This study aims to analyze the end-to-end management of complex processes by focusing on process efficiency, data utilization and automation potential to optimize value addition and business outcomes. It addresses an environment in which traditional workflow-centric process optimization is challenged by trends that significantly increase the amount of data in service and service delivery processes.
The approach involves analyzing process management in the insurance field by combining scoping reviews, qualitative semistructured interviews and data stream mapping.
The key findings indicate that the effective management of complex processes end-to-end requires a paradigm shift from a workflow-centric approach to dataflow-centric process management to achieve end-to-end effectiveness and efficiency. A dataflow-centric approach shows a better potential for data quality for process decision points, which would support decision-point automation. The downstream impacts of data quality and automation capability on downstream processes are highlighted as significant considerations. The importance of addressing data quality among the process effectiveness metrics is highlighted.
Originality involves a proposal for approaching a data-centric model that bridges the gap between workflow-centric process management and the complex data-intensive demands of modern service delivery processes. This is an environment with increasing amounts of data in service and service delivery processes.
1. Introduction
Traditionally, companies have been structured into functions dedicated to specific business processes, making the management and optimization of these processes a significant challenge for maintaining operational efficiency and agility (Telukdarie, 2019). Process optimization has traditionally focused on managing process workflows and leveraging methodologies, such as lean process management, to enhance performance (De Ramón Fernández et al., 2020). Although these approaches have brought incremental improvements, they are constrained by a persistent reliance on functional silos, which limits their ability to address broader system-level inefficiencies. This siloed optimization is particularly prevalent in service and service delivery processes, which are predominantly workflow-centric. Process automation has been introduced to increase efficiency and reduce costs (Blahušiaková, 2023). However, as the complexity and volume of data increase, traditional workflow-centric approaches face critical limitations in addressing the demands of modern service environments.
Three transformative trends disrupt traditional workflow-centric process optimization, particularly in service and service delivery contexts: the shift toward customer-centric end-to-end (E2E) processes, the increasing demand for service customization and the rapid adoption of service automation. These trends have driven an unprecedented accumulation of data, challenging the ability of workflow-centric models to effectively scale and adapt. In cross-functional services such as service and service delivery processes, organizational boundaries are crossed (Enz and Lambert, 2023), resulting in E2E processes that consolidate vast amounts of data (Abeysekara et al., 2023; Maddern et al., 2014). Customization further intensifies data demand by introducing requirements for service configuration, personalization and rule management (Pech and Vrchota, 2022). Service automation exponentially increases the need to capture, store and analyze data, adding to the complexity of process optimization (Willcocks et al., 2023).
As service and service delivery processes become increasingly data-rich, they involve the management of data related to the service configuration, delivery and customer profiles. Traditional workflow-centric models that emphasize task execution rather than data utilization struggle to address the dual challenge of optimizing both dataflow and process efficiency. Moreover, the potential automation in these processes often relies on decision trees, which require the integration of vast quantities of historical and reference data to build accurate automated decision models (Ng et al., 2021). This surge in data and decision-making complexity necessitates the reconsideration of process management logic to achieve E2E efficiency while effectively leveraging data.
Despite the growing importance of dataflow-centric approaches in these contexts, limited research has addressed how such frameworks can optimize the E2E process efficiency and automation in data-rich environments. Existing literature has primarily focused on workflow-centric methodologies, with insufficient exploration of how data-centric approaches can better support the demands of modern service processes. This creates a critical gap in our understanding of inhibitors and opportunities associated with transitioning to dataflow-centric process management.
The present study seeks to bridge this gap by analyzing how to manage complex processes E2E, considering process efficiency, data utilization and automation potential to optimize value-added and business outcomes. Specifically, it aims to address the following research questions:
Which process or data management approach best serves the E2E process efficiency and automation?
What are the main inhibitors of a data-centric process-management approach?
The remainder of this paper is organized as follows. First, the Section 2. Literature review examines complex E2E processes with a focus on data-centric and workflowcentric process management approaches and their impact on optimizing value-add. Second, it empirically analyzes process management in the insurance industry in Section 4, emphasizing data utilization and its implications for outcomes. Finally, the findings are presented and discussed in the context of advancing process-management methodologies in Sections 4 and 5.
2. Literature review
Previous significant work on process management includes frameworks such as Lean Six Sigma (Lameijer et al., 2024; Näslund, 2008), business process management (BPM) (Bazan and Estevez, 2022; Reijers, 2021) and other traditional models (Feversani et al., 2022). These methodologies are widely used for optimizing operational processes by identifying inefficiencies, reducing waste and ensuring alignment with organizational goals. For instance, Lean Six Sigma is valued for its ability to streamline operations and enhance decision-making by addressing inefficiencies and waste (Lameijer et al., 2024). On the other hand, BPM emphasizes the design, monitoring and optimization of business processes to ensure task efficiency and process quality (Bazan and Estevez, 2022).
Despite their effectiveness in improving workflow-centric processes, these approaches have limitations when applied to data-rich cross-functional processes. Lean Six Sigma, while successful in achieving efficiency gains, struggles to address the real-time flow and utilization of large-scale, diverse data in E2E processes (Antony et al., 2017; Koppel and Chang, 2021). Similarly, BPM, with its focus on workflow and siloed task execution, often fails to capture the dynamic and complex nature of dataflows across organizational boundaries, which are central to today’s interconnected service environments (Pech and Vrchota, 2022).
Complex processes in modern service delivery involve multiple stakeholders (Hvam et al., 2019; Walsh and Gordon, 2010; Zhou et al., 2023) require dynamic customization to meet customer needs (Danaher and Mattsson, 1998; Walsh and Gordon, 2010; Hvam et al., 2019; Zhou et al., 2023) and demand seamless coordination among participants (Walsh and Gordon, 2010). Effective management is essential to ensure both quality and efficiency (Walsh and Gordon, 2010). These processes also require the evaluation of multiple attributes, reflecting significant data requirements (Danaher and Mattsson, 1998) and face challenges related to coordination, governance and integration (Danaher and Mattsson, 1998; Hvam et al., 2019; Zhou et al., 2023). Workflows often span organizational boundaries and increase complexity (Danaher and Mattsson, 1998; Hvam et al., 2019; Zhou et al., 2023). Ensuring trustworthy data is critical for maintaining accuracy and effectiveness (Hvam et al., 2019; Zhou et al., 2023).
Process complexity spans governance, managerial and operational levels, with distinct responsibilities and challenges at each level. At the governance level, Zhou et al. (2023) and Hvam et al. (2019) emphasize ensuring data accuracy, transparency and strategic decision-making for long-term process integrity. Danaher and Mattsson (1998) highlight the role of governance in tailoring processes to balance long-term goals such as loyalty and profitability, while Walsh and Gordon (2010) focus on aligning professional identities with strategic objectives.
