexts (Baptista, 2013; de Almeida Moraes et al., 2017), and is dynamic, unpredictable as determined by individual, organizational, and inter-organizational issues, actor- and time-dependent, thus customized (Restuccia & Legoux, 2019), situational (Lau & Goh, 2005), and non-definitive (Ferreira et al., 2017). Importantly, as the proposed model for the multipath development of IOR integrates prior qualitative findings from field research run in different national, cultural, and industry contexts, it can be seen as context-free and thus more generalizable (Finfgeld-Connett, 2010) than the models (c.f. Table A in appendix) used to develop this framework.

We argue that the adopted methodological approach and the identified significant differences between particular relationship development paths (Appendix I) together allow us to draw conclusions on a general but not absolutely universal, generic, fully-defined, or constitutive framework for the IOR dynamics. As shown in Table 2, the relationship development path should be thought of as ranging from relationship establishment (i.e., initiation and early development) to relationship dissolution, considered either as an unending continuation (i.e., without formal cooperation on given goals but with informal bonds) or as a complete ending (i.e., without either formal cooperation or informal bonds).

The proposed multipath framework adopts a situational approach. Thus the IOR development path can take different shapes (i.e., in terms of the content of the relationship and its specific features, and the order and number of phases as these are not constitutive), may take different trajectories (i.e., go through different forms of particular phases, namely active, passive, or re-activated), and may be unique and inimitable due to being highly conditioned by time, different external contexts, and the different characteristics of both relationship development and relationships per se.

Such a multidirectional development pattern of IORs supports prior claims made by Havila and Wilkinson (2002) pointing at three possible, alternative scenarios for the development of the phase of the relationship aftermath (i.e., decreased trust, vanished, and sleeping) determined for instance by the industry type or by the features of IOR. All in all, we argue that it is hard (if even possible) to design an unambiguous development path as relationships develop in an unpredictable, dynamic, and flexible way. Thus every relationship can be seen as unique, unrepeatable, and inimitable.

Our reviewing study addresses two research questions.

RQ₁ : What phases of the pattern(s) make up the IOR? It is hard to define a generic model for the IOR life cycle. Instead, IOR development is characterized by extremely high contextuality, iterativeness (and regressiveness), evolutionary dynamics, and dependency on actors’ adjustments, and finally time-dependency. Therefore, we propose adopting a situational approach under which IORs develop along a flexible multipath route (Table 2). We do believe that our situational approach supports prior arguments that relationships seldom go through a definite step-by-step development process (Batonda & Perry, 2003; Harmeling & Palmatier, 2019).

RQ₂ What is the pattern of relationship development like? In general, IORs start, develop, stabilize and decline. Nonetheless, although the general shape of IOR development can be outlined, it is impossible to adopt one generic approach and identify the building blocks that make up the life cycle of every single IOR. IORs develop in different ways as they are not independent of external factors (e.g., culture, industry specificity) and remain impacted by the characteristics of each specific IOR, e.g., dynamics, non-linearity, flexibility, etc.

This study adds to the popularization of the adoption of metasynthesis in social sciences (Dekker & Bekkers, 2015) which is gaining recently increasing popularity (e.g., organizational psychology – Bazzoli & Probst, 2022; management – Habersang et al., 2019; entrepreneurship – Rauch et al., 2014) but remains quite often overlooked when discussing rigorous reviewing methods in the field of management (Elsbach & van Knippenberg, 2020; Post et al., 2020; Snyder, 2019). By using metasynthesis our study focuses mostly on the interpretation of a reviewed, integrated, and summarized stock of knowledge (Erwin et al., 2011) and may be a preliminary step to applying quantitative research that will provide “the big picture” of the IOR life cycle. The analyzed qualitative studies include narrow scopes and restricted research contexts, as the investigations were usually limited to one country, one industry, or one organization. As a result, in the literature, there are a great many “>little islands of knowledge< separated from each other and doomed ultimately never to be visited” (Sandelowski et al., 1997: 367). We hope that by using metasynthesis we were able to offer solutions which remove these limitations (Garip & Kamal, 2019), thus add to cumulative knowledge creation (Elsbach & van Knippenberg, 2020) and theory advancement (Post et al., 2020). Moreover, as metasynthesis is claimed to allow researchers to make cross-case generalizations (Sandelowski et al., 1997), our findings can be seen as offering moderatum generalization (Finfgeld-Connett, 2010).

