Context-Aware Business Process Management

The CAMAS Method consists of a Classification Framework, an Assessment Process, and a Selection Process. Linking the CAMAS Method to the method attributes from Table 1, it focuses on the context awareness of BPM methods (goal orientation). Therefore, the Classification Framework structures context along three dimensions that build on the BPM lifecycle and the BPM context framework introduced in Sect. 2 (principle orientation). Guidelines for the application of the Classification Framework are embedded in the Assessment and the Selection Process (systematic approach). The Assessment Process guides BPM method engineers and users to assess existing or newly developed BPM methods regarding their applicability to specific contexts (use case 1). Assessed BPM methods feed into the Method Base of the CAMAS Method, so the Assessment Process is an important prerequisite for the application of the Selection Process. The Selection Process, in turn, guides BPM method users to select BPM methods that fit their contexts at hand (use case 2). Both processes are further specified in terms of activities, which comprise techniques, tools, roles, and outputs that support their execution in various contexts and among various users (repeatability). Figure 4 illustrates the structure of the CAMAS Method and the relationship among its components. Details on each component are provided in Sect. 4.2 (Classification Framework), Sect. 4.3 (Assessment Process), and Sect. 4.4 (Selection Process).

At the center of the CAMAS Method, the Classification Framework facilitates the assessment of BPM methods’ applicability in terms of BPM lifecycle stages (lifecycle dimension), goal orientation (goal dimension), and various context dimensions of the BPM context framework (context dimension). The Classification Framework extends an existing assessment scheme (Denner et al. 2018a) and builds on the BPM lifecycle (Rosemann and vom Brocke 2015) as well as the BPM context framework (vom Brocke et al. 2016). Figure 5 illustrates the Classification Framework as a three-dimensional cuboid that serves as joint meta model for the Assessment and the Selection Process.

The lifecycle dimension represents the BPM lifecycle stages, so BPM methods can be categorized along five elements: process design and modelling, implementation and execution, monitoring and control, improvement and innovation, and project and program management (Rosemann and vom Brocke 2015). For reasons of simplicity, we shortened the names to design, implementation, monitoring, improvement and innovation, and project management.

The goal dimension refers to a BPM method’s goal orientation, differentiating exploitation and exploration (Benner and Tushman 2003). In line with its strategy, an organization decides whether to exploit (i.e., improve), explore (i.e., innovate), or both (vom Brocke et al. 2016). The other dimensions of the BPM context framework refer to the given context and cannot be modified, so the Classification Framework separates the goal dimension from the process, organization, and environment dimensions.

The context dimension covers the last three dimensions of the BPM context framework, which consist of additional hierarchically structured context factors and characteristics. Unlike the BPM lifecycle and the goal dimension, most characteristics of the context dimension are not assessed in a yes-or-no but a high-or-low logic. In contrast to vom Brocke et al.’s (2016) BPM context framework, we assess each context factor on a two-point scale and avoid medium-level characteristics (e.g., medium knowledge intensity) to achieve a binary categorization of BPM methods. The medium-level characteristics of the context factors lack a clear definition, so they depend heavily on subjective interpretation that could bias assessment and selection results (Christenfeld 1995). As unified measures and thresholds for context characteristics yet need to be developed, we used generic definitions (vom Brocke et al. 2016).

Linking the Assessment Process to the method attributes (Table 1), it strives for assessing the applicability of existing or newly developed BPM methods in specific contexts (goal orientation) and, using the Classification Framework, it integrates the lifecycle dimension and the BPM context dimension outlined in Sect. 2 (principle orientation). Comprised of four consecutive steps, the Assessment Process starts with determining which BPM method to assess. This method is then assessed in terms of the lifecycle dimension. Finally, the method is classified based on the characteristics of the goal dimension and other context dimensions (systematic approach). Completing each step (Table 2) supports the execution of the Assessment Process in various contexts and among various users (repeatability). We provide more detailed insights below. The evaluation of the Assessment Process is discussed in Sect. 5.1.

