Keywords

1 Introduction

An increasingly competitive environment and differentiated customer demands have shifted the focus from mass production to individualization of products and services over the last decades [57]. However, individualization entails increased complexity costs, as it requires different business process variants to be designed, implemented, managed and maintained. The more diverse the customer needs a company wants to address, the more process variants are required [51], leading to higher costs-to-serve. This creates a dilemma between revenue-sensitive process individualization and cost-sensitive process standardization, which can be observed in a number of industries (e.g. see detailed elaboration for the car manufacturing industry in [25]).

Established Business Process Management (BPM) methods and process-oriented improvement programs such as Six Sigma, Total Quality Management, Lean Management, and popular process management life-cycle models (e.g. [16]), tend to focus on process standardization over individualization. These methods are driven by measures such as processing time, cost-per-outcome unit or minimal variation (Six Sigma), but rarely give guidance in terms of how to individualize processes leading to higher variety of services and products [17, 41]. Ignoring individualization will no longer be viable for organizations [57], creating an increased demand for insights on how to approach it.

The rapid developments in the digital age, especially in the form of advanced manufacturing, robotic workflow management and data analytics brings tremendous potential to make process individualization cost-effective. These new affordances materialize in a higher variety of process outcomes and a reduced time to individualize. Thus, we argue that digital technologies have broadened the design space of business processes, providing the means for firms to individualize their products and services. However, the existing BPM method set does not provide sufficient support for capitalizing on this emerging affordance. We contribute towards addressing this gap by developing a defined set of design patterns for process individualization. In this way, we address the following research question: What are design patterns for process individualization?

This paper presents a conceptual model that explains how business processes can be individualized in the digital age. In particular, we derive four different design patterns for business process individualization. These patterns cover the essential constructs making up a business process. Contemporary case examples are provided in support of each design pattern as a means of illustrative evidence. The resulting framework contributes towards theory building by presenting a forming typology of process individualization with actionable design options. Thereby, we contribute towards forming theory of analyzing (Type I) and theory of design and action (Type V), as described in [21].

This paper is structured as follows. First, we outline the research background. Second, we introduce the design method that explains how we derived our theoretical model of process individualization patterns. Third, we present the conceptual model and explain the different design patterns for process individualization. We conclude the paper by summarizing its key contribution, and pointing to limitations and future work.

2 Research Background

Creating diverse process variants in an efficient way is a major barrier to the individualization of products and services. The developments in the digital age, especially robotic workflow management, sophisticated data analytics and advanced manufacturing (e.g. 3D printing) provide an entire new level of cost effective capabilities and make previously impossible forms of scalable process individualization accessible. In this section we summarize the contextual background to the key concepts underlying this study.

2.1 Stages of Individualization

Individualization refers to the degree to which products, services, and processes are configured to meet explicit as well as latent customer needs [57]. We distinguish between three stages of individualization (see Fig. 1).

Fig. 1.
figure 1

Stages of individualization

Full size image

The first stage consists of only standard products with no or very little individualization. Products and services that fall into this category are off-the-shelf products that follow a “one size fits all” approach. These products and services are uniform, as variation is either technically or economically not feasible, or there is no market for these offerings. On this stage of product and service individualization the customer does not play a role other than deciding and buying one of the products available in the market [31]. An example is Ford’s “any color as long as it’s black” approach in the early 20th century. By eliminating variation in the product, Ford was able to scale production in a way that was unprecedented. Nowadays, this level of individualization is predominant for daily products, where the costs of individualization outweigh the potential mark-up in price.

The second stage encompasses mass customized products or services that exhibit a certain level of individualization by offering different variants of the same product or service. As outlined by [19], there are different strategies that can be employed. Mass customization heavily relies on component-based manufacturing that enables companies to offer limited variation while still realizing returns of scale. In mass customization, the role of the customer is to choose from a variety of different modules that are produced and resembled by the manufacturer on large scale [31]. For example, Dell and many other computer manufactures allow customers to configure their device based on a pre-selection of components. In this stage, individualization is characterized by the plethora of options that are provided to the customer and the sophistication of product configuration and pricing engines that guide the user through the process.