At the managerial level, Zhou et al. (2023) and Hvam et al. (2019) discussed managing roles, resources and workflows, fostering collaboration and ensuring regular updates for continuous improvement. Danaher and Mattsson (1998) emphasized monitoring, evaluating and optimizing process attributes, while Walsh and Gordon (2010) addressed aligning professional behaviors with client needs and organizational strategies.
Zhou et al. (2023) and Hvam et al. (2019) addressed task execution, data dependencies and resource management within a defined scope at the operational level. Danaher and Mattsson (1998) stressed tailoring execution to align with process-specific priorities and customer needs, whereas Walsh and Gordon (2010) focused on the nuances of individual service exchanges and client-specific adaptations.
Finally, complex processes must be responsive to changing conditions (Hvam et al., 2019), require cross-channel coordination to address evolving customer expectations and leverage technology for optimization (Hvam et al., 2019; Zhou et al., 2023).
A definition synthesized for complex processes: A complex process is characterized by the involvement of multiple stakeholders; substantial data requirements; challenges related to coordination, governance and integration; and workflows that span organizational boundaries. In modern service delivery, these processes encompass dynamic customization, real-time responsiveness and the need for cross-channel coordination to address evolving customer expectations. Process complexity spans the governance, managerial and operational levels, manifesting differently at each level: strategic decision-making at the governance level, resource and performance management at the managerial level and task execution at the operational level. Effective management often requires leveraging technology, such as automation and analytics, to optimize decision-making and adapt to the rapidly changing service landscape.
As modern service delivery processes grow increasingly complex and data-intensive, driven by automation, customization and the shift toward customer-centric E2E processes, traditional methodologies are often insufficient. A dataflow-centric process management model addresses this gap by focusing on the integration and utilization of data throughout the process lifecycle, enabling improvements in efficiency, automation potential and business outcomes.
The remainder of this literature review explores how process management approaches, including data-centric and workflow-centric models, influence process efficiency, data utilization and automation potential. This involves analyzing the management of complex E2E processes, highlighting the limitations of traditional methodologies and assessing the metrics necessary for optimizing value-added outcomes.
2.1 Data-centric process management
Data-centric process management is an approach to managing business processes by focusing on the central role of data within these processes (Bhattacharya et al., 2009). The focus was on the flow and manipulation of the data throughout the process (Haddar et al., 2016). Data are integrated into all aspects of process design, execution and optimization (Reijers et al., 2017). This involves designing processes around the data flow while ensuring data quality and consistency (Liu et al., 2017; López Martínez et al., 2021). Data-centric process management leverages data to improve processes within an organization and is also performed by data-driven process management. However, there is a difference in the emphasis of these two concepts. Data-centric process management emphasizes the central role of data in process design and execution (Bhattacharya et al., 2009), whereas data-driven process management focuses on using data to drive decision-making and optimization (Czvetkó et al., 2022). Other related concepts include data-centric workflow management, which applies data-centric principles (Kougka et al., 2018) but has a narrower scope. Additionally, data-centric BPM emphasizes the role of data in designing, executing and optimizing business processes (van der Aalst et al., 2015), whereas data-centric process management can be seen as having a broader scope, encompassing organizational processes more widely. Table 1 illustrates the key aspects of data-centric process management.
Key aspects of data-centric process management
| Key aspects | Description | References |
|---|---|---|
| Leveraging data | Data are leveraged at every stage of the process lifecycle, including design, execution, improvement and optimization. The primary focus is on data rather than activities and tasks | Andrews et al. (2021); Di Ciccio et al. (2015); Eshuis (2023); Eshuis and Van Gorp (2016); Reijers et al. (2017); Russo and Mecella (2013); Steinau et al. (2019a, 2019b) |
| Data modeling | Data models define relationships and attributes of the data used in the process | Andrews et al. (2020, 2021); Di Ciccio et al. (2015); Russo and Mecella (2013); Snoeck et al. (2023) |
| Data integration | Data are integrated from various sources and systems to support process execution | Di Ciccio et al. (2015); Russo and Mecella (2013); Stahl et al. (2023); Snoeck et al. (2023) |
| Data governance/ data quality | Data governance ensures that data meet defined standards and business rules. Emphasis is placed on data quality, security and compliance | Cappiello et al. (2013); Mahanti (2019); Paasivirta et al. (2025); Reis and Kenett (2018); Stahl et al. (2023) |
| Data-driven decisions | Decision-making is guided by data insights, with real-time analytics, reporting and visualization playing a key role | Hannila et al. (2019b, 2022b); Noh (2018); Steinau et al. (2019b) |
| Flexible processes | Process flexibility is critical, focusing on data rather than the sequence of activities. Processes are adapted to evolving business requirements | Di Ciccio et al. (2015); Eshuis and Van Gorp (2016); Russo and Mecella (2013); Steinau et al. (2019a, 2019b) |
| Automation | Repetitive tasks are automated to improve efficiency, reduce errors and streamline processes | Di Ciccio et al. (2015); Eshuis and Van Gorp (2016); Russo and Mecella (2013); Steinau et al. (2019b) |
| Orchestration of dataflow | Comprehensive consideration is given to dataflow orchestration to avoid data-related challenges | Hannila et al. (2019a, 2022a, 2022b) |
Source(s): Authors’ own work
Data-centric business process modeling is closely related to data-centric process management, as both approaches emphasize integrating data into business processes to enhance efficiency, effectiveness and decision-making. This methodology adopts a data-centric perspective to ensure that data are effectively managed, used and transformed within business operations (Rietzke et al., 2021).
A distinguishing feature of data-centric business process modeling is its ability to obtain the same data through multiple pathways; different tasks within the process can generate or lead to the same data (Maletzki et al., 2019). This flexibility highlights the centrality of data in designing and optimizing business workflows, allowing for redundancy and robustness in data collection and usage.
Challenges in data-centric process management include the usability of data-centric approaches, which, while valuable, are often considered a concern (Snoeck et al., 2023). Fundamental challenges may arise in determining the processes that exist, their interrelations and how to effectively coordinate them (Steinau et al., 2021). Adopting a data-centric perspective can introduce additional complexity to process management (Steinau et al., 2019b). Key issues such as data quality, consistency, integration and governance are anticipated hurdles that require robust frameworks and strategies to address them.
Lean Six Sigma can play a pivotal role in addressing these challenges in data-centric process management. Although its significance in data-rich environments has been acknowledged (Antony et al., 2017; Koppel and Chang, 2021), detailed discussions of its specific application in such contexts remain limited. The potential to adapt and enhance the Six Sigma framework for complex, data-intensive problems has been recognized (Fahey et al., 2020).
Lean Six Sigma methodologies can support efficiency gains (Maia et al., 2024; Rajić et al., 2023), improve decision-making (Belhadi et al., 2021), reduce costs and enhance customer satisfaction. However, further exploration is required to fully understand its application and impact on data-centric process management.