Identifying a BPM method (A1) requires determining an existing or newly developed BPM method whose applicability to a specific context should be assessed (technique/output). Therefore, a literature review can be helpful (tool). To ensure that the identified BPM method is suitable to go through the Assessment Process (technique) and, thus, to be an appropriate input for the Selection Process, it should match the definition of a BPM method (Sect. 2) (tool). Depending on the used technique, activity A1 is performed by a BPM method engineer who developed a new BPM method or by a BPM method user who comes across an unclassified BPM method in the course of their daily business (role).

Classifying the lifecycle dimension (A2) requires classifying the BPM method with respect to the targeted BPM lifecycle stage in line with the Classification Framework (technique/output). If a BPM method is applicable to more than one BPM lifecycle stage, a multiple assessment can be performed. This activity is supported by the proposed Excel prototype (tool). For a better understanding on how the Excel prototype works, please find a blank version in the electronic supplementary material. Activity A2 is performed by a BPM method engineer for a newly developed BPM method or a BPM method user for an existing BPM method (role).

Classifying the goal dimension (A3) requires classifying the BPM method with respect to its characteristic(s) in the goal dimension (output), that is, to assess whether the BPM method is geared to incremental improvement (exploitation), radical (re-)design (exploration), or both (Rosemann 2014) (technique). This activity is supported by the proposed Excel prototype (tool). Again, activity A3 is performed by a BPM method engineer or BPM method user (role).

Classifying the context dimension (A4) requires using the process, organization, and environment dimensions of the BPM context framework to classify the BPM method according to the Classification Framework (output). The BPM method user determines whether the BPM method is applicable to the underlying characteristics (technique). The applicability of a BPM method to a specific context is expressed by a nominal scale that consists of three assessment criteria: not applicable (na), applicable (a), and not assessable (−) (tool). The last criterion serves as an auxiliary value for external assessors who do not know the original BPM method engineer’s intention. In summary, the assessment criteria have the following semantics:

If the Assessment Process is not performed by the original BPM method engineer, we recommend ensuring the assessment’s reliability by involving at least two independent judges (technique). Therefore, the Excel prototype calculates Cohen’s Kappa (Cohen 1960) per BPM method (online Appendix 2) (tool). Besides ensuring the reliability of the assessment, we recommend analyzing the assessed BPM methods with respect to their degree of context specificity (DCS), i.e., an indicator to classify whether the BPM method follows a special or a general purpose (technique). This addresses the need for selecting suitable BPM methods that support the efficient use of organizational resources. As no suitable indicator for measuring the DCS is available in the literature, we developed the formula shown in Eq. (1) (tool). An exemplary calculation can be found in online Appendix 2.with \(F\) Set of context factors included in the context dimension of the Classification Framework, \({C}_{f}\) Set of characteristics per context factor \(f\in F\), \({\alpha }_{f}\) Number of characteristics assessed with (a) for context factor \(f\in F\), \(\gamma\) Number of characteristics assessed with (−) across all context factors \(F\)

According to Eq. (1), the DCS represents a weighted fraction of the characteristics per context factor in which a specific BPM method can be applied. To make the DCS comparable across different BPM methods in the case that some characteristics cannot be assessed, the second factor of Eq. (1) adjusts the weighted fraction based on the number of characteristics assessed with (−). Hence, in the absence of characteristics that cannot be assessed, a DCS of 100% means that a BPM method is applicable to only one characteristic per context factor (i.e., special-purpose method), while a DCS of 0% means that a BPM method applies to all characteristics of all context factors (i.e., general-purpose method). Again, activity A4 is performed by a BPM method engineer or method user (role).