The third and last stage reflects the highest level of individualization. Here, products and services are bespoke to an individual customer or purpose. Mass personalization combines the following four key properties as according to [57]. (1) As these products and services are tailored to specific customer needs, they are one of a kind. This is also referred to as the market-of-one. (2) Mass personalization needs to be paired with mass efficiency in order to be economically viable. (3) Companies need to employ customer co-creation to integrate the customer in all phases of the product life-cycle. (4) With an increasing level of product and service individualization, it becomes more important to detect and understand explicit as well as latent customer needs [34]. Modern technology enables this level of individualization. For example, hospitals can produce body parts by use of additive manufacturing (colloquially know as 3D printing), considering a patient’s unique physical characteristics [33].

2.2 Business Process Variant Management

Process variants are created to configure processes to diverse contexts due to varying environmental and market conditions [10]. To respond to customer needs in different markets, products and services are adapted. In turn, underlying processes often need to be altered in order to reflect these changes [51].

To allow for the generation and management of different process variants, research has investigated configurable process models [20], software product lines [39], and assembly system design [26], to name only but a few. These approaches have in common that they exclusively focus on the sequence of activities and their causal relationship to create variation, but do not consider other components of business processes. Thus, the more customized products and services a firm wants to offer, the more diverse its business processes need to be leading to inefficiency and increased complexity in the management of the business process portfolio [51]. This is, why many companies refrain from competing via business process individualization, but aim at standardizing their business processes instead [53].

Furthermore, process variants operate on the level of mass customization, i.e. they result in a set of product options the customer can choose from. However, process variants are limited in that they cannot (and do not aim to) provide a unique customer experience and tailoring of the process.

2.3 Business Process Improvement and Redesign Patterns

Prior research provides a rich set of process improvement and process re-design methods and patterns. These patterns “target the resolution or mitigation of problems” [18, p. 8] to increase efficiency and improve other metrics of the devil’s quadrangle [18] by addressing the “mechanics of the process” [42, p. 283]. Yet, how to provide companies with a set of design guidelines to differentiate their processes and customer touch points from competitors has so far been neglected.

We address these limitations by developing design patterns for process individualization. The patterns contribute to the body of knowledge in BPM by explaining and demonstrating how different components of business processes can be manipulated to individualize business processes and broaden their design space.

3 Design Method

In this section we explain how we derived design patterns for process individualization. We first introduce the key elements forming a business process, as all of these can potentially be manipulated to increase process variation. Second, we introduce the notion of design patterns [4] and describe how we utilized this concept for our research.

3.1 Conceptual Framework

Business processes are a sequence of activities leading to an output that generates value to an internal or external customer of an organization [16] by transforming inputs to outputs [32]. Figure 2 visualizes the different components of business processes and their interplay. A business process is composed of (1) process activities and buffers together with their respective sequence. Activities and steps of the process are carried out by (2) resources that are either capital assets or labor. Further, (3) the information and data associated with the process help to make process decisions and trigger process activities or sub-processes [32]. (4) The flow unit is a transient entity “that proceeds through various activities and finally exits the process as finished output” [32, p. 19].

Fig. 2.
figure 2

(adapted from [32])

Business process meta-model

Full size image

Traditionally, companies focused on creating changes to the output of the process by using different inputs or adapting process activities and the sequence of the process. However, this increases complexity and costs. How business processes can be individualized by changes to other components such as resources, data, and the flow unit of the business process has received limited attention.

In the remainder of this paper, we address business process individualization and how digital technologies can contribute to resolving the dilemma of increased complexity costs versus satisfying external demands for individualized services. First, variation to a process cannot only be achieved by creating diverse process sequences and activities, but also by manipulation of resources, data, and the flow unit of the business process. Second, technological developments enable lower costs of providing process variation. Especially, ready access to and sophisticated analyses of vast amounts of data can increase the cost effective feasibility of process variations and change the organizing logic [54] of process design.

3.2 Synthesis of Design Patterns

The approach we employ in this study is based on the notion of design patterns introduced by Alexander [3]. With the term pattern, Alexander describes the description of an artefact endowed with a guideline that can be used for creating the artefact [4]. Further, “the term pattern appeals to the replicated similarity in a design, and in particular to similarity that makes room for variability and customization in each of the elements” [15, p. 1]. Thus, design patterns serve as general solutions to reoccurring problems, while leaving room for creative freedom [3]. In BPM, patterns have been discussed, amongst others, in the context of process models [49], control flow [2], and data flow [45].