In summary, data-centric process management plays a crucial role in optimizing business operations by leveraging the data. The key takeaway is as follows:
Key insight: Data-centric process management adopts a broad scope encompassing organizational processes to optimize and manage business operations by leveraging data. In this approach, data serves as a fundamental component of business processes, with workflows designed around data to enhance flexibility and drive efficiency.
2.2 Workflow-centric process management
Workflow-centric BPM entails managing business processes by emphasizing the detailed sequencing, coordination and optimization of tasks and activities within workflows to reach specific outcomes (Arantes et al., 2023; Cheung and Hidders, 2011). This approach is characterized by enterprise-wide integration, strategic alignment, process mapping, automation and governance. Enterprise-wide integration ensures that workflow principles are applied across all departments and functions (Vallejo et al., 2011), whereas strategic alignment focuses on ensuring that workflow supports broader business objectives and strategies (Lizano-Mora et al., 2021). Process mapping emphasizes E2E process visualization, which spans multiple workflows and departments (Stark, 2024). Automation, using workflow automation tools, streamlines various processes (Bolcer and Taylor, 1998; Szelągowski and Berniak-Woźny, 2024). Governance plays a critical role by assigning responsibilities to ensure standardization and compliance (Rosemann and vom Brocke, 2015).
The workflow-centric approach offers several advantages. It contributes to efficiency gains and improved organizational performance (Reijers et al., 2016) and facilitates the holistic optimization of business processes (Zhang and Perry, 2014). It ensures alignment with organizational goals and strategies (Ubaid and Dweiri, 2020) and enhances collaboration and coordination across departments, depending on the BPM structure (Szelągowski and Lupeikiene, 2020). Moreover, it improves the scalability and adaptability of business processes while increasing the visibility of process performance and its overall impact (Bartlett et al., 2023).
However, this approach presents several challenges. Integrating systems and processes can be complex and requires strong governance to ensure standardization and seamless operations (Alotaibi and Liu, 2017). The increasing volume and variety of data, combined with the complexity of operations, further strains the scalability of the workflow-centric BPM. As workflows become more complex, designing and executing them become increasingly time-consuming, making handling errors and workflow adjustments more challenging (Filatov and Kantere, 2018). These limitations highlight the need to address the challenges posed by the growing amount of data in workflow-centric process management.
2.3 Impact of process metrics
Metrics play a critical role in managing business processes by helping organizations understand, evaluate and improve their operations (Van Looy and Shafagatova, 2016). Consequently, metrics are one of the primary drivers of operational management decisions (Amzil et al., 2022). As illustrated in Figure 1, process management applies three primary metrics: effectiveness, efficiency and quality. These metrics are interdependent and must be balanced to achieve an optimal process performance:
Effectiveness measures whether a process delivers the intended value-add (Davidow, 2018).
Efficiency includes both resource efficiency (cost-efficiency) and time efficiency (cycle time) (Afy-Shararah and Rich, 2018; de Mast et al., 2011; Vera-Baquero et al., 2022).
Quality refers to the ability of a process to satisfy predefined quality standards (Vanderfeesten et al., 2007).
Figure 1 visually represents the relationship between these metrics, with the effectiveness at the center of the triangle. This central placement highlights the fact that effectiveness is the ultimate goal of any process, with both efficiency and quality contributing to its achievement. For instance, neglecting quality may result in failure of the process to meet its intended value, thereby undermining overall effectiveness. Similarly, insufficient time efficiency can delay value delivery, directly impacting the effectiveness.
By maintaining a balance across these metrics, organizations can improve their processes without sacrificing one area for another. The size of the circle for effectiveness in the center also holds significance, as it relates to the necessary quality and affects how the efficiency metrics are defined. This emphasizes that metrics must align with the intended outcomes, with effectiveness being a key driver.
Process improvement areas can be identified by applying these relevant metrics through methodologies such as lean (Breen et al., 2022) and Lean Six Sigma (Oliver et al., 2019) or by benchmarking against industry standards or competitors (Siha and Saad, 2008). Additionally, methods such as robotic process automation (RPA) (Siderska, 2020) and artificial intelligence (AI)-based process automation (Javaid et al., 2022) can play a significant role in enhancing process efficiency, effectiveness and quality.
2.4 Synthesis: value-add drivers
A proper focus is essential for optimizing value addition or business outcomes. Generally, processes can be modeled in multiple ways (Figure 2). The most common approaches include material flow, workflow and dataflows. Process models originating from the manufacturing industry typically describe processes as material flow. As nonmanufacturing processes have gained importance, the concept of workflow has become more prominent. In addition to the increasing process complexity, control processes that guide manufacturing and service processes have become increasingly intricate. These control processes are primarily composed of decision points and are best modeled as decision trees. When decisions are low in complexity, processes that consist mainly of decisions can still be effectively modeled as workflows. However, when decisions are more complex and require data from multiple sources, modeling processes such as dataflows become more relevant. This reflects the idea that data are not just a process resource but also a primary asset. A process comprising decisions is driven by data and the process output depends heavily on the data used in the decisions.
In workflow-centric process management, actions are seen as the primary drivers of value-added, while data (data objects) are considered merely a resource. Value-added and business outcomes are optimized primarily by optimizing actions. In contrast, data-centric process management focuses on the data objects (data) used throughout the process workflow. In this approach, value-added is derived from the completeness and consistency of the data objects. The primary goal of the process is to assign attribute values to data objects and create relationships between them. The enriched data ultimately constitute the value added. Based on this enriched data, decisions are made that drive value for both the customer and the company. Data are treated as primary assets in a data-centric process-management approach. Instead of modeling the business logic within the workflow, this is reflected in the data streams. Therefore, the process workflows are shaped by the data streams. Process management focuses on creating and maintaining consistent and complete data objects, thereby ensuring that data lineage remains intact. This data lineage is mirrored in the process workflow as an unbroken audit trail.
Finding: In data-centric process management, enriched data constitute value-add. The focus is on the data objects (data) used in the process. Assigning attribute values to data objects and creating relationships between them are essential. Decisions are made based on the enriched data to realize value for both the customer and the company.
3. Research process
This study adopted a qualitative conceptual research design to explore the management of complex E2E processes and their associated data streams. The analysis was conducted in the context of a selected insurance company that forms the nonlife insurance (NLI) division of a larger Finnish financial services group. This financial services group, a market leader in Finland, comprises the divisions of corporate and retail banking, life insurance and NLI. The chosen company was particularly suitable for this study because of its ongoing large-scale system renewal, which served as a critical driver for process redesign and automation projects. The company offers a unique opportunity to investigate E2E process management in a real-world, data-rich environment. Furthermore, extensive access to key personnel and internal documentation enhanced the depth of analysis and supported triangulation.
Figure 3 outlines the research process, which integrates multiple qualitative methods to achieve the objectives of the study. These include a limited scoping review of the literature (Munn et al., 2018), semistructured interviews (Merton et al., 1990), data stream mapping and a review of internal company documentation. This combination of methods ensures a robust and triangulated understanding of process-management challenges and opportunities.