Linking the Selection Process to the method attributes (Table 1), it assists BPM method users to select BPM methods applicable to their organizational contexts (goal orientation) and, just like the Assessment Process, relies on the Classification Framework (principle orientation). Comprised of four necessary and one optional step, the Selection Process starts with defining the BPM lifecycle stage to which BPM methods should be applicable. Then, the method’s overall target is determined. After that, the characteristics of the other context dimensions to which BPM methods should be applicable are defined. The last task is to identify and select the BPM method(s) that performs best across all dimensions. An additional optional step accounts for the fact that several contexts need to be considered in a single organization (Kerpedzhiev et al. 2020), so the query of the Method Base can vary depending on the method user’s specific role. For example, a BPM process owner or manager may be searching for BPM methods for a specific process in a defined BPM lifecycle stage, while a Head of BPM or a process portfolio manager may take a multi-context perspective, aiming to identify BPM methods that meets as many contextual needs as possible. As the Selection Process refers to only one context, it must be applied repeatedly in that case (systematic approach). The detailed description of each step (Table 3) supports the execution of the Selection Process in various contexts and among various users (repeatability). Again, we provide deeper insights below. An evaluation of the Selection Process is given in Sect. 5.2.

Defining the lifecycle dimension (S1) involves defining the BPM lifecycle stage to which BPM methods should be applicable according to the Classification Framework (technique). This activity is supported by the Excel prototype (tool). Just like for the Assessment Process, the Excel prototype including a preliminary Method Base of 103 BPM methods can be found in the electronic supplementary material. Activity S1 is usually performed by a process owner or manager (role). All subsequent activities consider only those methods that fit the defined BPM lifecycle stage (output).

Defining the goal dimension (S2) requires defining to which characteristic(s) of the goal dimension BPM methods should be applicable (technique). As in the previous activity, one filters the characteristics of exploitation and/or exploration within the goal dimension (vom Brocke et al. 2016) in line with the Classification Framework (tool). As this decision may be relevant to the organization’s strategy, activity S2 is probably not only performed by a process owner or manager, but also supported by the Head of BPM or similar roles (role). All subsequent activities consider only those BPM methods that fit the defined characteristic(s) of the goal dimension (output).

Defining the context dimension (S3) requires defining to which of the context dimension’s characteristics BPM methods should be applicable (i.e., considering all characteristics of the process, organization, and environment dimensions) (technique). BPM method users must go through all characteristics and decide which characteristics represent their organizational contexts. As all characteristics are mutually exclusive, one characteristic per context factor, at most, can be chosen (e.g., scope: intra-organizational processes or inter-organizational processes). In the Excel prototype, the characteristics to which BPM methods should be applicable are assessed with the value of “1”, all others with “0”. If a method should be applicable to more than one characteristic, the Selection Process must be applied repeatedly. Afterwards, the characteristics can be prioritized by determining relative weights. As the weighting happens on two hierarchy levels (i.e., context factors and dimensions), we draw from knowledge on multi-criteria decision analysis, such as the Analytical Hierarchy Process (Saaty 1990). To reduce complexity, the Excel prototype proposes an initial configuration, assuming that all characteristics are equally important, but the configuration can be changed as required (tool). Since this activity is related to the organization’s strategy, it should be performed by several stakeholders, such as a process owner, process manager, or Head of BPM (role). If many stakeholders are involved, they may use techniques like brainstorming, moderated group discussion, and team estimation games to determine appropriate ratings (Schwaber 1997; Yoo et al. 2009). All subsequent activities consider only those BPM methods that fit the defined BPM lifecycle stage (activity S1) as well as the goal dimension’s (activity S2) and context dimension’s defined characteristics (output).

