Conference PaperPDF Available


Cyber-physical systems generate and collect huge amounts of usage data during operation. Analyzing these data may enable manufacturing companies to identify weaknesses and learn about the users of their products. Such insights are valuable in the early phases of product development like product planning, as they facilitate decision-making for product improvement. The analysis and exploitation of usage data in product planning, however, is a new task for manufacturing companies. To reduce mistakes and improve the results, companies should build upon a suitable reference process model. Unfortunately, established models for analyzing data cannot be easily applied for product planning. In this paper, we propose a reference process model for usage data-driven product planning. It builds on three well-established models for analyzing data and addresses the unique characteristics of usage data-driven product planning. Finally, we customize the model for a manufacturing company and demonstrate how it could be implemented in practice.
A Reference Process Model for Usage Data-Driven Product Planning
Maurice Meyer
Paderborn University
Ingrid Wiederkehr
Paderborn University
Melina Panzner
Paderborn University
Dr.-Ing. Christian Koldewey
Paderborn University
Prof. Dr.-Ing. Roman Dumitrescu
Paderborn University
Cyber-physical systems generate and collect huge
amounts of usage data during operation. Analyzing
these data may enable manufacturing companies to
identify weaknesses and learn about the users of their
products. Such insights are valuable in the early phases
of product development like product planning, as they
facilitate decision-making for product improvement.
The analysis and exploitation of usage data in product
planning, however, is a new task for manufacturing
companies. To reduce mistakes and improve the results,
companies should build upon a suitable reference
process model. Unfortunately, established models for
analyzing data cannot be easily applied for product
planning. In this paper, we propose a reference process
model for usage data-driven product planning. It builds
on three well-established models for analyzing data and
addresses the unique characteristics of usage data-
driven product planning. Finally, we customize the
model for a manufacturing company and demonstrate
how it could be implemented in practice.
1. Introduction
The further development of products is the main
development focus of engineers. In a survey with 247
engineers, ALBERS et al. found that only 7% of
developments are true new developments without carry
overs or adjustments of existing products. In contrast,
93% of developments represent further developments of
existing products [1]. To further develop a product,
engineers need to answer questions like: How does the
product perform in the field? What are its strengths and
weaknesses? How do users utilize the product? What are
new requirements for the product? Answering such
questions is not trivial, as companies often lack high
quality information about the product usage phase [2].
A new solution for this problem emerges from the
progressive digitalization of products. In recent years,
the digitalization has transformed mechatronic products
into so-called cyber-physical systems that integrate
hardware, sensors, data storage, microprocessors,
software, and connectivity. These systems can collect
data about themselves and their environment during
their utilization [3].
The analysis of such usage data promises to be
especially valuable during the early stages of product
development like product planning [4]. As the initial
phase or phase zero of product development, product
planning aims at finding success potentials of the future
[5] and promising products to be developed [6]. In this
regard, usage data analytics can help exploit insights
about predecessor products and their users that lead to
new success potentials and set the agenda for product
improvements. In conjunction, usage data analytics and
product planning span the new research area usage data-
driven product planning [7].
At present, the analysis of usage data in product
planning is not widely researched. For example,
BERTONI found that research on analyzing data to
identify customer needs mainly focuses on data from
social media and online reviews [8]. In addition,
approaches that address usage data often have a broader
perspective than product planning. For example,
WILBERG et al. present a stakeholder-oriented procedure
for the development of a use phase data strategy [10].
The approach addresses stakeholders in different
business functions, e.g., service or marketing. While
such approaches help to identify the potential value of
usage data for the different stakeholders of a company,
they do not help analyze usage data in product planning.
As there are no existing approaches, the creation of
a reference process model for usage data-driven product
planning is required. In this paper, the term reference
process model describes a generic process model that
formalizes recommended practices for a certain domain
[11]. Our research question is as follows: How does a
reference process model for usage data-driven product
planning have to be designed?
Proceedings of the 55th Hawaii International Conference on System Sciences | 2022
Page 6105
(CC BY-NC-ND 4.0)
Utilizing a design science approach, the result of
this paper is a reference process model for usage data-
driven product planning. It guides companies with a
stepwise procedure and helps structure projects in usage
data-driven product planning. It is especially useful for
manufacturing companies that want to start exploiting
usage data in product planning.
The paper is structured as follows: In section 2, we
describe our research design for the construction of the
reference process model. Section 3 presents the
reference process model for usage data-driven product
planning in detail. In section 4, we show the
customization of the model for a manufacturing
company. Section 5 concludes the paper.
2. Research design
For the development of the reference process model
for usage data-driven product planning, we followed the
design science research (DSR) guidelines. According to
HEVNER et al., DSR aims at developing viable artifacts
(e.g., models or methods) for relevant business
problems. These artifacts must be novel and provide a
clear contribution through solving so far unsolved
problems. They are created and evaluated by applying
rigorous methods and utilizing the knowledge base [12].
While design science research represents the
foundation of our research design, our concrete
procedure broadly derives from the Process Model for
an Empirically Grounded Reference Model
Construction by AHLEMANN and GASTL. This process
model emphasizes the importance of consulting domain
experts when constructing and validating reference
process models [13]. In addition to that, we analyzed
established models following DE LA HIDALGA et al. [14]
and FRANK et al. [15]. To carve out a suitable reference
process model from all the input data, we utilized VOM
BROCKE’s design principles [16]. Overall, our
procedure comprises four phases: Domain analysis,
Model construction, Empirical validation, and
Customization (see Figure 1). Subsequently, the phases
are described in detail.