To derive the design patterns we employed heuristic theorizing as outlined by Gregory and Muntermann [22]. This approach is suitable, since design patterns are a form of heuristics that help to reduce the search for a satisfactory solution [22, p. 642]. First, we entered the heuristic search process and defined the problem at hand, i.e. the generation of design patterns for business process individualization. Since we soon realized that process individualization can be approached from different angles, we decomposed the problem into simpler problems that could be approached individually. That is, we reformulated the problem to derive patterns for individualization for each of a business process’ components, as defined above [32]. Next, we reviewed literature from real-life and published sources pertaining to case examples and the theoretical analysis of individualization. This extraction of prescriptive design knowledge from existing artifacts is also referred to as design archaeology [12]. Based on this information, we started generating the design patterns, followed by multiple rounds of heuristic synthesis and what Sein and associates call ‘reflection and learning’ [48]. When our patterns became stable, i.e. new cases did not change the derived design components, we finalized the design patterns and exited the heuristic search process. This abductive line of reasoning is common in design synthesis [30] and similar to the line of argument by Reijers et al. [42], who derive redesign patterns based on principles observed in practice and described in literature.

4 Design Patterns for Process Individualization

Based on the cases and literature analyzed, we derived four design patterns: (1) process sequence and activity individualization, (2) flow unit individualization, (3), resource individualization, and (4) data individualization. These patterns are distinct as each of them addresses a different facet of a business process. Thus, each design pattern can be used individually as well as in combination with one or more of the other design patterns. We capture this distinctive, yet interacting behavior of the patterns in Fig. 3, which depicts the resulting conceptual model.

Fig. 3.
figure 3

Design patterns for business process individualization

Full size image

Process sequence and activity individualization captures how technological developments can be used to more efficiently and effectively adjust the orchestration of tasks/steps and their sequence to provide an individualized outcome. The other three design patterns (flow unit, resource and data individualization) are control flow-agnostic, i.e. they do not change the control flow. We define control flow-agnostic individualization as: Individualization that is characterized by the manipulation of data, resources, and/or the flow unit of a business process in order to tailor process outcomes to a specific customer or purpose, while at the same time, sequence and activities of the respective business process remain unaltered.

Each design pattern is clustered in opportunity-driven and demand-driven strategies of business process individualization. Demand-driven individualization addresses explicit as well as implicit customer needs. This allows organizations to cope with increasingly differentiated customer demands [57]. In contrast, opportunity-driven individualization covers new ways that companies can use to individualize their processes by capitalizing on the rich pool of their internal resources. By aligning the business process with its context [10], effectiveness and efficiency can be further increased. In the following sections, we explain each design pattern in detail and provide evidence from practice how they can be utilized.

4.1 Sequence and Activity Individualization

Description. Sequence and activity individualization refers to the adaptation of process sequence and/or activities without changing the actual product or service. Accordingly, process sequence and execution of activities are bespoke to a particular customer, situation or condition.

Demand-Driven Sequence and Activity Individualization. First, process sequence and activity individualization can serve as a means to respond to customer needs. For example, consider the hotel brand Ritz-Carlton. Even though every guest receives the same service, i.e. an overnight stay at a hotel, the way different activities are executed is very much tailored to each individual customer [24]. Wilder, Collier and Barnes [52] discuss in detail how frontline employees can be supported in providing an adaptive service experience to customers. Beyond the question of how activities are carried out, Artificial Intelligence can provide support in choosing the next most suitable activity, i.e. answering the question what activity should be carried out. For example, IBM Watson for oncology can recommend the best next treatment option for cancer patients based on patient information and historical treatment data [27].

Opportunity-Driven Sequence and Activity Individualization. Second, companies can use process-related characteristics and information, such as process logs, to increase early individualization of their business processes. Based on historic and run-time information about the business process, more (precise) business rules and conditions can be formulated to trigger alternative process sequences. Pattern recognition can be used to derive detailed business rules given prior process instances and help configuring more individualized, robotic workflows that would not have been possible with human resources. Using current information including information about the flow unit (e.g. customer history), run-time adjustments to the process can be made. For example, banks use a combination of various attributes to calculate the loan default risk employing neural networks. These attributes can be re-incorporated into the process. Depending on the characteristics of a loan applicant, the extent of the process varies. Thus, applicants with high default risk run through a comprehensive screening, while for wealthy individuals with negligible default risk this process is shortened.