Selecting BPM method(s) for a single context (S4) requires analyzing all previous results and identifying the most suitable BPM method(s) (technique). Therefore, we recommend analyzing the degree of applicability (DA) of each BPM method for the context at hand. Again, as no suitable indicator is available in the literature, we developed the measure shown in Eq. (2), which reflects the extent to which a given BPM method is applicable to the context specified in the Selection Process (i.e., how often the criteria (a) or (na) match the specified context), considering that context dimensions and factors can differ in importance. In case a BPM method has not been assessed by the original method engineer, we offer two calculation modes. In the risk-averse mode, all ( −) are treated as (na), i.e., context characteristics which could not be assessed based on publicly available data are treated as if the method were not applicable. In the risk-taking mode, however, all ( −) are treated as (a). To prepare the calculation of the DA, all (a) and (na) values are replaced by 1 and 0, respectively. The DA of a given BPM method for a specified context is calculated as shown in Eq. (2). An illustrative example is included in online Appendix 2.with \(D\) Set of dimensions included in the context dimension of the Classification Framework, \({F}_{d}\) Set of context factors per context dimension \(d\in D\), \({C}_{f}\) Set of characteristics per context factor \(f\in F\), \({\omega }_{d}\) Weight of context dimension \(d\in D\) with \({\omega }_{d}\in \left[0;1\right]\) and \({\sum }_{d\in D}{\omega }_{d}=1\), \({\varphi }_{f}\) Weight of context factor \(f\in {F}_{d}\) with \({\varphi }_{f}\in \left[0;1\right]\) and \({\sum }_{f\in {F}_{d}}{\varphi }_{f}=1\), \({\delta }_{c}\) Assessed context characteristic \(c\in {C}_{f}\) of given BPM method with \(\delta \in \left\{0, 1\right\}\), \({\varepsilon }_{c}\) Assessed context characteristic \(c\in {C}_{f}\) of given context with \(\varepsilon \in \left\{0, 1\right\}\)

According to Eq. (2), a DA value of 100% means that a BPM method perfectly fits the specified context, while 0% means that a BPM method is not applicable at all. The DA is meant to be the main evaluation criterion for selecting suitable BPM methods, as it increases inter-subjectivity when comparing methods based on a consistent calculation logic. Depending on the DA, all remaining BPM methods are ranked so users can shortlist the BPM methods that will be subject to a detailed assessment. To support this narrowing-down, the DCS indicator, which we introduced in activity A4, should also be considered to assess whether a BPM method follows a special or a general purpose. Details on comparing BPM methods based on their DCS and DA values are provided in Sect. 5.2. If desired and necessary, organizations can also integrate further criteria like the methods’ ease of use, training effort (Recker et al. 2009), or required upfront investments (Neubauer and Stummer 2007) (tool). Depending on how the BPM method will be applied, we recommend involving all relevant stakeholders in the shortlisting process (role). The result is the identification of BPM method(s) that meet the given context needs, thus taking a single-context perspective (output).

Selecting BPM method(s) for multiple contexts (S5, optional) instructs the user of the Selection Process to perform activities S1, S2, and S3 repeatedly if there are several contexts in one organization to be considered. Unlike activity S4, activity S5 requires comparing multiple results and selecting the BPM method(s) that fit various contexts, not just one context (technique). The DA and DCS of multiple iterations are compared using a cross-context validation. If each iteration yields similar BPM methods in its shortlist, it is not necessary to implement various methods to account for all contexts. If each iteration yields different shortlisted methods, it might be necessary to implement multiple BPM methods. As outlined in activity S4, the DA and DCS are automatically calculated by the Excel prototype to support the selection decision (tool). Activity S5 is performed by roles such as Head of BPM or process portfolio manager (role). The result is the identification of BPM method(s) that meet as many of the context’s needs as possible, thus taking a cross-context perspective (output).

To evaluate the Assessment Process and the Classification Framework, we applied the Assessment Process in two phases. In the first phase, we applied it to a sample of 103 BPM methods (102 BPM methods from the literature and the CAMAS Method) to gain preliminary insights into its ease of use, real-world fidelity, effectiveness, and efficiency. Two co-authors independently assessed each BPM method. This setup is sensible as a two-assessor setting is commonly used in the literature (Montazemi and Qahri-Saremi 2015; Paré et al. 2015; Wolfswinkel et al. 2013) and as the co-authors are both academic BPM method engineers and users, developing new BPM methods and using BPM methods in projects. As organizations typically do not have access to the original engineers of the BPM methods included in the Method Base, we deliberately decided not to involve them in the first evaluation phase.