3.1 Research steps
3.1.1 Literature review.
The study began with a focused scoping review to clarify key concepts such as E2E process management, data utilization and relevant process metrics. Articles were identified using keyword searches, followed by selection based on their relevance to the research objectives. The literature review provides a theoretical foundation for understanding the interplay between workflow-centric and dataflow-centric management approaches. Although this review primarily addressed theoretical concepts, it laid the groundwork for identifying gaps in the current approaches and the potential benefits of integrating data-driven models into process management.
3.1.2 Empirical context and focus.
The empirical study focused on customer-initiated claims processes selected for their cross-channel and cross-functional characteristics. These processes span multiple customer interaction points (e.g. mobile, online, phone and agent channels) and organizational units (e.g. customer service, claims handling, investigative services and payment services). This focus is justified by the following three factors:
Claims processing is the most personnel-intensive customer process, offering significant potential for cost savings through automation.
The claims process involves numerous decision-making steps, making it ideal for assessing the role of automated or analytic-aided decision-making.
Initial company assessments identified the claims process as having the highest automation potential, enabling a clear comparison between the workflow-centric and dataflow-centric management approaches.
By selecting a claims process that crosses multiple functional boundaries and involves substantial decision-making, this study indirectly compares the challenges faced in workflow-centric and dataflow-centric management models. The intention was to explore how the inherent differences in these models impact process efficiency, decision-making and automation potential.
3.1.3 Interviews and data collection.
Semistructured interviews (Merton et al., 1990) were conducted with a diverse group of stakeholders, including process owners, business leads, data scientists, analysts and architects. The wide range of roles ensured the comprehensive coverage of both process management and dataflow perspectives. A total of 24 interviewees participated. The following roles were interviewed:
Tribe lead, claims shared capabilities;
Business lead, claims shared capabilities;
Process owner, claims;
Process development managers, claims;
Data scientists and analysts;
Business analysts;
Solution analysts;
Product owners (data warehouse managers); and
Data leads and information architects.
These interviews provide invaluable insights into the practical challenges and opportunities that arise when managing data-rich E2E processes. Respondents’ perspectives were used to triangulate findings from other data sources, strengthen the analysis and allow for a more holistic view of the claims process.
3.1.4 Documentation review and data stream mapping.
Internal documentation was reviewed to provide additional context and validate the interview findings. This included:
process descriptions and standard operation procedures for key steps in the claims process;
reports on claims processing performance and automation initiatives; and
architecture diagrams (system and business architectures).
The documentation informed us of the data stream mapping exercise, which involved more than 100 employees. This mapping identifies and visualizes key data flows within the claims process, supporting the analysis of dataflow management in the E2E processes. MS Excel was used to support data stream mapping to compare data attributes against decision-making points. By mapping the data streams, this study aims to reveal how data are used across processes, potentially highlighting areas where workflow-centric models fall short of capturing the dynamic movement of data.
3.1.5 Data analysis.
The data analysis followed a structured approach, applying manual thematic coding to groups and categorizing responses from the interview data. The authors independently reviewed the interpretations and any discrepancies or ambiguities were discussed collaboratively to ensure clarity and consensus. Conflicting visions within the interview data were analyzed to identify their sources and categorized by root cause to minimize the influence of subjective perceptions. Triangulation was used to validate the findings by comparing the insights from different data sources, thereby enhancing the robustness and reliability of the analysis.
3.1.6 Rationale for the research approach.
This methodology was applied to address the complexities of the data-rich E2E processes in modern service environments. By combining literature insights, stakeholder perspectives and empirical data, this study provides a comprehensive analysis of how traditional workflow-centric management compares with dataflow-centric approaches. While this study does not explicitly apply both models in a side-by-side comparison within a controlled framework, the focus on claims processing, an area identified by the company as having the highest potential for automation, offers a practical setting to assess the advantages of a dataflow-centric approach. The study’s empirical focus on customer-initiated claims processes ensures that the analysis remains grounded in real-world challenges while addressing broader implications for process automation and efficiency.
Additionally, the mixed use of interviews, internal documentation and mapping techniques ensured methodological rigor and reliability. The approach acknowledges the complexity of the insurance claims process, which involves both task-oriented workflows and data-centric decision-making and seeks to illuminate the interplay between these dimensions. Although this study does not directly compare both models in a single experiment, a combination of methods provides a nuanced understanding of how each model contributes to the management of complex processes in practice.
4. Analysis results
The results obtained from the study were a mix of expected outcomes and surprising discoveries. It was anticipated that fragmented process management would pose challenges to the design and management of the E2E process. However, the downstream impact of the data generated by these processes is more significant than expected. Additionally, although the fragmented system landscape was predicted to create challenges for data management, its effect on the automation capability and dataflows in downstream processes was far more pronounced than expected. The following will delve into the analysis results related to process management and data governance from various relevant perspectives.
4.1 Process management
Process management was analyzed from the E2E viewpoint, with a focus on process efficiency and its impact on overall costs. The two main drivers of process management at the company were 1) process efficiency and 2) customer experience. Process efficiency was prioritized as the main driver for cost savings, with all subprocesses evaluated based on both time- and resource-based efficiency metrics (see Table 2). These metrics were used to incentivize performance, ensuring that process owners, managers, developers and claims-handling agents aligned with the company’s goals.
Process metrics
| Process efficiency metrics | Customer experience metrics |
|---|---|
| # calls answered per hour | Waiting time before a call |
| # calls answered per agent | # days required for claim processing |
| # claims tickets closed per hour | Speed of claim payment |
| # claims tickets closed per agent | |
| Qualitative metric: employee satisfaction (use NPS survey) | Qualitative metrics, e.g. net promoter score (NPS) |
Source(s): Authors’ own work
Finding: The main focus on efficiency through time- and resource-based metrics (e.g. calls answered per hour and claims tickets closed per agent) has driven improvements in operational performance.
4.1.1 Fragmentation of process management.
The claims processing process initially consisted of three subprocesses: claims registration, handling and settlement. Figure 4 illustrates this initial process, in which the relevant systems are linked to each subprocess. The claims registration and handling system (OPAS) was used by the agents for both the initial registration and claims handling. Another system, LARE, was specifically used for claim settlement. Initially, claims registration was performed by phone or in person, requiring direct interaction with an insurance agent. During these early stages, both claims registration and handling phases were managed using the same system (OPAS).
Later, claims registrations were made available online. This shift introduced a dedicated system for the online registration of claims, separate from the phone-based system. While both the online and phone channels had dedicated systems for registering claims, the claims-handling phase remained largely unchanged using OPAS. In the claims-handling phase, a claim was analyzed and compared against the insurance policy as well as other relevant terms and conditions. A decision regarding the claim settlement was then made based on this analysis.