Figure 1: Procedure model for the development of the
reference process model
2.1. Domain analysis
In the domain analysis phase, relevant knowledge
about the domain shall be captured and prepared for the
model construction [13].
For the considered domain, MEYER et al. provide a
recent comprehensive systematic literature review [7].
In their work, the authors derive the main concepts,
advantages, success factors and challenges of usage
data-driven product planning [7] und thus present a
substantial overview of the topic. Therefore, we used the
literature review as our knowledge base.
Next, we investigated our knowledge base in search
of necessary process steps for the reference process
model. AHLEMANN and GASTL propose this step as a
suitable way to structure and aggregate the knowledge
available [13]. We analyzed the knowledge base using
the question: Which process steps does the desired
reference process model need to include? Table 1 shows
the derived process steps for the reference process
model. They represent domain requirements for the
model construction and conclude our domain analysis.
Table 1: Necessary process steps for the reference process
model derived from the knowledge base
Process step
Analysis of the product and strategy [10, 17]
Analysis of the data analytics capabilities [10]
Definition of use cases [4, 17, 18]
Definition of data needs [3, 19]
Collection of usage data [1921]
Pre-processing of usage data [19, 2123]
Analysis of usage data, i.a. [19, 22, 24, 25]
Validity-check of data analysis results [19, 26]
Interpretation of the analysis results [19, 27]
Creation of new ideas [28]
Identification of requirements [4, 27, 29, 30]
Planning the improvement of existing and
future products, i.a. [17, 21, 25, 26, 28, 31, 32]
2.2. Model construction
The model construction phase aims at developing a
first version of the reference process model [13]. This
version is based on the necessary process steps from the
knowledge base and on existing process models.
As our desired reference process model shall
describe how to analyze usage data of existing products
to find improvement potentials, we focused our analysis
of existing process models on models for analyzing data.
For this, numerous process models exist. MARISCAL et
al. and PLOTNIKOVA et al. provide an overview about
existing process models and show that almost all
identified approaches are based on two original models:
Page 6106
the Cross Industry Standard Process for Data Mining
(CRISP-DM) and the Knowledge Discovery in
Databases model (KDD) [33, 34]. Therefore, these two
models are subsequently described in more detail.
CRISP-DM is an industry-, tool- and application-
neutral process model for data mining. It was created by
the CRISP-DM CONSORTIUM and is based on the
experience of numerous data mining practitioners [35,
36]. For several years, it has been considered the
de facto standard for data mining projects.
The process model consists of six iterative phases:
(1) Business understanding aims at understanding the
business objectives and requirements and converting
them into a concrete data mining problem definition.
(2) In data understanding, the data are collected and
investigated to spot, for example, data quality problems.
(3) The data preparation includes tasks like the
transformation and cleaning of the data to construct the
final dataset. (4) During modeling, models are built and
refined with the data. (5) In the evaluation, the model
must be evaluated regarding the business objectives.
(6) Finally, in deployment, the model is put into
operation and a report is generated [35].
KDD refers to the overall process of discovering
useful knowledge from data [37]. FAYYAD et al. define
the KDD process as “the nontrivial process of
identifying valid, novel, potentially useful, and
ultimately understandable patterns in data” [38].
The process consists of nine iterative steps: (1) In
learning the application domain, relevant domain
knowledge and the desired goals are captured.
(2) Creating a target dataset aims at selecting the
dataset to be analyzed. (3) In data cleaning and pre-
processing, basic operations like noise reduction and
outlier removal take place. (4) Data reduction and
projection focuses on tasks like feature engineering and
dimensionality reduction. (5) In choosing the function of
data mining, the purpose of the desired model is decided
(e.g., classification, regression). (6) Choosing the data
mining algorithm(s) addresses the selection of suitable
method(s) for searching patterns in the data, e.g., by
comparing different models and parameters. (7) Data
mining describes the search for patterns in the data. (8)
In interpretation of the results, the patterns discovered
are interpreted and translated into the domain language.
(9) The final step usage of the discovered knowledge
covers documenting the new knowledge, reporting it
and taking action [38].
In addition to these industry-neutral models like
CRISP-DM, KDD, and their derivatives, we decided to
also analyze one manufacturing-focused approach to
account for any industry-specific aspects. The
VDI/VDE 3714 guideline presents a standard for the
implementation and operation of big data applications in
the manufacturing industry. It aims at aggregating the
numerous contributions towards big data analytics in the
manufacturing industry and at unifying them into one
model [39].
The guideline describes seven iterative phases:
(1) In the definition phase, the questions and objectives
to be answered or achieved need to be specified.
(2) Next, exploring the data situation aims at describing
and structuring the available data and defining
additionally required data. (3) In data management, data
from different sources are merged. (4) Modeling deals
with creating an evaluable model from the data.
(5) Subsequently, an initial evaluation of the data
analysis results with respect to the project goals is
necessary. (6) Implementation and rollout aims at
transferring the big data application into continuous
operation. (7) The final phase sustainability addresses
the project documentation as well as an assessment of
economic, technical and social aspects to ensure a
sustainable impact of the big data project [39].
