Currently, most research endeavors are of collaborative nature (Bozeman and Boardman, 2014; Bozeman and Youtie, 2017; Baker et al., 2017), which presents researchers with challenges that they would not be able to overcome working alone (Falk-Krzesinski et al., 2011). As a result, research has begun to focus on the problems that arise from collaborative work and possible ways to solve them (e.g., Youtie and Bozeman, 2014; Bozeman et al., 2016; Hückstädt, 2022; Meißner et al., 2022). The situation is complicated by the fact that to an ever growing degree, research collaborations (RCs) do not only include academic institutions but also non-academic organizations, especially from the business sector (Teirlinck and Spithoven, 2015; Tijssen and Wong, 2016). Consequently, there is a growing amount of studies on this type of collaboration, i.e., university-industry collaboration (UIC; e.g., Ankrah and AL-Tabbaa, 2015; Mascarenhas et al., 2018; Yegros-Yegros et al., 2016; Azagra-Caro et al., 2010; Tijssen et al., 2009). However, while research has tackled collaboration problems and potential solutions at a general level, there is still little knowledge about this issue with respect to UIC in specific. We address this research gap in this article.

As no systematic analyses of UIC that comprise all three aspects of expectations (RQ1), problems (RQ2), and solutions for self-governance (RQ3) of RC participants and their different perception by university or industry members exist, this calls for an exploratory procedure. Thus, we operated qualitatively and conducted focus group discussions. In our study, we followed a deductive-inductive approach with the aim to test established analytical categories that we borrow from Meißner et al. (2022) as well as to iteratively develop new categories that allow for an improved in-depth insight into RC, especially into UIC. We chose this qualitative design to ensure applicability to the complex dynamics of UIC. Focus groups were particularly useful for capturing the subjective interpretation of expectations, problems, and solutions rather than just their frequency. By adjusting the theoretical framework via a deductive-inductive approach, we could uncover how identical behaviors are interpreted through divergent organizational logics—a nuance that purely quantitative methods likely miss. The deductive categories were based on the seven problems and twelve solutions derived from the Meißner et al. (2022) study exploring a first framework of research collaboration problems and how to address them from a self-governance perspective. The inductive categories were then developed based on the focus group discussions according to the Qualitative Content Analysis (QCA) method (Kuckartz, 2019) until saturation occurred.

The study included three focus group discussions with ten participants: eight academic researchers employed as professors, engineers, or graduated research assistants at universities or organizations for applied research and two industry members from five RCs in Germany. The participants are experts in applied mechanics, gas technology, thermodynamics, robotics, plant sciences, biochemistry, tourism, and alternative fuel development. One of the industry members is a lawyer on management level in a small-scale tourism company and the other a project coordinator for industry-university collaboration in a large-scale automotive company. They were recruited via email in the period from January 10 to February 23, 2021. The average participant is male, middle-aged, and highly educated. In choosing these participants for the focus group discussions, we considered the extent of interdisciplinarity and the degrees of network complexity (team sizes and constellations). To obtain data relevant for UIC, most of the participants are members of heterogeneous RCs that partner with large-scale companies as well as small- and medium-sized enterprises (SMEs). Only two participants are members of a homogenous RC (i.e., university-only). The ten academic researchers and industry members occupy different roles in the UIC: speakers (n = 4), principal investigators (n = 3), and managing coordinators (n = 3). They were divided into focus groups according to their role (i.e., the groups included both university and industry members). This separation allowed for three focus group discussions that enabled the participants to discuss the research questions together from the perspective of their role. Furthermore, through this composition we were able to guarantee largely open and non-hierarchical discussions.

Two authors of this paper moderated the semi-structured focus group discussions in a semi-directive manner. Due to the ongoing COVID-19 pandemic, all discussions were held online on March 23 and 29 and on April 1, 2021. Beforehand, all participants were asked to sign written declarations of consent and information on data protection. The discussion questions were premised on the research questions. After some opening questions, the respondents were asked to report on their expectations concerning collaboration in research. After that, the seven problems in RC (Meißner et al., 2022) were explained to the participants and they were asked to rank them according to perceived relevance. Moreover, they were encouraged to report on additional problems to expand the existing catalogue of problems. Additionally, the utility of previously identified solutions in the context of RC (Meißner et al., 2022) was discussed and the catalogue of solutions was augmented. In the last step, we asked the participants to relate the enhanced list of problems and solutions to each other and determine who is responsible for initiating solutions as well as when. All these questions supported an advanced examination of UIC as further inquiries reiterated differences between university and industry that participants detected.