Originally, all claim processing was conducted manually using claims-handling agents. Over time, several changes and enhancements have been introduced to increase process efficiency. These include strategic partnerships with service providers to process claims externally up to a certain EUR limit, which covers car insurance and health insurance claims. Notably, externally processed health insurance claims are limited to nonaccident types. These external service providers have contributed to improving operational efficiency.
In addition, fraud detection analytics were implemented to categorize claims into three distinct groups: 1) claims that were valid and below a certain EUR limit; 2) claims that were suspected to be fraudulent; and 3) claims requiring manual processing.
Claims in the first category were automatically approved and forwarded to the claims-settlement phase. Claims in the second category were referred to the investigation department for further examination, whereas third-category claims were sent to claims-handling agents for manual processing. These changes aimed to enhance process efficiency, reduce process costs and improve customer experience by reducing claims processing time and accelerating claims settlement payments.
Figure 5 illustrates the evolution of the claims processing process, showing the transition from manual to automated handling with the integration of fraud detection and external partnerships. As the claims handling process evolved, process efficiency goals were met but at the cost of increased process complexity. The introduction of automated handling required the implementation of an intelligent claims handling (ICH) system. Additionally, the deployment of dedicated systems for each subprocess further complicates the system landscape, leading to fragmentation. This fragmentation meant that process management efforts were often focused primarily on individual subprocesses, each served by its own operational or support system. As a result, the E2E process management view was lost and efforts were concentrated on improving each subprocess independently.
Finding: Fragmented systems and dedicated subprocess tools have improved process efficiency through automation, strategic partnerships and fraud-detection analytics. However, this increases the system complexity and dilutes the E2E management view, hindering the full optimization of the claims process.
4.1.2 Workflow-centric process optimization.
Continuous efforts to optimize cost structures have driven process efficiency measures across individual claims-processing subprocesses. Specific initiatives include applying lean process management to identify and eliminate waste, RPA to automate steps within workflows and a Business Automation Workflow (BAW) framework to automate all subprocesses. These methodologies represent a progression toward streamlined operations.
To reduce waste, claim handling agents are instructed to collect only the essential data required for processing claims, omitting approximately 50% of claims location data and 30% of loss-cause data from the OPAS system. This approach minimizes call durations but compromises data completeness. Similarly, efficiency gains are achieved in external partnerships, such as towing services, which use compound billing (e.g. consolidating 50 tow cases into a single invoice) instead of individual billing.
RPA facilitates data transfer between claims handling systems, avoiding the need for time-intensive system integration by automating the steps in the user interface. However, this approach perpetuates the system silos. BAW further streamlines processes by automating all subprocesses, such as standard claims handling, which reduces manual intervention but is limited to less complex cases.
Claims handling remains inherently complex, owing to the varied terms and conditions across multiple insurance covers. Although automation improves efficiency, it is constrained by system fragmentation. Purpose-built systems largely determine workflows and efforts to optimize the focus on subprocesses and an E2E perspective is lacking. Recent steps, such as appointing E2E process ownership, have aimed to address this limitation.
Finding: Efforts to optimize workflows have reduced waste and improved localized efficiency but have exacerbated system and process fragmentation. The lack of an integrated E2E perspective hinders holistic process optimization, signaling the need for better alignment between subprocesses and a unified management approach.
4.1.3 Rule-based decision-making.
Automated decision-making has emerged as a key driver of productivity gains in claims processing, supported by the bespoke ICH system. ICH integrates a rule engine and workflow automation to manage decision points, such as directing standard claims for automated handling, routing potentially fraudulent claims for investigation and forwarding nonstandard claims for manual processing.
The rule engine focuses primarily on identifying standard claims with low complexity and minimal fraud risk. Claims were scored based on predefined rules and those below the stipulated threshold were processed automatically. For example, claims involving personal injury, exceeding a monetary value limit or requiring adherence to specific insurance terms are flagged for manual handling. Fraudulent claims, which account for approximately 15% of insurance claims in Europe, are identified through rules that assess factors such as the claimant’s history with the company, policy tenure and anomalies in claim amounts.
Although rule-based decisions improve efficiency and cost savings, the challenges include the rapid complexity of rule management. Maintaining and updating the rules is resource-intensive and risks diminishing the system’s agility over time.
Finding: Automated decision-making delivers significant productivity and cost-efficiency gains but is limited by the complexity of rule management. A more scalable approach may mitigate these limitations and enhance decision-making accuracy.
4.2 Data governance
Data governance has been overlooked in the past in favor of prioritizing process management and efficiency gains. Traditionally, data has been regarded as a process resource rather than a strategic asset. This approach resulted in limited governance practices, primarily confined to master data management (MDM) and product data management (PDM). While Party MDM, driven by general data protection regulation (GDPR) requirements, is centrally managed by the “Data Governance Tribe,” PDM remains decentralized within individual business units.
Recently, management has recognized the need to strengthen data governance. A notable effort is the Data Governance 2.0 (DG2.0) project, which aims to define new governance practices and appoint business data owners for central data assets. These steps indicate a shift toward treating data as a strategic asset to support business objectives and compliance.
Finding: The traditional lack of focus on data governance hinders the strategic utilization of data. However, recent initiatives represent a necessary evolution in aligning data governance with modern business demands and regulatory requirements. However, consistent and centralized data governance efforts are essential for realizing the full potential of data as strategic assets.
4.2.1 Data as a separate entity from the process.
Previously, data were treated as resources to support processes rather than as strategic assets. Business managers often view operational reports, analytics and dashboards as “the data,” failing to recognize the role of processes in creating data. This perspective has resulted in limited attention being paid to data governance and quality metrics, particularly in the data-creation stage, data warehouses and data pipelines. Table 3 illustrates the data quality metrics measured and followed at different stages from data creation to data utilization.
Data management metrics
| Data quality dimension | Metric | Data creation | Data storage | Data distribution | Data utilization |
|---|---|---|---|---|---|
| Intrinsic data quality dimensions | Completeness | (X) | X | X | |
| Correctness | (X) | X | X | X | |
| Timeliness | X | ||||
| Consistency | (X) | X | |||
| Use-case-specific data quality dimensions | Discoverability | NA | (X) | (X) | |
| Accessibility | NA | X | X | ||
| Availability | NA | X | (X) | X | |
| Other | NA |
Note(s): X = fully realized, (X) = partially realized; NA = not applicable
Source(s): Authors’ own work
In the claims processing process, only correctness is measured during data creation, which is limited to core customer information validated against the Party MDM. Completeness and consistency are partially addressed during data storage but remain insufficient for downstream needs. In some cases, missing data are looked up from internal MDM systems or external databases. Data completeness further downstream will not improve unless missing data are manually collected from customers or cooperation partners. Productized data assets (e.g. customers, claims and policy data) benefit from enhanced consistency and discoverability as they are harmonized and cataloged. However, nonproductized data assets such as product profitability remain inconsistent and challenging to compare across product groups.