To construct our first version of the reference
process model, first, we analyzed the three existing
process models CRISP-DM, KDD, and the VDI/VDE
3714 guideline. We followed DE LA HIDALGA et al. and
FRANK et al. and created a detailed process model for
each, including all tasks and steps mentioned in their
descriptions [14, 15]. We also synchronized their steps,
thereby highlighting gaps and similarities in the models.
Second, we aggregated the established reference
models to build one exhaustive model from the three
original models [16]. We found that we could sort all
process steps into four main processes: (1) Planning of
the data analysis, (2) Analytics and data preparation,
(3) Analytics workflow design and modeling and
(4) Exploitation of the data analysis results. In detail:
(1) Planning of the data analysis contains process
steps related to the early business perspective on the
project: Parts of CRISP-DM’s business understanding
(e.g., the determination of business objectives and
requirements); KDD’s learning the application domain;
VDI/VDE 3714 guideline’s definition.
(2) Analytics and data preparation includes
process steps concerning the clarification of the
analytics task as well as the selection and collection of
the data: Parts of CRISP-DM’s business understanding
(e.g., the determination of data mining goals), CRISP-
DM’s data understanding; KDD’s creating a target
dataset; VDI/VDE 3714 guideline’s exploring the data
situation and data management.
(3) Analytics workflow design and modeling
summarizes the process steps addressing data pre-
processing and analysis: CRISP-DM’s data preparation
and modelling; KDD’s data cleaning and pre-
processing, data reduction and projection, choosing the
function of data mining, choosing the data mining
algorithm(s), and data mining; VDI/VDE 3714
Page 6107
guideline’s modeling and parts of the evaluation of the
data analysis results.
(4) Exploitation of the data analysis results
consists of process steps for the interpretation and
utilization of the data analysis results: CRISP-DM’s
evaluation; KDD’s interpretation of the results and
usage of the discovered knowledge; Parts of VDI/VDE
3714 guideline’s evaluation of the data analysis results.
We omitted the following process steps, as they aim
at the continuous operation of the built models and
therefore do not contribute to the purpose of our
reference process model: CRISP-DM’s deployment;
VDI/VDE 3714 guideline’s implementation and rollout
and sustainability.
Third, again following VOM BROCKE’s design
principles, we specialized the aggregated model for our
considered domain [16]. For each necessary process
step from Table 1, we analyzed if it was already
sufficiently addressed in the model. If not, we added a
new process step (e.g., derivation of use cases)
following DE LA HIDALGA et al. and FRANK et al. [14,
15]. As a result, we obtained a first version of the
reference process model with four main and 16 sub
processes. Figure 2 shows the schematic procedure.
Figure 2: Schematic procedure for the model construction
2.3. Empirical validation
For the validation of the reference process model,
AHLEMANN and GASTL recommend conducting
interviews with topic experts [13]. We performed three
semi-structured interviews of 60-90 mins duration. Our
interview partners are characterized in Table 2.
Table 2: Overview about interview partners
Innovation and process consultant
6 years
Digitalization consultant and
15 years
Head of data science department
11 years
The interviews were aimed at answering the following
questions: Is the reference process model complete or is
something missing? Are the sub processes structured in
a logical sequence? Are the sub processes clearly
separated from each other?
We discussed each main and each sub-process with
all interviewees and protocolled all remarks and
questions raised in the interviews.
After each interview, we refined the reference
process model according to the remarks of the experts.
For example, we added the sub-process Update product
strategy in the Exploitation main process. After all
interviews were conducted, we showed the updated
reference process model to all experts again. We asked
for further remarks, but all experts were satisfied and
confident that it fulfills its requirements. Section 3
presents the resulting model.
2.4. Customization
After the validation, AHLEMANN and GASTL
suggest to test the reference process model [13]. We
performed this practical test with a machinery and plant
engineering company, which produces machine tools
and production equipment in the field of forming
In the test, we customized the reference process
model for the company. We used the RACI method,
which is also known as RAM (Responsibility
Assignment Matrix) [40]. The method’s goal is to assign
tasks and responsibilities. For that, it uses a matrix with
processes in the rows and roles in the columns [40].
For the assignment of tasks and responsibilities,
four options are possible. They are derived from the
name of the method: R, A, C, and I. The letter R stands
for responsible and describes which person is
responsible for a task. The letter A stands for
accountable: It shows who decides whether a task has
been performed correctly. For example, this could be a
supervisor. The letter C describes that a person should
be consulted when the task is performed. Lastly, the
letter I names all people who need to be informed about
the events and the results for that specific task [41].
While working with the matrix, several assignment rules
Page 6108
must be respected, e.g., only a single person can be
responsible for a given task [41].
At the start of the workshop, we collected the
company’s relevant roles for usage data-driven product
planning. Then, we successively assigned the roles to
each sub-process of the reference process model. The
results are shown in section 4.
3. Reference process model for usage data-
driven product planning
The reference process model for usage data-driven
product planning consists of four main processes:
(1) Planning of the usage data analysis, (2) Analytics
and data preparation, (3) Analytics workflow design
and modeling and (4) Exploitation of the data analysis
results. Each main process comprises four sub-
processes. In the following, all sub-processes are
described. The sources are given with the corresponding
IDs, e.g., [I-1] and [P-1] (see Table 1 and Table 2), and
abbreviations, e.g., [CRISP-DM]. Figure 3 shows the
reference process model. Please note that the arrows
represent sequence flows and not information flows.