Based on 295 min of discussion material, which was transcribed into 96 pages of written text, we analyzed our data using the five steps of Kuckartz’s (2019) Qualitative Content Analysis. This qualitative approach allows for a detailed coding and subsequent categorization of the transcribed data to consolidate relevant information regarding our research questions. As this study follows a deductive-inductive approach, we developed both concept-driven (deductive) and data-driven (inductive) categories to evaluate our material. To achieve an in-depth understanding of the focus group discussions and relevant insights into UIC, we (1) intensively read and studied the text as well as wrote summaries of their content, (2) formed main categories that corresponded to the questions asked in the focus group discussions, (3) consensually coded the data according to the main categories, (4) compiled text passages of the main categories and inductively formed subcategories based on this compilation and coded all data according to the main categories as well as subcategories (Kuckartz, 2019). The entire coding process was conducted with the software MAXQDA and allowed us to apply the categories deduced from the previous Meißner et al. (2022) study as well as to detect new categories that significantly enrich our understanding of expectations (RQ1), problems (RQ2) and solutions (RQ3) which participants of UIC deem relevant. In contrast to quantitative content analysis in which coding units are defined in advance, Kuckartz’s QCA allowed for a definition of coding units by the coding process. Therefore, the analysis was carried out on a circular basis to accurately code and categorize all relevant data in repeated coding processes. The main coder independently completed a first round of coding, which then was surveyed by another author of this article. Upon agreement of the coded elements, the remaining circular coding was carried out by the main coder. To code the transcribed text, it was split into precisely defined parts of material covering content relevant for the research questions and either subsequently assigned an established category or designated a new category. Saturation during analysis was ensured through reading the entirety of the transcribed text repeatedly and either assigning its elements to previously established deductive categories or establishing new inductive categories. In sum, 97 elements of the text were coded. In a final step of the QCA process, the analyzed data is (5) reported and documented in this publication to retain our findings (Kuckartz, 2019).

Based on our three focus group discussions, we formed 14 main categories, derived from our research questions and further discussion questions as well as 131 subcategories. Since this paper concentrates on UIC, the following four categories are of particular relevance: (1) expectations from industry members, (2) expectations from university members, (3) perceived relevance of problems in RC as well as (4) solution approaches perceived as especially useful. For each of these main categories, we developed a range of subcategories whose frequency of mention (i.e., number of coded elements) we ranked to obtain the three most relevant subcategories for university as well as for industry members. See Table 3 for a description of the subcategories.

We present our findings in three main parts: First, we discuss the expectations from industry and university members (RQ1), then we outline the problems perceived as most relevant in the context of UIC (RQ2), and lastly, we demonstrate the solutions perceived as most useful to solve these problems potentially arising in UIC (RQ3). There are more findings for the university members’ perception of RC since they constitute most of our focus group participants.

Based on the assumption that successful collaborative research depends on the solving or prevention of collaboration problems (Youtie and Bozeman, 2014; Bozeman et al., 2016; Meißner et al., 2022), the present study sought to identify the expectations, problems, and solutions to these problems in the context of UIC. Specifically, we aimed to carve out the differences between academic and non-academic organizations through focus group discussions with researchers from UICs. Indeed, we found more differences than commonalities. First of all, concerning the expectations with respect to how collaboration should work and what should result from it, industry members lay a strong emphasis on gaining financial profit. Researchers from universities, in contrast, primarily expect the exchange of knowledge and education for junior researchers. Consequently, university and industry members also differ when it comes to the most exigent problems in collaboration. Researchers from universities report the commitment problem as the most important one and designate the industry members as the ones lacking commitment, thus impeding successful collaboration. For industry members, the difference problem is the most prominent one. This primarily results from their perceived priority of university members to focus on theory, while industry members prioritise practicality. Lastly, concerning the solutions to collaboration problems, university members’ primary recommendation is to select participants based on their proven ability to collaborate. For members of the industry, the most promising solution is an integrative and competent personality for leadership.

A particularly striking finding is the discrepancy in the perception of the commitment problem. This problem was ranked as the most critical issue by university members but was not highlighted by industry partners. This divergence may not necessarily reflect a lack of self-awareness from industry members, but rather a difference in how each side interprets the root cause of collaboration challenges. From the university perspective, which is often process-oriented and focused on continuous engagement, a lack of participation from industry partners is perceived as a failure of commitment. Conversely, industry members, who are typically results-oriented, may view the same disengagement not as a commitment issue, but as a symptom of the difference problem. Thus, they may perceive the academic activities as insufficiently practical or relevant to their commercial goals. In this light, what one partner labels a “commitment problem,” the other may experience and describe as a “difference problem,” highlighting a fundamental disconnect in collaborative expectations and priorities.

To deepen the interpretation of this discrepancy, it is helpful to look beyond the project level and apply the lens of “research duos” (Glebova, 2024). As Glebova (2024) argues, the collaborative essence of research lies in dyads, which form the elementary building blocks of wider networks. These “research duos”—for instance, a university PI and a specific industry project lead—are the locus where trust, communication patterns, and mutual expectations are negotiated. Consequently, the clash we observed is not just an abstract institutional conflict but a friction experienced within these specific dyads. In a duo where the academic partner expects continuous interpersonal exchange, a lack of constant feedback may be interpreted as a personal commitment problem. Conversely, the industry partner in that same duo may view the academic’s expectations as a difference problem. Thus, the divergent problem perceptions can be understood as emergent properties of misaligned expectations within specific research duos.