The data distribution stage ensures accessibility and addresses leaky pipeline issues but does not enhance the data quality dimensions. In the utilization stage, data quality is measured for analytics and reporting reliability; however, this cannot be improved. Intrinsic data quality issues from earlier stages persist, particularly completeness and consistency, affecting downstream processes such as risk management and profitability evaluation.
Finding: Data governance remains siloed, with intrinsic data quality dimensions such as completeness and consistency inadequately addressed during data creation and storage. The lack of unified governance across the data lifecycle stages limits the reliability of the downstream processes. To maximize value, intrinsic data quality should be the focal point from the creation stage onward, with a strategic emphasis on harmonizing data assets across the organization.
4.2.2 Inconsistent data structures.
The evolution of the claim processing process has resulted in subprocesses using distinct operational and supporting systems, leading to inconsistent data structures across the E2E process. These inconsistencies include variations in the attribute definitions, formats and logical relationships of data objects. Although minor differences can be remedied with reasonable effort, fundamental structural disparities represent a significant challenge in creating a unified data set for E2E process management.
For example, in the claims registration subprocess, data from the phone channel are gathered in the OPAS system, whereas data from the online channel are collected in a separate backend system. The online channel captures more detailed information owing to its comprehensive claims registration form, whereas the OPAS system prioritizes shorter call durations, leading to limited data coverage. Consequently, analytics from these two channels are not directly comparable.
Initially, mapping tables were used to bridge the differences between the systems. However, as the system landscape became more complex, the maintenance of these tables became impractical. A data warehouse was introduced to consolidate the data set for the claims process. Although effective for claim processing, the warehouse does not address the data needs of downstream processes, further highlighting the limitations of inconsistent data structures.
Finding: A consistent data set with unified attribute definitions, formats and structures is essential for effective E2E process management. The lack of data consistency, particularly across systems, remains the most significant barrier in creating a unified data set, inhibiting seamless management and optimization of the E2E process.
4.2.3 Lack of visibility on downstream data requirements.
Claims processing and sales are the primary upstream processes that create data, while downstream processes, including risk management, rely heavily on these data for core insurance business functions such as pricing and coverage evaluation. Risk management requires detailed claims histories to derive accurate risk evaluation data. Internal sources, such as claims data, are more critical than external ones for this purpose. From the actuary team’s perspective, having access to more claims data would be highly beneficial. Pricing is determined based on the actuary team’s evaluation of risks, including their outstanding risk position, likelihood and impact of claims, re-insurance costs, risk distribution and other relevant factors.
However, prioritizing process efficiency during claims registration, particularly in the phone channel, has led to the omission of essential data fields, such as loss location and loss, to reduce call durations. Customer service agents collecting claims data are unaware of the actuary team’s requirements, leading to incomplete data sets. Although mandatory field definitions have been introduced to meet downstream data needs, they have been overridden by operational management’s focus on efficiency.
This issue is exacerbated by a lack of direct communication between the operational management of claims processing and the actuary team, further deepening the disconnect between upstream and downstream data needs. Additionally, newly appointed business data owners focus primarily on regulatory compliance, leaving the upstream/downstream imbalance unaddressed.
Finding: The lack of visibility over downstream data requirements significantly undermines the utility of the data created during claim registration. This gap between upstream and downstream processes, driven by efficiency-focused operational decisions, results in incomplete data sets that fail to meet the critical needs of downstream stakeholders such as the actuary team.
Analysis key insights: The analysis revealed that the current process and data management approach is insufficient for achieving optimal E2E process automation and efficiency. While workflow optimization has delivered efficiency gains at the subprocess level, it has also resulted in poor data quality at the data creation stage. The most affected dimension is data completeness, as incomplete data impacts decision points beyond the creation stage. This issue is particularly evident in claims registration, in which missing data hinders effective decision-making in claims handling. Consequently, poor data quality adversely affects all the downstream processes linked to claim processing.
4.3 From workflow to data-centric process management
The insurance company’s processes are primarily service-oriented, but claim processing is driven by decision points optimized for efficiency. Automated decision-making, particularly through a rule engine in the BAW, plays a key role in improving process efficiency by categorizing claims as standard or suspicious for appropriate handling.
The company followed a workflow-centric approach (Figure 6), with efforts focused on optimizing the workflow and reducing cycle times. For example, the emphasis in claim registration has been on minimizing the call duration rather than gathering all the necessary data for handling claims. This resource efficiency approach maximizes the value per resource allocated but neglects the comprehensive data needed for downstream processes.
In contrast, a data-centric approach (Figure 7) shifts the focus to data quality, specifically the value-added from the completeness and consistency of data objects. The value comes from enriching these data objects, making them the foundation for decision-making and ensuring that business logic is reflected through data streams, not just workflows.
Finding: Transitioning from a workflow-centric to a data-centric process management model that prioritizes the completeness and consistency of data is essential for optimizing E2E process efficiency and decision-making in claims processing.
4.3.1 Impact of process metrics.
Process metrics are commonly used in companies to align strategic and management decisions with operational execution. Claims processing was managed through a set of metrics, as listed in Table 4. A key focus has been on reducing call duration, which leads claim agents at call centers to aim for shorter calls and the closure of as many tickets as possible. Although these process efficiency metrics are designed to optimize resource use, they do not initially seem to affect the process effectiveness or quality. However, this is primarily owing to the lack of visibility in downstream processes and data quality metrics, which limits a comprehensive view of the overall process performance.
Process management metrics at the operational level
| Category | Metric |
|---|---|
| Effectiveness | # Claim settlement decisions made |
| Efficiency (resources) | # Tickets closed per agent |
| Efficiency (time) | # Tickets closed per hour |
| Call duration | |
| Claim processing time | |
| Quality | Customer satisfaction: NPS (net promoter score) |
| Waiting time in the phone queue |
Source(s): Authors’ own work
Finding: Overemphasis on efficiency metrics, such as reducing call duration, can inadvertently neglect the importance of data quality and downstream process performance, potentially impacting the overall effectiveness and quality of claims processing.
4.4 Data quality and its downstream impact
Data quality plays a crucial role in shaping the effectiveness of the downstream processes. In the analyzed company, data are primarily created during the claim registration process, and these data are then used downstream in the claims handling process. The same claims data later form part of the claim history, which is used by the actuary team for risk evaluation and pricing and is later used for financial reporting. These data are critical for assessing the balance of the insurance product portfolio and calculating the product and customer profitability.
The downstream impact of the data quality is illustrated in Figure 8. To understand this impact, it is important to consider the entire data lifecycle, which includes data creation, storage, distribution and utilization. Some processes primarily create data, whereas others use it. Processes that use data also generate new data, which are then fed back to the lifecycle.
The two main data quality categories, intrinsic data quality and use-case-specific data quality were emphasized at different stages of the data lifecycle. Intrinsic data quality, such as correctness and completeness, should be ensured during the data creation stage because correcting these dimensions later is costly or impossible. Consistency can be improved during the data storage stage, although this requires considerable effort. Completeness can be addressed after creation through lookups from internal or external data sources, which are typically limited to MDM objects.