Following the Assessment Process, activity A1 revealed that all 103 identified BPM methods are suitable for being assessed, as they comprise all method components and apply to at least one BPM lifecycle stage (Table 1, Sect. 2). However, as most BPM methods do not explicitly state each component, the authors discussed their fit based on indications and decided on their suitability for being assessed. Conducting activity A2 to A4 revealed that especially activity A4 was challenging, leading to discrepancies in assessing the applicability of BPM methods. As almost no BPM method explicitly stated its applicability to specific contexts, the assessment is mainly based on indications, which leave room for interpretation and differing assessments. Thus, we only used the criteria (a) and (na) if, in the respective research paper, a method explicitly stated or clearly indicated its applicability to a specific context characteristic. Otherwise, we used the auxiliary criterion (−). In case of disagreement, the authors discussed all mismatches and decided on a single criterion for the final assessment. To account for discrepancies and ensure the reliability of the classification as performed by the co-authors, we calculated Cohen’s Kappa (Cohen 1960). The Kappa ranged from 61 to 100% with an average of 77%, which indicates substantial agreement (Landis and Koch 1977). All assessment results are presented in online Appendix 3. Figure 6 shows two methods per BPM lifecycle stage and goal characteristics assessed along A2 to A4.

In the second phase, we asked original BPM method engineers to assess selected BPM methods based on the Assessment Process. We asked 20 BPM method engineers to cover around 20% of BPM methods per lifecycle stage from our sample (30 in total). All BPM method engineers received a questionnaire (online Appendix 4) including questions related to activity 2 to 4 of the Assessment Process. Additionally, the method engineers were asked to comment on and assess the ease of use of the Assessment Process using a 7-point scale. Overall, we received first-hand classifications from 12 BPM method engineers who assessed 20 BPM methods. To get insights into the validity of the assessment performed by the co-authors, we compared their classification with that performed by the original BPM method engineers and calculated hit ratios (Moore and Benbasat 1991), which measure the frequency of correctly assigned objects (Nahm et al. 2002). In case we used the auxiliary criterion (−), we assumed a match of both assessments, and in case of a different assessment using (a) and (na), we assumed a mismatch. The co-authors achieved hit ratios between 90 and 100%, yielding an average of 97%, which reflects significant agreement (Moore and Benbasat 1991). All assessment results are presented in online Appendix 5.

The application of the Assessment Process 226 times with co-authors and original BPM method engineers showed its effectiveness and efficiency to assess BPM methods in a context-aware manner (use case 1). Even though slight discrepancies between the co-authors occurred when conducting activity A4, a circumstance that would also happen in industry settings, the achieved Cohen’s Kappa values confirm reliability. Moreover, the hit ratios between the co-authors and the original BPM method engineers underpin the validity of the Assessment Process. This also led us to conclude that the Method Base with 103 exemplary BPM methods was a solid basis for evaluating the Selection Process (Sect. 5.2). As for real-world fidelity, the evaluation showed that the assessed methods cover diverse contexts in which BPM methods are applied. Regarding ease of use, the original BPM method engineers assessed the ease of use with 5 out of 7 points. Even though the co-authors stated that the classification of some methods was challenging, it is reasonably easy for BPM method engineers. Most original BPM method engineers confirmed the detailed description of each activity, corresponding techniques, tools as well as definitions per dimension, context factor, and characteristic as sufficient to assess BPM methods. Nevertheless, some of them asked for extended guidelines in the sense of a manual, which we plan to provide when publicly sharing the CAMAS Method after publication. Overall, the evaluation confirmed that the Assessment Process is appropriate to assess BPM methods in a context-aware manner.

Two BPM method users from two different organizations applied the Selection Process to gain insights into its ease of use, real-world fidelity, effectiveness, and efficiency. We chose experts from two organizations that differ widely in terms of their organizational setup. In both organizations, we interviewed the key expert responsible for BPM (Table 4) using qualitative semi-structured interviews (Myers and Newman 2007) along the activities of the Selection Process, also using the Excel prototype. Each interview took about two hours and was attended by two co-authors. After introducing the Selection Process, we asked the experts to select three real processes from their organizations, to apply each of the Selection Process’ activities. Besides, we asked for comments on ease of use, real-world fidelity, effectiveness, and efficiency. To obtain results, we pre-filled the Method Base with the 103 BPM methods we had assessed during the evaluation of the Assessment Process (Sect. 5.1). An overview of the two organizations, their experts, and the processes in focus is presented in Table 4.