Summing up, the application of metasynthesis allowed us to integrate the fragmented and diverse (or even contradictory, as visible in Appendix I) views on the IOR life cycle and thus to offer – based on their critical and in-depth analysis and synthesis – a multidirectional trajectory of IOR development. Taking into account the methodological advantages of metasynthesis, the proposed framework can be seen as valuable because it is a single, coherent, cumulative proposal developed on the basis of results obtained from desk research of past studies.

Additionally, our research provides a theoretical contribution to the relational view of a firm’s competitive advantage in several ways. We compare and integrate qualitatively explored models of the RLC and we provide definitive evidence for the managerial futility of modeling the life cycle of relationships. As our analyses follow a systematic approach to reviewing the literature, we provide a more in-depth, extensive, and comprehensive summary and comparison of prior works dealing with RLC. In particular, we do not describe and compare the classical approaches (e.g., Ford, 1980; Dwyer et al., 1987; Wilson, 1995) as they are described, compared, and commented on in-depth by Batonda and Perry (2003) and by Mandják et al. (2015). Instead, we show much more specific differences and similarities between prior models than simply the number of phases, theoretical approach, and industry context (Abosag & Lee, 2013; Davis & Love, 2011; Hastings et al., 2016). Indeed, we emphasize the rich description of the IOR development path through presentation and comparisons of different meanings, interpretations, and contexts for what is claimed as required in any kind of qualitative, scientific investigations (Guba & Lincoln, 1994).

Moreover, as we use a study-based (Finfgeld-Connett, 2018) and integrative (Elsbach & van Knippenberg, 2020) approach, we see the proposed multipath framework for the development of inter-organizational relationships as comprehensive and context-free, thus much more general (Finfgeld-Connett, 2010) and applicable in practice. Indeed, the existing (more or less holistic) frameworks resulted from qualitative investigations should be seen usually as highly “exploratory” and “contextualizing” (Andriopoulos & Slater, 2013) while their integration – here achieved using metasynthesis (Garip & Kamal, 2019) – can give ground for cohesive conclusions (Durach et al., 2017; Okoli, 2015) which should be tested in quantitative studies.

Next to the theoretical and methodological contributions, this paper offers managerial contributions as well. Providing a synthesis of prior findings develops managers’ understanding and awareness of the different approaches to modeling the IOR life cycle. We see this contribution to be crucial, as leveraging such awareness is shown to be important in improving the inter-organizational process by building more effective relational activities (Harmeling & Palmatier, 2019; Hastings et al., 2016; Quinton & Wilson, 2016) leading to higher exchange performance (Shen et al., 2020). Using our findings (Table 2) managers can identify the phase of the particular business relationship, recognize possible options for its further development, and navigate to the best one. Moreover, it should be noted that increasing managers’ awareness increases protection from managerial myopia (including strategic myopia in particular) and thus favors more effective managerial decision-making processes. Moreover, the leveraged understanding and awareness seem to be relevant managerial implications, “as better understanding of the dynamic aspects of relationships is needed, because in practice the majority of collaborations appear to be unsuccessful” (Mitręga & Pfajfar, 2015).