Ensuring use case-specific data quality is important during the data storage stage. Productized data assets that undergo version and release management tend to have better data quality. However, the later stages of the data lifecycle, such as data distribution, have little influence on improving data quality, with the “leaky pipeline syndrome” causing some data quality degradation. For example, in the analyzed company, claims data created during the phone channel’s claims registration process lacks completeness, as claims agents omit the required data fields. By contrast, the data collected through the online channel were more complete. The collected data suffer from consistency issues owing to the differing data structures between phone and online channels. This inconsistency leads to poorer quality data available for decision-making in the subsequent stages of claims handling.
As an example of the significance of data quality, statistical information revealed that with workflow-centric process management, the data coverage for accident location information was below 50%, with over 50% of accident locations left blank. Measures to improve data coverage, along with the transition toward data-centric process management, led to an improvement, increasing data coverage to approximately 70%. While accident location is not necessary for making a claim payment, it is important for the actuary team to build a risk profile. In addition, customers may expect location information, which can impact customer service. Missing data may also affect the reporting and other downstream processes. Other examples include missing details such as the time of the accident or categorizing a vehicle as an “other vehicle” rather than specifying whether it was a train, truck, or electric scooter. While missing data may not impact claims registration, handling or payments, it can significantly affect downstream processes and their overall effectiveness.
Finding: Data quality significantly impacts downstream processes, with poor data quality at the data creation stage affecting the entire data lifecycle and subsequent processes. Inconsistent data from different channels (e.g. phone and online) exacerbate this issue. This underscores the benefits of adopting a data-centric process management approach over a workflow-centric one, as the former places greater emphasis on the accuracy and consistency of data across the entire process.
Key insight: Data quality, even in areas where immediate operations are not directly affected, plays a crucial role in the governance, managerial and operational levels of downstream processes. Transitioning to a data-centric approach can lead to significant improvements in data quality across these levels, resulting in enhanced overall process effectiveness.
4.4.1 Data quality among process metrics.
Data quality is a central issue in claims processing and its role in process metrics has become increasingly emphasized. However, in a workflow-centric approach, the focus on cost efficiency often leads to trade-offs, with processes executed at the lowest acceptable quality to achieve maximum efficiency. This creates conflicts when attempting to prioritize data quality. By contrast, a data-centric approach views data as a primary asset for value-creation, where low data quality becomes a barrier to downstream value. Therefore, data quality must be considered not only in the data creation phase but also throughout the entire process lifecycle, including downstream stages, because it is an integral part of the value-add (process effectiveness). This indicates that data quality should be integrated into the process effectiveness metrics (Figure 9).
Finding: Integrating data quality metrics into process effectiveness metrics is essential for E2E process management using a data-centric approach. Low data quality limits downstream value-creation, highlighting the importance of considering the data quality across the entire process.
4.5 Synthesis: workflow-centric vs dataflow-centric process management
Effectively managing complex E2E processes may require a paradigm shift, as a workflow-centric approach often falls short of processes that primarily consist of data-driven decision points or decision trees. In such cases, a dataflow-centric process management approach may be better suited to enhance E2E effectiveness and efficiency. The key differences between the workflow-centric and dataflow-centric process management approaches are summarized in Table 5.
Workflow-centric vs dataflow-centric process management
| Aspect | Workflow-centric | Dataflow-centric |
|---|---|---|
| Process management | Focus on optimizing the workflow | Workflow follows the dataflow |
| Process flow | Sequential/parallel actions | A series of decision points |
| Role of data | Data is a resource | Data is an asset |
| Process effectiveness | Considered mainly within the process silo | Considered E2E |
| Process efficiency | Focused on resource efficiency | A balanced view of upstream/downstream related metrics |
| Process quality | Measured solely based on process output | Includes intrinsic data quality (completeness, correctness, timeliness and consistency) |
| Process maturity | Dependent on managing individual actions and their sequence in the process | Strongly dependent on data maturity |
Source(s): Authors’ own work
The benefits of dataflow-centric process management include improved data quality at decision points, which supports more effective decision automation. In workflow-centric processes, intrinsic data quality may suffer owing to the prioritization of resource efficiency, which limits automation capabilities to basic rule-based methods. In contrast, a dataflow-centric approach enables the use of clustering or AI-based methodologies, yielding fewer false negatives and positives, reducing overall process costs and enhancing decision accuracy.
Key insight: Transitioning from workflow-centric to dataflow-centric process management can enhance E2E effectiveness and efficiency by prioritizing data as an asset, improving decision quality, enabling advanced automation methods, ultimately reducing process costs and enhancing outcomes.
5. Discussion
This study highlights the need for a dataflow-centric process management approach to optimize E2E process efficiency and automation. Findings indicate that traditional workflow-centric approaches, while effective for task-level efficiency, fail to address the complexities of modern data-driven processes. These processes, such as the handling of insurance claims, involve multiple decision points that rely on high-quality data. Transitioning to a data-centric model enables better data utilization, improves decision-making and enhances the effectiveness of the E2E process.
Prior research has emphasized the importance of process optimization through methodologies such as Lean Six Sigma (Lameijer et al., 2024) and BPM (Reijers, 2021). However, these approaches predominantly focus on workflow-centric improvements, which have been shown to be insufficient for handling large-scale data complexity (Pech and Vrchota, 2022). Our findings support recent studies suggesting that dataflow-centric models are better suited for managing data-intensive decision-making processes (Ng et al., 2021; Zhou et al., 2023). The results extend previous work by demonstrating how data quality influences automation potential and decision accuracy.
A critical insight is that the downstream impact of poor data quality, particularly inconsistent and incomplete data from different channels of claims registration, such as missing accident location information, can hinder decision-making in downstream processes, such as risk evaluation and pricing. Although workflow-centric approaches have optimized operational efficiency (Telukdarie, 2019), they often overlook the importance of comprehensive data. A shift to a dataflow-centric model that prioritizes data quality (Liu et al., 2017; López Martínez et al., 2021) and integrates it throughout the process lifecycle can improve decision accuracy and reduce errors. Earlier studies have focused on the role of data governance in maintaining process integrity (Mahanti, 2019; Stahl et al., 2023), but our findings suggest that data completeness and consistency must be incorporated into process effectiveness metrics to optimize business outcomes.
5.1 Theoretical implications
This study contributes to the theoretical advancements in process management by proposing a shift from workflow-centric to dataflow-centric models. While prior research (Bhattacharya et al., 2009; Reijers et al., 2017) has explored data-centric frameworks, our findings uniquely highlight how this shift influences decision automation, data governance and efficiency trade-offs.
The theoretical contribution of this study lies in the proposal of a data-centric model that bridges the gap between traditional workflow-centric process management and increasing data-intensive demands of modern service delivery processes. These findings offer new insights into integrating data governance using data-driven metrics and applying automation methodologies to enhance E2E process efficiency and effectiveness. Moreover, this study discusses the barriers to adopting such a model and suggests practical solutions for overcoming these challenges.