We present as an example the results of the evaluation related to the product development process (P4) at the PRODUCT organization (Table 4), while the results of all other processes are shown in online Appendix 6. As P4 has already been designed, implemented, and improved at PRODUCT, the expert searched for a method with which to “monitor” the process (S1). Moreover, PRODUCT requested an exploitative BPM method (S2). Regarding the context characteristics, all relevant characteristics were marked. The context factors and dimensions were also weighted. PRODUCT was more interested in considering the process (0.6) and the organization dimension (0.4) than the environment dimension (S3). Finally, the results were analyzed (S4). The results of activities S1 to S4 are presented in Fig. 7.

To compare the most suitable BPM methods for the given context, all BPM methods are ranked according to the indicator DA, i.e., the extent to which they are applicable to the specified context. PRODUCT calculated the DA based on the risk-averse mode. The results show that the DA of BPM methods assessed by the original BPM method engineers tend to be higher than that of methods assessed by the co-authors. This is because, in the risk-averse mode, all (−) are treated as (na) as no statement about the applicability of a BPM method to a specific context can be made based on publicly available data. Importantly, the Selection Process does not aim at estimating the DA as precisely as possible, but to compare BPM methods based on a consistent calculation logic. Hence, method users applying the Selection Process decide which calculation mode (i.e., risk-averse or risk-taking) they prefer. In concert with PRODUCT’s expert, we analyzed a shortlist of seven methods that reached a DA above 48%, a mix of BPM methods assessed by BPM method engineers and the co-authors. To ensure a detailed analysis, the DCS indicator was considered as well, which ranged between 6 and 24%, indicating general-purpose methods. Accordingly, these BPM methods can be considered sufficient for the context at hand. The expert also stated that no other criteria (e.g., training effort) are necessary to identify a suitable BPM method. Based on the results of applying the Selection Process, PRODUCT examined the top seven methods in detail and selected one of them.

As for effectiveness and efficiency, the experts confirmed that the Selection Process is a well-founded, yet pragmatic, way to reason about how to select BPM methods in a context-aware manner (use case 2). It reduces time and uncertainty in selecting suitable BPM methods and, thus, facilitates an efficient use of resources for their implementation. The experts acknowledged that the Method Base includes not only well-known BPM methods, but also unknown BPM methods that inspire to consider context from various perspectives and support the exploration goal of BPM. As for real-world fidelity, the experts confirmed that the Selection Process is suitable for various contexts as the multi-dimensional architecture of the Classification Framework allows for a comprehensive analysis. As for ease of use, the experts confirmed that the activities of the Selection Process are understandable for people typically involved in BPM. The experts particularly appreciated the detailed description of each activity, corresponding tools (e.g., Excel prototype), and definitions of each dimension, context factor, and characteristic (vom Brocke et al. 2016). Even though the determination of the relative importance of factors and dimensions (S3), the analysis of DA and DCS (S3), and the cross-context validation of various BPM methods (S5) was challenging, the experts appreciated the possibility to prioritize specific contexts and analyze the results in detail. Additionally, the experts emphasized the need of activity S5 as due to scarce resources and training effort for new BPM methods, it is indispensable to select BPM methods that fit various contexts. To overcome these challenges, techniques like brainstorming, moderated group discussion, and team estimation games (Schwaber 1997; Yoo et al. 2009) are appreciated. Nevertheless, the experts also saw room for improvement in the visualization of the Excel prototype, e.g., by directly providing access to definitions of all dimensions and descriptions of all BPM methods. We outline respective ideas for future research in Sect. 6.3.