What can be concluded from these findings and what is their added value for research and practice? On the one hand, our findings have an epistemological value in that they reveal that the type of organization has substantial consequences with respect to what researchers expect from collaborations, what they regard as the most relevant problems, and what they recommend as solutions to these problems. This underlines that there is an urgent need to investigate UIC because universities no longer conduct research among themselves, but together with organizations from other sectors (e.g., Falk-Krzesinski et al., 2011; Teirlinck and Spithoven, 2015; Tijssen and Wong, 2016). In a nutshell, researchers from universities seem to focus on cooperation and working together, intellectually as well as personally. Collaboration members from the industry, in contrast, seem to mostly appreciate the organizational and structural aspects of collaborative research projects. To put it pointedly, academic researchers seem to think RCs in persons (i.e.,” thinking in persons”), non-academics in structures (i.e.,” thinking in structures”). Our study thus follows on from previous research that has shown that researchers from universities and industry approach research collaborations with very specific, different motivations (e.g., Ankrah and AL-Tabbaa, 2015; Cugmas et al., 2020; Ryan et al., 2008). Hence, our findings suggest a conceptual model connecting organizational background to self-governance: The type of organization (university vs. industry) shapes expectations, which in turn determines perceived problems (e.g., commitment vs. difference) and preferred solutions (e.g., selection of participants vs. leadership personality). Furthermore, the insights of our study add to the empirical knowledge within research on UIC insofar as our study is to the best of our knowledge, the first one that took a comparative perspective on the expectations, problems, as well as on the solutions in the context of UIC.

On the other hand, our findings provide practical guidance for researchers and other actors involved in UIC. Researchers can rely on this knowledge to better select and understand their partners and to improve their collaboration. For example, university members of a RC could pay more attention to the potential financial merit of their research (e.g., in form of patent applications or public relations), because this is an important goal for the industry. Furthermore, the results of our study can also be used by funding agencies and policy makers, for example, to revise existing criteria for the reception of funding. For funding agencies, this may imply moving beyond checking boxes for industry partners. Calls could require a specific “collaboration governance plan” that explicitly addresses how the consortium plans to bridge organizational logics. The consortia could in turn organize preparatory workshops to raise awareness among participants of the different logics. As previous research has shown, there is great potential for optimization, for example with regard to state funding of UIC (e.g., Ryan et al., 2008; Veletanlić and Sá, 2019). Furthermore, apart from specific measures taken, our findings show that all actors involved need to be sensitive to the different organizational background of researchers and the resulting implications. Thus, our findings may also enable collaboration to become increasingly successful in researchers’ everyday working life.

Of course, our study has a number of limitations that need to be discussed. Before we come to the conclusion, we would like to focus on what we consider to be the three most important limitations. First, our findings are based on three focus group discussions with German researchers, and only two out of ten participants were industry members. We made sure that the participants had different backgrounds in terms of disciplines, organizational affiliation, and socio-demographics. This means that the various types relevant to our research questions were covered. In addition, we made sure that the industry’s views were saturated adequately in each of the discussions by asking specific questions and follow-up questions. Nevertheless, our findings are based on a modest and asymmetric sample, and we cannot rule out the possibility that the views of academia carried disproportionate weight in our study, which would distort the comparison between the two types of organizations. Thus, to ensure full empirical as well as theoretical saturation, more studies, specifically ones with quantitative designs, are needed to examine the validity of our results on a broader basis and in other contexts (e.g., different countries and academic environments) and increase generalizability. Future research should also more explicitly consider the specific context of the German research landscape. For instance, cultural and institutional norms, such as Germany’s federally organized research funding structure—in contrast to more centralized systems in countries like France—or its traditionally hierarchical university systems compared to the flatter departmental structures common in the US, may shape collaboration dynamics in ways not fully captured by our study. Furthermore, it is crucial to note that this study was conducted during the COVID-19 pandemic. The “distance perspective” (Skute et al., 2019) was thus not only cognitive but physical. The lack of informal face-to-face interaction and informal exchange likely exacerbated the commitment problem perceived by university as well as industry members. It should therefore be emphasized that our study does not claim to be universally valid. However, our results offer a good starting point for formulating and testing specific hypotheses, improving deficiencies in terms of standard quality criteria (such as validity and reliability), and testing our findings on a broader basis. Second, our study was closely based on the findings by Meißner et al. (2022), that is, on a pre-existing catalogue of problems and solutions identified in the context of collaborative research. Consequently, there may be other collaboration problems and solutions to these problems we may have overlooked, for example, the problem to develop common goals and make joint decisions (Volk, 2021). Third, we focused on the top three expectations, problems, and solutions. Most certainly there are more differences to be found when looking at the whole picture of the different motivations, problems, and solutions. However, our study does not explain the empirically observed differences in expectations and perceptions of problems and solutions. This can now be addressed in further studies with different methodological designs, relying on specific theoretical approaches (e.g., self-governance, motivation theory) with the necessary explanatory potential.

To summarize, like other studies before ours have shown, studying the details that account for the success of UICs is a complex matter. Specifically, our findings reveal that focusing on the differences that emerge from researchers’ organizational backgrounds is pivotal in order to understand the cornerstones of this success. We hope that our study not only informs further research on this issue but may also guide researchers through their collaborative journey.