(1) Planning of the usage data analysis
(1.1) Identification of investigation needs:
Input: Product strategy
Procedure: This sub-process aims at finding critical
needs for investigation concerning the product or its
users. For this, the business objectives specified in the
product strategy are analyzed [CRISP-DM, I-1, P-1].
This strategy contains details about the product program
design, e.g., about the product’s strategic focus and its
planned evolution [42, 43]. In addition to the strategy,
the status quo of the product under investigation is
examined [KDD, CRISP-DM, P-1]. From these
analyses, critical investigation needs are derived.
Output: Investigation needs
(1.2) Analysis of data analytics capabilities:
Input: -
Procedure: In this sub-process, the capabilities of the
organization and the product concerning usage data-
driven product planning are assessed [P-2]. For
example, the available resources (e.g., data mining
experts, usage data) need to be analyzed [CRISP-DM].
Output: Data analytics capabilities
(1.3) Definition of boundary conditions and goals:
Input: Investigation needs (1.1); Data analytics
capabilities (1.2)
Procedure: Next, boundary conditions and goals need
to be defined [CRISP-DM, KDD, VDI/VDE]. Boundary
conditions include requirements, prerequisites,
assumptions, and constraints of the project. Just as the
goals, they can refer to the organization or the product
context. Goals help achieve the business objectives and
guide the analytics activities. The boundary conditions
and goals set the direction for the subsequent steps.
Output: Boundary conditions and goals
(1.4) Derivation of use cases:
Input: Boundary conditions and goals (1.3)
Procedure: Use cases address the previously defined
content goals. They link possible influential variables to
a goal or ask questions about the relationships of goals
and influential variables [VDI/VDE, I-2, P-3]. For each
use case, costs and benefits must be estimated
Output: Use cases
(2) Analytics and data preparation
(2.1) Specification of analytics objectives:
Input: Data analytics capabilities (1.2); Use case (1.4)
Procedure: This sub-process transforms a use case into
concrete data analytics objectives [CRISP-DM]. If
domain knowledge is missing, it needs to be captured to
derive relevant variables and adequate data analytics
approaches (e.g., clustering or association rule mining)
[KDD]. For this, the data scientists stay in close contact
with the domain experts [I-3].
Output: Data analytics objectives
(2.2) Definition of data needs:
Input: Data analytics objectives (2.1)
Procedure: This sub-process specifies the data and their
sources [P-4]. Also, it defines specific measurements or
data objects [VDI/VDE]. This includes the
transformation of the defined physical variables into
concrete data attributes in the target dataset. For
example, from the domain knowledge and demand (e.g.,
noise behavior), raw data and data sources (e.g., sound
vibrations via a sound level meter) need to be derived.
Output: Data needs
(2.3) Collection of data:
Input: Data analytics capabilities (1.2); Data needs
Procedure: In this sub-process, the data needs are
compared with the actual existing data in the company.
If the required data are not available, an acquisition
concept must be developed and implemented [P-5].
Output: Raw data set
(2.4) Description of data:
Input: Raw data set (2.3)
Procedure: This sub-process is about understanding the
collected data and analyzing them regarding their
processing options [CRISP-DM]. Here, the goal is data
literacy. Steps are a first exploration, a holistic
Page 6109
description of the data with relevant meta data, and a
classification of the data [VDI/VDE]. If data quality is
not sufficient, it is necessary to go back to the previous
step and acquire the data with the required quality [I-3].
Output: Data characteristics
(3) Analytics workflow design and modeling
(3.1) Workflow design:
Input: Data analytics objectives (2.1); Data
characteristics (2.4)
Procedure: In this sub-process, pre-processing and
modeling techniques are selected and composed into
analytics workflows [I-3]. Typically, there are several
techniques for the same problem type. Some modeling
techniques need specific data formats or have certain
model assumptions. This requires a close link between
data preparation and modeling as well as an alignment
of these steps as a coherent workflow [I-3].
Output: Conceptual workflows
(3.2) Data pre-processing:
Input: Raw data set (2.3); Conceptual workflows (3.1)
Procedure: In this sub-process, analysts need to refine
the data to prepare it for modeling [CRISP-DM, KDD,
P-6]. Here, the pre-processing methods defined in the
workflows must be implemented (e.g., record and
attribute selection, integration of different data sets, data
cleaning, data transformation, and feature engineering).
Output: Pre-processed data
Figure 3: Reference process model for usage data-driven product planning
Page 6110
(3.3) Model building:
Input: Conceptual workflows (3.1); Pre-processed
data (3.2)
Procedure: This sub-process is about implementing the
algorithms of the workflows together with an adequate
test design [CRISP-DM, KDD, VDI/VDE, P-7]. The
latter examines and compares the workflows and their
parameter settings.
Output: Analysis results
(3.4) Model validation:
Input: Analysis results (3.3)
Procedure: The models and their results need to be
validated [CRISP-DM, VDI/VDE, P-8]. For this,
suitable technical metrics and evaluation criteria must
be selected. Model building and validation iterate until
results no longer improve and reach satisfaction.