The literature review (online Appendix 1) we used to evaluate the Classification Framework and the Assessment Process also provided general insights into the applicability of the identified 103 BPM methods to specific contexts. We structured these insights along the lifecycle, goal, and context dimension (Fig. 8). First, when analyzing the 103 BPM methods per lifecycle stage (Fig. 8A), we found that BPM methods for implementation (n = 10), improvement and innovation (n = 16), and project management (n = 12) are rare compared to other BPM lifecycle stages. These findings confirm prior investigations, indicating that these stages still need more attention from BPM researchers (Recker and Mendling 2016). Second, regarding the goal dimension (Fig. 8B), we found that 102 BPM methods only apply to exploitation, while two BPM methods are for exploration only and seven fit both exploitation and exploration. Also these findings comply with prior studies, indicating the lack of explorative BPM methods (Gross et al. 2019; Kohlborn et al. 2014). Third, regarding the context dimension, we investigated the DCS (Fig. 8C). The DCS ranges between 0 and 45% with an average of 16% indicating that most methods included in our sample rather follow a general-purpose approach. This finding supports our research problem and calls for further research (Sect. 6.3).

Existing research offers limited insights into the question how BPM methods can be assessed and selected in a context-aware manner. In particular, research on the application possibilities of BPM methods and their design for specific contexts is missing (Dumas et al. 2018; Rosemann and vom Brocke 2015; vom Brocke et al. 2016). At the same time, guidance on how to assess and select BPM methods in a context-aware manner is lacking (Zelt et al. 2018). The CAMAS Method proposed in this research is the first to conceptualize and operationalize the context-aware assessment and selection of BPM methods, so it contributes to both theory and practice.

We distinguish two theoretical implications that add to the descriptive and prescriptive knowledge on context-aware BPM. Regarding descriptive knowledge, our study extends the assessment scheme by Denner et al. (2018a) in two different ways: first, the Classification Framework included in the CAMAS Method draws from the original assessment scheme and extends the context perspective (vom Brocke et al. 2016) through the lifecycle perspective (Rosemann and vom Brocke 2015). With the combination of both perspectives, we expect a more precise assessment of existing or newly developed BPM methods. Thereby, the Classification Framework’s modular design enables high flexibility. As existing context dimensions can be dropped or new ones added easily, we account for future developments of context-aware BPM. Second, we extended the number of BPM methods in the Method Base from 25 (Denner et al. 2018a) to 103 by applying the Assessment Process (Sect. 5.1). Analyzing the status quo of this sample as a byproduct of our evaluation, we revealed various insights into existing BPM methods. Regarding the lifecycle perspective, we found that BPM methods for implementation, improvement and innovation, and project management are rare compared to the other lifecycle stages. As for the goal dimension, we identified a lack of explorative BPM methods. While existing research introduces ambidextrous BPM only on a high-level abstraction, we identified a few specific BPM methods recognizing and confirming the importance of distinguishing exploitative and explorative BPM methods. Finally, investigating the context perspective revealed an overall lack of context-specific BPM methods. So far, BPM method engineers have not explicated specific application contexts for most existing methods. These insights serve as a starting point for further discussions that will strengthen research on context-specific BPM methods (Sect. 6.3).

Regarding prescriptive knowledge, our study used SME to build an ensemble artifact (i.e., the CAMAS Method), which not only includes the Classification Framework but also offers guidance on its application in terms of a newly developed Assessment and Selection Process. In the end, the CAMAS Method helps compare BPM methods based on a common set of context-related dimensions and characteristics. The Assessment and Selection Process provide guidance on how to assess BPM methods regarding their applicability to specific contexts (use case 1) and on how to select BPM methods that fit their contexts at hand (use case 2). As the CAMAS method is not an entirely new end-to-end method but composes existing fragments against the background of context awareness in BPM, it accounts for the DSR contribution type ‘exaptation’, extending known solutions to new problems (Gregor and Hevner 2013).