Third, environmental events or conditions can affect the execution of business processes. For example, Rosemann et al. [44] describe the claims handling process at an Australian insurance provider, which is highly context-sensitive. As some of Australia’s regions are prone to storms in summer, there are considerable more claims requests in summer compared to winter. Due to the large variance in numbers of cases (more than double), the insurance provider operates two season-dependent variants of the claims handling process. While in winter there is a full registration of the claim, in summer (when there is storm season) the claims handling process is shortened to include a rapid lodgment of the claim. Additionally, staff from other departments is re-deployed and casual staff are hired. This way, the insurance provider can offer a similar processing time, irrespective of the season. For a more detailed description of this case see [44]. The availability of micro-data has made such process individualization even more specific. For example, context-specific data about the situation of a customer (e.g. location of a car) can be used to trigger specific processes (e.g. automated remote speed-control).

Technical Realization. This type of individualization makes use of available environmental, processual, and customer related information. As processes become increasingly branched and individualized early, Artificial Intelligence can help route the respective flow units through the process. AI-algorithms can determine patterns in the data and create new process variants as the process is being executed. New variants can be compared to the existing process ‘in the shadow’ without impacting customers or process workers [46]. This allows not only for the formulation of more decision points, but also more complex business logic. For example, multiple attributes can be combined in order to make a decision. We can imagine that this eventually causes a shift from ex-ante process design to ongoing and even run-time and predictive process design. At the same time, declarative process modelling [38] can serve as an important tool to specify constraints in which the process has to be executed.

4.2 Flow Unit Individualization

Description. Flow unit individualization refers to the manipulation of the flow unit of a business process to deliver a unique outcome. This is the first of the three control flow-agnostic forms of individualization we present here. The sequence and activities of the process remains the same, but the flow unit is manipulated and exits the process as final individualized output.

Demand-Driven Flow Unit Individualization. Unique demands of customers can be met using flow unit individualization. For example, in the medical industry, biofabrication enables to print cell fibers, bones and even organs [33]. In case of an emergency, doctors and paramedics might soon be able to collect the details related to the injury of the patient and print the required body part. Therefore, a fiber, bone, or any other body part can be produced based on patient (customer) needs. The asset sharing platform launched by Healx is another example of demand driven flow unit individualization. It uses a machine learning algorithm based on patient’s biological information to match drugs to disease symptoms and also reveal the level of effectiveness for that particular patient [29].

Koren et al. [31] discuss how the development of so-called open products allows for the re-configuration of products even after production. This product class is characterized by its customizability that allows adding physical components once the product has been purchased by the customer. Further, where the physical and the digital meet, products become re-programmable [55]. This allows producers as well as third party providers to constantly add new features to an existing physical device. Any smart device that allows for the installation of apps, such as smartphones, smartwatches, smartspeakers, etc., falls under this category. This is referred to as ‘late individualization’ as it only occurs when the customer is using the product over time, e.g. a standard smart phone at the point of purchase becomes a highly individualized product due to customers’ downloads and settings.

Opportunity-Driven Flow Unit Individualization. Opportunity-driven individualization is initiated by internal needs of the organization. Internal needs cover various factors such as past knowledge, dearth or scads of resources or immediate need of a resource. The aerospace industry is utilizing the potential of 3D printing, as it provides unparalleled freedom in component design and fabrication [28]. Rolls-Royce is leveraging the potential of 3D printers to design and manufacture power systems for aircraft. The technology provides various advantages including design flexibility, quick iteration, and part consolidation [8]. 3D printing can also be used to produce spare parts and to repair machinery. This is of particular advantage, when idle time of capital assets result in large costs, as it is the case for trains [23], aircraft [5], different types of water-craft [40] as well as machines that are critical for production.

Flow unit individualization provides various advantages to consumers as well as businesses. Being able to produce a product based on unique consumer needs, increases customer satisfaction and trust [37]. Whereas, using available technology and resources to satisfy the internal needs of the organizations, results in reduced cost and time for businesses.