Besides the above contributions, we do notice some limitations to our work. First, we should consider a wide range of limitations resulting from the adoption of a systematic approach to literature reviewing (Okoli, 2015). For instance, given our study, we should note that when analyzing the papers we noticed that additional search criteria could have been used, namely “state,” sometimes used when describing theoretical lenses of the IOR life cycle. As summed up by Batonda and Perry (2003) the life cycle of IOR can be explained using the stages theory considering RLC as an evolutionary process or the states theory considering RLC as a result of unstructured strategic behaviors and movements made by the partners. Unfortunately, we have not used “state” among our search criteria (see Table 1), which could impact our reviewing process. Furthermore, when it comes to search criteria we used “interorganizational” but did not use “interorganisational,” which may have influenced the results as well. It should be emphasized that we found no significant differences between hyphenated and hyphen-free spellings. All in all, the set of search criteria was built by an experienced research team during the brainstorming session (Pinto et al., 2012). The team tried to adopt a broad approach to IOR (i.e., there were no limitations related to the type of IOR considered, e.g., horizontal/ vertical, business/non-business, long-term/short term, coopetitive/cooperative, etc.) while still focusing on their features. Next, this review followed the typical systematic (Tranfield et al., 2003) and qualitative approach to metasynthesis (Finfgeld-Connett, 2018). Therefore, the findings take the form of conceptual propositions based on qualitative, thematic and concept-centric synthesis (Kraus et al., 2020) carried out by the research team, thus the conclusions drawn are prone to the risk of subjectivity (Denyer & Tranfield, 2009; Finfgeld-Connett, 2010; Melendez-Torres et al., 2015; Lachal et al., 2017; Snyder, 2019). In future it would be valuable to supplement our findings with quantitative analyses of the existing literature (e.g., bibliometric or at least descriptive ones – Kraus et al. 2020) as well as to test the developed multipath framework in quantitative surveying. Moreover, one should point at the main limitations related to the application of metasynthesis, referring to the focus on qualitative studies only, the subjectivity of the reviewing process (including its outputs), and the lack of quality assessment of the analyzed works during the reviewing process. Nevertheless, we saw this reviewing technique as relevant due to the deficit of quantitative studies in the field, which made the application of meta-analysis impossible. At the same time, we believe that the limited number of selected works confirms the legitimacy of the goal we have set here (the dynamics of IOR remains under-studied) as well as the use of metasysthesis to develop an integrated picture of the IOR dynamics phenomenon. Using our 18-element pool of work, we paid great attention to in-depth and critical analysis of the identified works in terms of the relationship type studied, examined sector, and country/cultural environment (Appendix I). Furthermore, to make our investigation more reliable and the findings more valid, the analysis of those 18 qualitative works was carried out independently by all of the team members, then the results (including differences in particular) were discussed by the entire research team, also during brainstorming sessions. Given the above limitations, including those typical for studies adopting qualitative metasynthesis of prior research (Finfgeld-Connett, 2010, 2018; Melendez-Torres et al., 2015), to make empirically grounded generalizations, quantitative studies are required. Next to the quantitative verification of the multipath framework of IOR development it would be interesting to explore (in further qualitative studies) or to test (in verification quantitative studies) potential moderators playing roles in IOR development and shaping the timespan, number, and type of phases in the specific IOR development scenarios. Our literature review showed us that there are some preliminary ideas for such potential moderating variables explored so far as contextual factors (see the second section presenting our findings). Interestingly, those moderators seem to be varied in terms of the level of analysis. Among the potential moderators one can point at those directly linked with: (1) the context in which IORs are exploited (e.g., type of industry, as suggested by Havila & Wilkinson, 2002), (2) organizations engaged in IOR (e.g., their responsiveness, flexibility, or cooperation perception – Harwood, 2006), (3) managers operationally responsible for the establishment, maintenance, and development of IOR (e.g., empathy – Harwood, 2006), or (4) IOR per se (e.g., relational attributes – Jap & Ganesan, 2000; relationship properties – Jap & Anderson, 2007; relationship features – Havila & Wilkinson, 2002). We perceive the need to investigate the aforementioned moderating variables (as well as other contextualities or even newly revealed aspects in further qualitative investigations) in future research as essential for a deeper understanding of the situational approach to the IOR development path, thus to cumulatively develop our knowledge in the field and generate more specific practical conclusions for managers interested in relational strategies adoption.