Further, our work has several managerial implications. First, our study supports BPM method engineers as it facilitates the targeted application of their methods, which may increase adoption in practice. Second, our study guides practitioners in assessing the applicability of existing BPM methods to specific contexts. As the CAMAS Method helps understand the nature of BPM methods in a structured and well-founded manner, practitioners may challenge the applicability of BPM methods currently used in their organizations. Third, our study guides practitioners in selecting suitable BPM methods for specific contexts, reducing risks related to an inefficient use of resources or, in some cases, the failure of BPM projects (Sect. 1). The Method Base comprising 103 BPM methods can also inspire organizations to use new, perhaps locally unknown BPM methods. Finally, our study helps practitioners handle multiple contexts at the same time by applying the Selection Process repeatedly, comparing results, and selecting those BPM methods that fit various contexts. Overall, our findings reduce the uncertainty related to the selection of BPM methods and increase the transparency of related decisions.

One limitation has to do with the assumptions we made for the CAMAS Method (Sect. 4) to reduce complexity. First, we assessed each context factor on a two-point scale, dropping medium-level characteristics (e.g., medium knowledge intensity). Future research may include more fine-granular specifications of the context characteristics, allowing for a more detailed assessment of BPM methods’ applicability. Second, we propose two indicators, i.e., the DA and DCS, to support the selection of suitable BPM methods. Future research may investigate additional indicators to enhance BPM method selection.

The evaluation of the CAMAS Method also has limitations (Sect. 5). First, two co-authors independently assessed a sample of 103 BPM methods and 12 BPM method engineers were involved in obtaining first-hand classifications of 20 BPM methods. Future research may involve all original method engineers in the assessment of existing methods. Second, the evaluation builds on a structured literature review of articles in recognized journals and conferences in the BPM and IS discipline. However, we consciously decided not to include BPM methods from (text) books and/or consulting companies. This design decision had two reasons. On the one hand, we did not aim for a complete sample of BPM methods, as the identified methods primarily served the purpose of evaluating the Assessment Process and as a basis for applying the Selection Process. This purpose has been confirmed by applying the Assessment Process 226 times and the Selection Process six times with two organizations for real processes. On the other hand, we could not rigorously decide which (text) books or consulting methods should be included, as many consulting methods are not publicly available. Thus, we started with those BPM methods that have been published in well-known and high-ranked journals and conferences in the field of IS and BPM. To address both limitations mentioned above and to facilitate a first-hand classification by BPM method engineers, we provide the current Method Base and the Excel prototype for the Assessment and Selection Process in the electronic supplementary material. That way, many people can contribute to extending our compilation of existing BPM methods. Third, we evaluated the Selection Process in two organizations to gain preliminary insights. Future research should involve additional organizations from various contexts. Finally, future research may also further develop the prototype in terms of visualization and analysis functionality. For example, to automatize activity S5, a decision model could be proposed and implemented which automates the compilation, valuation, and selection of alternative combinations of BPM method(s) considering multiple contexts. To facilitate the real-world application of the Selection Process and developments of the prototype, we publicly shared the respective Excel prototype.

Beyond addressing limitations and future research, we make a call for action. Our key findings related to the status quo of BPM methods disposes us to call for more context awareness in BPM method design. In particular, we request that BPM method engineers assess the applicability of their methods to specific contexts when the method is being designed. We also call for the development of context-specific BPM methods that, for example, address frequent combinations of context characteristics. Additionally, an extension of explorative BPM methods is required, as they become more important in today’s dynamic business environments (Grisold et al. 2019) and as most extant BPM methods are exploitative in nature. In doing so, BPM method engineers may consider methods from other disciplines (e.g., innovation management, design thinking, product engineering) to derive explorative methods applicable for BPM purposes. Besides developing new context-specific BPM methods, we call for broadening the knowledge about context-aware BPM with respect to additional context dimensions (e.g., customer dimension), for supplementing the proposed dimensions of the BPM lifecycle and the BPM context framework, or by expanding our approach to other core elements of BPM (e.g., governance, culture, or strategic alignment).