Technical Realization. Using technologies such as additive manufacturing and making use of open products and re-programmability, flow unit individualization can assist in producing a variety of products and services catered to individual consumer needs in an affordable manner. Flow unit individualization makes use of sophisticated technology to produce products based on internal needs, or cater for specific customer needs.

4.3 Resource Individualization

Description. Resource individualization determines the most appropriate resource for the activities a flow unit runs through. Because activities and process sequence remain unaltered, this type of individualization is control flow-agnostic. While resource specialization has undoubtedly led to a high level of efficiency in regards to activity execution, we argue that resources can also be used as a means to broaden the design space of business processes. Whether it is selecting an Uber driver with unique language skills for your ride to the airport, finding the right nanny for your child, or getting advice from someone with the same medical issue as yourself on patientslikeme.com, this type of individualization matches customer specific characteristics and requirements with available resources, while leaving sequence and activities of the process unchanged.

Demand-Driven Resource Individualization. Demand-driven resource individualization allows companies to respond to characteristics and likings of the customer by selecting a resource that matches these specifications. Applied correctly, resource individualization helps to increase customer satisfaction by tailoring services to customer needs. For this type of individualization, resources’ personal traits and characteristics that match with customers’ desires are of particular relevance. While complete business models of many start-ups and apps rely on matching resources, there are only few com-panies that use this from of resource individualization in their regular end-to-end business processes.

Platforms like Patientslikeme [36], Tinder [50], and CareGuide [11] allow customers to find patients with similar medical symptoms, a potential partner of their liking, or a perfect nanny to take care of their child. These companies provide a platform to choose resources as per one’s own requirements, making the customers feel privileged and taken care of. By providing a platform, these applications essentially connect people with a predefined set of characteristics.

Opportunity-Driven Resource Individualization. Opportunity-driven resource individualization makes use of the diversity that the company internal resource-pool provides. Companies can incorporate screening processes into their regular business processes to check for and optimize current resources’ utilization. When new process instances arrive, they can be routed to the most appropriate resource based on a set of predefined constraints. For example, an airline may route customer inquiries depending on the language proficiency of the customer calling. I.e. a Mandarin speaking customer will be matched with an agent that is familiar with Mandarin.

Resource individualization can also be used to ease critical process steps. This is the case when particular process activities determine the process outcome to a large extent. Often these parts of the process are non-routine and knowledge-intensive. By using specialized resources, companies are able to in-crease customer satisfaction, increase profitability, and guarantee that safety critical, legal or health related process aspects are executed correctly. In comparison to customer-oriented resource individualization, resources employed in process critical steps need to be highly specialized and have expert knowledge.

Insurance companies, for example, use resource individualization as part of a two-step procedure for their claims handling process. Customers calling to report on a claim, first report on some general information to an artificial call agent. The artificial call agent screens the information for inconsistencies and conspicuous patterns. If the algorithm detects any anomalies, the customer will be routed to an experienced, human call center agent. In a second step, the customer will be asked to provide more detailed information on the claimed case. A very experienced call agent can work on hard cases where fraud seems likely, while a new employee can work on easy cases that demonstrated low fraud potential in the first step of the process. As this case shows, resource individualization is most powerful, if combined with the analysis of customer information collected prior to or during process execution.

Technical Realization. From a technological perspective, resource individualization is enabled by data analytics and the information availability on customer characteristics and requirements as well as resource specific characteristics and features. With this information, companies can use classification algorithms to detect similarities and matches between customers and resources.

For demand-driven resource individualization, employing or buying all this different types of resources is not an economically viable strategy as this will cause high expenditures and yield low resource utilization. For this reason, companies can make use of crowdsourcing [9] and source tasks to individuals. For example, companies can use 99 Designs to commission different types of designs [1]. In many cases the outsourcing of tasks and sub-processes will require companies to build a community that they can default and delegate to.

For process critical steps, resource individualization helps to match already available resources with tasks that require their level of expertise. As these tasks often mark critical points of the process, require a high level of skills, experience, and trust, they cannot simply be sourced out. Also, there are only small incentives for highly experienced knowledge workers to participate in crowdsourcing.