Output: Model performance
(4) Exploitation of the data analysis results
(4.1) Results interpretation:
Input: Use case (1.4); Analysis results (3.3); Model
performance (3.4)
Procedure: The interpretation of the analysis results
seeks to reveal unknown, but valuable insights about the
product and its users [P-9]. It covers the evaluation of
the analysis results and their transfer into the product
context, e.g., the description, explanation, and
verification of the results by product experts [KDD,
VDI/VDE]. The insights form the starting point for the
identification of new success potentials for product
improvement [I-1].
Output: Success potentials for product improvement
(4.2) Idea generation:
Input: Investigation needs (1.1); Success potentials for
product improvement (4.1)
Procedure: In this step, promising ideas for product
improvement are generated [P-10]. They are aimed at
exploiting the identified success potentials. From all
ideas, the most promising ones are selected [I-1].
Output: Ideas for product improvement
(4.3) Requirements derivation:
Input: Success potentials for product improvement
(4.1); Ideas for product improvement (4.2)
Procedure: Building on the success potentials and ideas
for product improvement, this sub-process aims at
translating the new ideas into requirements for product
development [P-11].
Output: Requirements
(4.4) Product strategy update:
Input: Requirements (4.3)
Procedure: In the last sub-process, the product strategy
is updated based on the new requirements [I-1]. It
specifies the changes planned for new product
generations and existing products [P-12].
Output: Updated product strategy
4. Customized process model
To demonstrate how it could be implemented in
practice, we customized the reference process model
with a machinery and plant engineering company,
which produces machine tools and production
equipment in the field of forming technology (see
section 2.4). The company links nine roles to the
implementation of the process: Purchasing agent, sales
engineer, service technician, pre-developer, mechanical
developer, electrical developer, virtual developer, head
of development and CEO. In the following, tasks and
responsibilities of each role are described. The resulting
RACI-matrix is shown in Figure 4.
Of all roles, the purchasing agent is the least
involved. He only supports during the identification of
investigation needs by presenting current and future
purchasing challenges.
The activities of the sales engineer are limited to
the first main process. He is responsible for the
identification of investigation needs. After that, he is
consulted because of his close customer relationships.
Later, he receives information about the results
interpretation and the product strategy update.
The role of the service technician is especially
important in the first two main processes. As he is in
close contact with customers, he is consulted during the
planning and preparation main processes. Furthermore,
he supports in the results interpretation.
The pre-developer helps in the identification of
investigation needs. At the end, he is informed about the
product strategy update as it affects his future work.
Mechanical and electrical developers accompany
the process from the derivation of use cases to the
product strategy update. Early on, their expertise is
needed for the analytics and data preparation. While
they are only informed about the analytics workflow
design and modeling, they are again consulted for
results interpretation and idea generation.
The virtual developer is the most important role
for performing usage data-driven product planning in
the considered company. He is responsible for all sub-
processes of the first three main processes except the
identification of investigation needs (see sales
engineer). For the second and third main process, he also
oversees the results obtained. While he has a consulting
function for results interpretation, he is only informed
about the results of the sub-processes idea generation,
requirements derivation, and product strategy update.
Page 6111
The head of development has a consulting role
during the planning main process. While he is only
informed about the second and third main process, he is
responsible to exploit the identified insights and find
promising product improvements.
The last role is the CEO. He is informed about all
process activities. Furthermore, he oversees the
planning and exploitation main processes.
5. Conclusion
The result of this paper is a reference process model
for usage data-driven product planning. It consists of
four main processes: (1) Planning of the usage data
analysis, (2) Analytics and data preparation,
(3) Analytics workflow design and modeling and
(4) Exploitation of the data analysis results. Each main
process is divided into four sub-processes, which further
explain the tasks to be performed. The model’s
customization with a machinery and plant engineering
company shows how the reference process model could
be implemented in practice. In the following, our work’s
contributions to theory and practice are described.
Finally, its limitations as well as recommendations for
future research are presented.
5.1 Contributions to theory and practice
Usage data-driven product planning is a new and
promising research field. While there are numerous
established reference process models for performing
data mining or big data analyses, none of them can
easily be applied to product planning. The reference
process model presented in this paper addresses and fills
this gap. Thereby, the developed artifact represents a
valuable addition to the knowledge base as requested in
design science research [12]. The reference process
model contributes to the scientific discourse by
describing the utilization of data analytics approaches in
the early phase of innovation.
For the practical contribution, the model provides
managers an overview about how to perform usage data-
driven product planning. Managers can confidently
build on the model as it is based on three process models
that are widely used in practice. For the implementation,
our paper lays out how to customize the model by
assigning responsibilities to each sub-process.
5.2 Limitations and recommendations for
future research
The result of this work is subject to four main
limitations. (1) When analyzing existing reference
models, we focused on three models: CRISP-DM,
KDD, and VDI/VDE 3714. However, there are
numerous further reference models, many of which are
derivatives of the original CRISP-DM and KDD models
[34]. These models might have included further aspects
for our process. Yet, we are convinced that we captured
all critical aspects with our selection. (2) The validation
was performed with three interviews. A higher number
of interviews would probably have generated further
Figure 4: Results of the customization in a RACI-matrix
Page 6112
valuable suggestions. Yet, as our interviewees have
diverse experiences and perspectives, we are confident
that they pointed out the most important points. (3) The
reference process model conceptually describes the
required steps and tasks of usage data-driven product
planning. However, concrete methodical approaches are
not yet included in the reference process model.
(4) Until now, we have not yet fully applied the
reference process model in practice, but only parts of it.