Concerning the shift from workflow-centric to dataflow-centric process management, our findings introduce novel perspectives relevant to modern service processes that require data-intensive decision-making and automation. The move toward dataflow-centric management aligns with the growing need to leverage data for more accurate and efficient decision-making in complex, interconnected environments (Ng et al., 2021). This shift is aligned with the increasing complexity of modern processes (Abeysekara et al., 2023), in which managing data across organizational boundaries is central to improving outcomes. Our work challenges traditional process management approaches such as Lean Six Sigma (Lameijer et al., 2024) and BPM (Bazan and Estevez, 2022), which primarily focus on optimizing the flow of tasks rather than managing the flow of data (Pech and Vrchota, 2022).
Regarding the topic of data quality in E2E process management, our findings integrate data quality into process effectiveness metrics, representing an advancement over previous studies. Treating data quality as a core component of the process lifecycle emphasizes the need for evaluating and optimizing data quality at all levels – governance, managerial and operational – reflecting the increasing role of data in modern business processes (Hvam et al., 2019; Zhou et al., 2023). Previous studies, such as those by Haddar et al. (2016) and Liu et al. (2017), have highlighted the importance of data governance and quality but typically treat data quality separately from process management. Our theoretical contribution recontextualizes data quality within broader effectiveness metrics rather than just operational efficiency or task optimization (Di Ciccio et al., 2015).
The proposed data-centric governance model presents new insights into the role of centralized data governance, positioning data ownership as a strategic corporate-level responsibility. This approach aligns with the emerging theoretical understanding of data as a strategic asset (Mahanti, 2019) and underscores the need for a holistic governance strategy that spans multiple management levels to maximize data utility throughout its lifecycle (Andrews et al., 2021). While prior research (Stahl et al., 2023; Pech and Vrchota, 2022) has discussed the siloed nature of data governance, our study makes a contribution by proposing integration into the E2E process framework.
Regarding automation and decision-making, this study proposes that automated decision-making is a key driver of productivity and cost efficiency but remains constrained by rule management complexity and data quality issues. The findings suggest that AI and machine learning can optimize decision-making in data-rich environments, contributing new insights into scalable decision-making. Although previous studies (Ng et al., 2021; Willcocks et al., 2023) have explored automation, most have focused on rule-based systems and workflow automation (Bolcer and Taylor, 1998), often neglecting the limitations of such systems in handling large and complex data sets.
Finally, our study reinforces earlier findings by emphasizing how system fragmentation limits E2E optimization. It extends the discussion by proposing that the dataflow-centric approach can better align subprocesses and improve cross-functional coordination. The concept of using dataflow orchestration to address fragmented systems represents a fresh direction in solving E2E inefficiency, a challenge previously acknowledged but not fully explored in terms of holistic process optimization (Zhang and Perry, 2014; Kougka et al., 2018).
5.2 Managerial implications
Managers can leverage the transition to dataflow-centric process management to drive E2E process effectiveness and efficiency, particularly when increasing volumes of data are used in service and service delivery processes. A key understanding for managers is that complex processes span all management layers, governance, management and operations, each with unique challenges and solutions.
5.2.1 Lifecycle value of data across management layers.
Managers must recognize the value created through the lifecycle of data and how they are distributed across the operational, managerial and governance levels. For example:
Operational layer: Introduce guidelines to collect additional data attributes and improve data coverage.
Managerial layer: Incorporate data quality metrics into process effectiveness metrics to ensure alignment with process goals.
Governance layer: Transition data ownership from individual departments to a corporate level for better cross-departmental integration and regulatory compliance.
5.2.2 Data ownership models.
The ownership of data should align with its value-creation potential and regulatory requirements rather than being tied to the department where the data are created. For example:
Claims data: While generated by the Claims Handling Department, claims data form the foundation for risk evaluation and pricing by the actuary team. Ownership may need to shift to the corporate level for holistic value management.
Billing data: Although created by the Accounts Receivable Department, billing data are part of broader financial reporting and may necessitate ownership under business control to ensure consistency and compliance.
5.2.3 Implications for data domain structure.
A shift to dataflow-centric management may require restructuring of the data domain model. Traditionally, data domains might reflect organizational or departmental structures that roughly align with the workflow. Under a dataflow-centric approach, domains may need to mirror E2E data flows to support cross-departmental collaboration and better data utilization.
5.2.4 Process metrics and data quality.
To support dataflow-centric processes, data quality metrics may need to be integrated into the process effectiveness metrics. These metrics must be collaboratively determined by the department managing the process and downstream stakeholders, reliant on the data. Productizing key data assets as “data products” can further enhance both data and process maturity, addressing one of the major inhibitors of implementing data-centric management.
5.2.5 Practical steps for transition.
Practical steps for transitioning to dataflow-centric process management could include:
creating visibility for downstream data usage;
defining the downstream data requirements;
including data quality metrics in the process effectiveness metrics;
mapping the dataflow within the process; and
redesigning the process workflow to follow the dataflow to minimize data loss.
Adapting the process workflow to the dataflow is necessary to account for the data being the primary asset. These identified steps have implications at the architectural level, affecting business and information architecture.
5.2.6 Business benefits.
Managers must consider how a data-centric approach enhances capabilities such as customer lifecycle value analysis, proactive risk management and improved customer relationships. By shifting the focus to E2E data flows, managers can anticipate outcomes, enable smarter decision-making and unlock new value-creation opportunities.
Key managerial insight: Transitioning to dataflow-centric process management enables managers to redefine data ownership, restructure data domains and integrate data quality into process effectiveness metrics. These changes drive improved E2E process efficiency and support higher-value outcomes such as enhanced customer relationships and regulatory compliance.
6. Conclusions
This study highlights the need for a paradigm shift in managing complex E2E processes from a workflow-centric approach to a dataflow-centric model. This shift is vital to achieving both E2E effectiveness and efficiency. By prioritizing data as a strategic asset and incorporating data quality into process effectiveness metrics, organizations can potentially enhance process efficiency, decision accuracy and overall performance. This transition holds significant potential for optimizing E2E processes and automation, helping businesses to better meet evolving customer expectations while driving value-creation throughout the process lifecycle.
However, this study has some limitations. It focuses on a specific context and environment and does not assess which particular domains are most urgent to address when transitioning to dataflow-centric process management. Both processes and data are closely interconnected with various domains such as organization, cost accounting and enterprise architecture. Future research should explore these domains to better understand their roles and urgency during the transition.
Additionally, future studies could use statistical analysis to compare dataflow-centric and workflow-centric process-management approaches. Investigating how architectural elements (business, information, processes and system architecture) are affected by this shift would be valuable. Specifically, it would be worthwhile to examine how architecture changes when business processes are realized through data.