4.4 Data Individualization

Description. With big data at the forefront, data serves as a valuable resource to make decisions [43]. Firms have access to vast amounts of customer, processual, and environmental data that provides innovation opportunities [35]. Amongst others, data can be used in a more integrated manner, to enable more agile, more accurate decision-making, and modify digital products.

Demand-Driven Data Individualization. Data can be individualized to meet the unique demands of the customer. Products and services that do not have a physical representation (anymore) such as music and movie streaming, but also insurances can make use of data individualization. For example, YouI car insurance adapts insurance premia of policy takers depending on how they use their car [56].

With the use of smart contracts on blockchain [14] or any other decentralized ledger, contracts can be automatically enforced. Insurance provider AXA offers a flight insurance against delayed flights [7] based on customer related data saved on an Ethereum blockchain. As the underlying smart contract is connected to a global air traffic database, policy holders are automatically reimbursed, once their flight is delayed. This principle can be used to adjust insurance payments to biological information, as available from fitbits or any smartwatch. This enables the tailoring of health insurances to individual characteristics and behavior of the customer. From an economic perspective, this tailoring can contribute to resolving information asymmetry, adverse selection, and moral hazard leading to more fair insurance premia.

Opportunity-Driven Data Individualization. Data individualization can also be used to enhance the decision-making process, contributing to the effectiveness of process outcomes. E.g. the Australian Taxation Office uses information from a range of social media sources to ensure relevant information is provided. Third-party sources include government bodies, employers, online selling platforms, stock exchanges, amongst others [6]. Social media and other personal data may be reviewed further, to make the right decision, when reviewing tax applications. Therefore, individual data can be used to make enhanced decisions, enabling organizations to achieve their goals in an effective manner.

Technical Realization. Data individualization is enabled by capturing, storing, and analyzing large volumes of data. Data analytics in form of Machine Learning and Artificial Intelligence provide the means to do so. To reap the benefits associated with this pattern, a strong foundation in data management is obligatory. For instance, to transform to a data driven insurer, AXA required cooperation of the data analytics, data management office and data engineering units [47]. The 3Vs, respectively 4Vs feature of big data [13] summarize the key challenges. First, companies need to be able to integrate and process data of different types and from various sources. Second, the volume of data is constantly increasing. Third, new data is continuously created (e.g. by sensors and equipment) and streamed to assigned data bases.

5 Summary and Conclusion

This paper presented a conceptual framework of four distinct design patterns for process individualization, namely: (1) sequence and activity individualization, (2) flow unit individualization, (3) resource individualization, and (4) data individualization. Drawing upon design patterns theory [3, 4], the framework is built to extrapolate how different components of a process can be manipulated to derive individualization options that broaden the design space of business processes. The suggested design patterns also guide organizations on how to best apply digitalization to efficiently obtain individualized products and services at lower costs.

The resulting framework is the first to theorize about process individualization. First, by presenting a classification of process individualization options, we contribute towards theory of analyzing (Type 1), as explained by Gregor [21]. We identify and describe what different process individualization options are. We provide various examples that demonstrate how the patterns (can) materialize. Secondly, the actionable design options provide guidance on how to usefully deploy the design patterns. Presenting each design pattern in the form of sub-patterns, technical realizations and illustrative examples contributes towards theory of design and action (Type 5) [21] on process individualization.

The framework acts as a useful reference and guide for different stakeholders. For example, process owners/process change champions can use the framework to identify new opportunities to enhance their business processes with individualized products and services. Process architects can use the framework when (re-)designing a single process or a portfolio of processes; it can be applied to individualize existing processes or to design new individualized processes. It is also applicable for those engaged in product and service innovations, where a single or few of these patterns may be considered to create innovative customer experiences.

While there are many benefits, this work is still in its genesis and calls for further research. The empirical support for the design patterns were based on literature and (limited) case examples from practice. Detailed empirical validation, i.e. through in-depth case studies of existing individualization practices or Action Design research where these patterns are newly executed, is warranted to further validate the patterns. The design patterns also need further specification in order to enhance their utility; in terms of its contextual applicability (i.e. how the patterns may be differently applicable based on process, organization and external-environmental contexts), and having detailed and validated procedure guidelines on how to implement the design patterns. Also, given the reliance of data across all patterns, data management considerations need to be carefully thought through and managed in order to maintain customer trust.