Therefore, some real-world challenges could still be
Considering the limitations, future research should
focus on the following three recommendations:
(1) Methodical approaches need to be developed for
each sub-process. (2) The whole process should be
performed with multiple companies to find more
practical challenges. (3) The reference process model
should be regularly updated and improved, e.g., after its
first complete application in practice.
This work is funded by the German Federal Ministry of
Education and Research (BMBF). The authors are
responsible for the content of this publication.
[1] A. Albers, N. Bursac, J. Urbanec, R. Lüdcke, and G.
Rachenkova, “Knowledge Management in Product
Generation Development - an empirical study”, in Design
for X Proceedings of the 25th DfX-Symposium, TUHH
Universitätsbibliothek, 2014, pp. 1324.
[2] Q. Deng, S. Wellsandt, K. Hribernik, and K.-D. Thoben,
“Understanding Users and Products in Product
Development: The Application of Product Usage
Information and its Challenges, in Proceedings of the
International Conference on Engineering Design
(ICED21), Gothenburg, Sweden, 2021, pp. 32993308.
[3] M. E. Porter and J. E. Heppelmann, “How Smart,
Connected Products Are Transforming Competition,
Harvard Business Review, vol. 92, no. 11, pp. 6488,
[4] M. Holler, F. Uebernickel, and W. Brenner,
“Understanding the Business Value of Intelligent
Products for Product Development in Manufacturing
Industries, in Proceedings of the 8th International
Conference on Information Management and
Engineering, Istanbul, Turkey, 2016, pp. 1824.
[5] J. Gausemeier, R. Dumitrescu, S. Kahl, and D. Nordsiek,
“Integrative development of product and production
system for mechatronic products, Robotics and
Computer-Integrated Manufacturing, vol. 27, no. 4, pp.
772778, 2011.
[6] K. T. Ulrich and S. D. Eppinger, Product design and
development, 6th ed. New York, NY: McGraw-Hill,
[7] M. Meyer, I. Wiederkehr, C. Koldewey, and R.
Dumitrescu, “Understanding Data-Driven Product
Planning: A Systematic Literature Review, in
Proceedings of the International Conference on
Engineering Design (ICED21), Gothenburg, Sweden,
2021, pp. 32893298.
[8] A. Bertoni, “Data-Driven Design in Concept
Development: Systematic Review and Missed
Opportunities, Proceedings of the Design Society:
DESIGN Conference, vol. 1, pp. 101110, 2020, doi:
[9] J. Wilberg, T. Kalla, M. Fetscher, F. Rimbock, C.
Hollauer, and M. Omer, “Development of a Use Phase
Data Strategy for Connected Products: A Case Study in
Industry, in 2018 Portland International Conference on
Management of Engineering and Technology (PICMET),
Honolulu, HI, 2018, pp. 112.
[10] J. Wilberg, L. Fahrmeier, C. Hollauer, and M. Omer,
“Deriving a Use Phase Data Strategy for Connected
Products: A Process Model, in Proceedings of the
DESIGN 2018, 2018, pp. 14411452.
[11] J. Ling and L. Zhang, “Generating Hierarchical
Reference Process Model Using Fragments Clustering,
in Asia-Pacific Software Engineering Conference, 2015,
pp. 4047.
[12] A. Hevner, S. T. March, and J. Park, “Design Science in
Information Systems Research, MIS Quarterly, vol. 28,
no. 1, pp. 75105, 2004.
[13] F. Ahlemann and H. Gastl, “Process Model for an
Empirically Grounded Reference Model Construction,
in Reference modeling for business systems analysis, P.
Fettke and P. Loos, Eds., Hershey, PA., London,
Melbourne, Singapore: Idea Group Publishing, 2007, pp.
[14] A. N. de la Hidalga, A. Hardisty, P. Martin, B. Magagna,
and Z. Zhao, “The ENVRI Reference Model”, in Lecture
Notes in Computer Science, Towards Interoperable
Research Infrastructures for Environmental and Earth
Sciences, Z. Zhao and M. Hellström, Eds., Cham:
Springer International Publishing, 2020, pp. 6181.
[15] M. Frank, J. Gausemeier, N. Hennig-Cardinal von
Widdern, C. Koldewey, S. Menzefricke, and J. Reinhold,
“A reference process for the Smart Service business:
development and practical implications, in Proceedings
of the ISPIM connects, International Society for
Professional Innovation Management (ISPIM), 2020, pp.
[16] J. vom Brocke, “Design Principles for Reference
Modeling: Reusing Information Models by Means of
Aggregation, Specialisation, Instantiation, and Analogy,
in Reference Modeling for Business Systems Analysis:
IGI Global, 2007, pp. 4776.
[17] J. Wilberg, I. Triep, C. Hollauer, and M. Omer, “Big Data
in product development: Need for a data strategy, in
Proceedings of PICMET '17: Technology Management
for Interconnected World, Portland, USA, 2017, pp. 1
[18] Hofmann, P., Jöhnk, J., Protschky, D., & Urbach, N.,
“Developing purposeful AI use cases – a structured
method and its application in project management, in
Page 6113
15th International Conference on Wirtschaftsinformatik
(WI), Potsdam, Germany, 2020.
[19] L. Hou and R. J. Jiao, “Data-informed inverse design by
product usage information: a review, framework and
outlook, J Intell Manuf, vol. 31, no. 3, pp. 529552,
2020, doi: 10.1007/s10845-019-01463-2.
[20] Y. M. Goh and C. McMahon, “Improving reuse of in-
service information capture and feedback, Journal of
Manufacturing Technology Management 20(5), 2009.
[21] M. Abramovici, P. Gebus, J. C. Göbel, and P. Savarino,
“Utilizing Unstructured Feedback Data from MRO
Reports for the Continuous Improvement of Standard
Products, in Proceedings of the 21st International
Conference on Engineering Design (ICED17),
Vancouver, Canada, 2017.
[22] M. E. Porter and J. E. Heppelmann, “How Smart,
Connected Products Are Transforming Companies,
Harvard Business Review, vol. 93, no. 10, pp. 96114,
[23] M. Shahbaz, M. Srinivas, J. A. Harding, and M. Turner,
“Product Design and Manufacturing Process
Improvement Using Association Rules, Proceedings of
the Institution of Mechanical Engineers, Part B: Journal
of Engineering Manufacture, vol. 220, no. 2, pp. 243
254, 2006, doi: 10.1243/095440506X78183.
[24] M. Cantamessa, F. Montagna, S. Altavilla, and A.
Casagrande-Seretti, “Data-driven design: the new
challenges of digitalization on product design and
development, Design Science, vol. 6, 2020, doi:
[25] J. Igba, K. Alemzadeh, P. M. Gibbons, and K.
Henningsen, “A framework for optimising product
performance through feedback and reuse of in-service
experience, Robotics and Computer-Integrated
Manufacturing, vol. 36, pp. 212, 2015, doi:
[26] D. van Horn, A. Olewnik, and K. Lewis, “Design
Analytics: Capturing, Understanding, and Meeting
Customer Needs Using Big Data, in Proceedings of the
ASME 2012 International Design Engineering Technical
Conferences & Computers and Information in
Engineering Conference, Chicago, Illinois, USA, 2012,
pp. 863875.
[27] J. Wilberg, F. Schäfer, P. Kandlbinder, C. Hollauer, M.
Omer, and U. Lindemann, “Data Analytics in Product
Development: Implications from Expert Interviews, in
IEEE International Conference on Industrial Engineering
and Engineering Management (IEEM), Singapore, 2017,
pp. 818822.
[28] L. Wu, L. Hitt, and B. Lou, “Data Analytics, Innovation,
and Firm Productivity, Management Science, vol. 66,
no. 5, pp. 20172039, 2020, doi:
[29] T. Wuest, K. Hribernik, and K.-D. Thoben, “Capturing,
Managing and Sharing Product Information along the
Lifecycle for Design Improvement, in Proceedings of
the 10th International Workshop on Integrated Design
Engineering. IDE'14, pp. 107115.
[30] M. Holler, G. Neiditsch, F. Uebernickel, and W. Brenner,
“Digital Product Innovation in Manufacturing Industries
Towards a Taxonomy for Feedback-driven Product
Development Scenarios, in Proceedings of the 50th
Hawaii International Conference on System Sciences
(HICSS), Hawaii, USA, 2017, pp. 47264735.
[31] H. Holmström Olsson and J. Bosch, “Towards Data-
Driven Product Development: A Multiple Case Study on
Post-deployment Data Usage in Software-Intensive
Embedded Systems, in vol. 167, Lean Enterprise
Software and Systems: LESS 2013. Lecture Notes in
Business Information Processing, B. Fitzgerald, K.
Conboy, K. Power, R. Valerdi, L. Morgan, and K.-J. Stol,
Eds., Berlin, Heidelberg: Springer-Verlag, 2013, pp.
[32] S. A. Chowdhery, M. Bertoni, J. Wall, and T. Larsson,
“A data-driven design framework for early stage PSS
design exploration, Design Science, 2020.
[33] G. Mariscal, Ó. Marbán, and C. Fernández, “A survey of
data mining and knowledge discovery process models
and methodologies, The Knowledge Engineering
Review, vol. 25, no. 2, pp. 137166, 2010, doi:
[34] V. Plotnikova, M. Dumas, and F. Milani, “Adaptations of
data mining methodologies: a systematic literature
review, PeerJ Computer Science, vol. 6, e267, 2020, doi:
[35] P. Chapman et al., “CRISP-DM 1.0 Step-by-step data
mining guide, SPSS, vol. 9, 2000.
[36] C. Shearer, “The CRISP-DM Model: The New Blueprint
for Data Mining, Journal of Data Warehousing, vol. 5,
no. 4, pp. 1322, 2000.
[37] U. Fayyad, G. Piatetsky-Shapiro, and P. Smyth, “From
Data Mining to Knowledge Discovery in Databases,
AIMag, vol. 17, no. 3, p. 37, 1996, doi:
[38] U. Fayyad, G. Piatetsky-Shapiro, and P. Smyth, “The
KDD process for extracting useful knowledge from
volumes of data, Communications of the ACM, vol. 39,
no. 11, pp. 2734, 1996, doi: 10.1145/240455.240464.
[39] VDI/VDE 3714 Blatt 1, Implementation and operation of
big data applications in the manufacturing industry
Implementation of big data projects. Berlin: Beuth
Verlag, 2020.
[40] Project Management Institute, A Guide to the Project
Management Body of Knowledge (PMBOK® Guide),
6th ed. Newtown Square, PA: Project Management
Institute, 2017.
[41] T. Costello, “RACI—Getting Projects "Unstuck", IT
Professional, vol. 14, no. 2, 64-63, 2012, doi:
[42] J. Gausemeier, R. Dumitrescu, J. Echterfeld, T. Pfänder,
D. Steffen, and F. Thielemann, Innovationen für die
Märkte von morgen: Strategische Planung von
Produkten, Dienstleistungen und Geschäftsmodellen.
München: Hanser, 2019.
[43] R. G. Cooper, “The drivers of success in new-product
development, Industrial Marketing Management, vol.
76, pp. 3647, 2019, doi:
Page 6114
... For utilizing data analytics in the knowledge value stream, to which product planning belongs, Meyer et al. (2022b) propose a reference process model that includes four process steps for the exploitation of data analytics results, which the authors derived from the literature and expert interviews: Results interpretation, requirements derivation, idea generation, and product strategy update. As a concrete application of data analytics in the knowledge value stream, Riesener et al. (2021) show, for example, how they derive new requirements for a product based on undesired correlations uncovered in the analyzed data. ...
... The workshop was conducted in advance of the exploitation of data analytics results with the consortium and served to collect the expectations, experiences, and requirements of the participants. The workshop was based on the four process steps for the exploitation of data analytics results according to Meyer et al. (2022b). Using the "World Café" method, the participants worked on two questions for each process: "What are the requirements for this process?" ...
... For enabling the effective instantiation and application of the design principles, we organized them into a method. To support this, we used the necessary process steps for the exploitation of data analytics results proposed by Meyer et al. (2022b) as a framework: Results interpretation, requirements derivation, idea generation, and product strategy update. We matched these four phases and the 13 design principles to obtain the method. ...
Conference Paper
Full-text available
The megatrend digitalization turns mechatronic products into continuous collectors and generators of use phase data. By analyzing this data, manufacturers can uncover valuable insights about the products and the users. Especially in product planning, these insights could be used to plan promising future product generations. The systematic exploitation of data analytics results, however, represents a serious challenge, as research on the topic is still scarce. In this paper, we present 13 design principles for exploiting data analytics results in product planning. The results are based on a systematic literature review and a workshop with a research consortium. The evaluation of the design principles is demonstrated with a real case of a manufacturing company. The identified design principles represent a first contribution to a still scarcely explored research field.
... Figure 1 illustrates these concepts. Meyer et al. (2022) propose a reference process model for the analysis of use phase data in product planning. It consists of four main processes: (1) Planning of the use phase data analysis, (2) Analysis and data preparation, (3) Analytics workflow design and modeling, and (4) Exploitation of the data analysis results (Meyer et al., 2022). ...
... Meyer et al. (2022) propose a reference process model for the analysis of use phase data in product planning. It consists of four main processes: (1) Planning of the use phase data analysis, (2) Analysis and data preparation, (3) Analytics workflow design and modeling, and (4) Exploitation of the data analysis results (Meyer et al., 2022). ...
... Then, ideas for improving the product and related requirements are developed. Finally, the product strategy and other documents are updated based on the new insights, ideas, and requirements (Meyer et al., 2022). Figure 2 shows the reference process model. ...
Full-text available
The ongoing digitalization of products offers product managers new potentials to plan future product generations based on data from the use phase instead of assumptions. However, product managers often face difficulties in identifying promising opportunities for analyzing use phase data. In this paper, we propose a method for planning the analysis of use phase data in product planning. It leads product managers from the identification of promising investigation needs to the derivation of specific use cases. The application of the method is shown using the example of a manufacturing company.
... However, problems often arise at this intersection [5]. When using data analytics in product planning, the use case definition and derivation of data analytics tasks is also an important and challenging step [6]. While product managers often have a clear understanding of their strategic goals (e.g., identifying the weak points and causes to improve the product), they are not clear about how data-driven analytics methods can help achieve those goals and formulate appropriate use cases and questions or hypotheses. ...
Full-text available
Cyber-physical systems (CPS) generate huge amounts of data during the usage phase. By analysing these data, CPS providers can systematically uncover hidden product improvement potentials for future product generations. The successful implementation of such analytics use cases depends to a large extent on whether the stakeholders involved succeed in coordinating their goals and procedures. In particular, product managers and data scientists must come to a common understanding in the context of defining and concretizing the use cases. A common vocabulary is necessary so that the data scientists or those responsible for analysis can determine target-oriented, analysis-capable use cases with which the processing of the data can start quickly and successfully. The research question that arises at this point is: How can business goals or use cases be translated into realizable analytics use cases or tasks? In this paper we present the Busines-to-Analytics Canvas as a result of an action design research approach. It supports the translation of business use cases and goals into concrete data analytics tasks for product planning. By providing various information elements and guiding questions, the canvas helps data scientists translate the business goal into a data analytics approach, i.e., an algorithm class, and gather the necessary information to start processing data.
Full-text available
Digitalization and the momentous role being assumed by data are commonly viewed as pervasive phenomena whose impact is felt in all aspects of society and the economy. Design activity is by no means immune from this trend, and the relationship between digitalization and design is decades old. However, what is the current impact of this ‘data revolution’ on design? How will the design activity change? What are the resulting research questions of interest to academics? What are the main challenges for firms and for educational institutions having to cope with this change? The paper provides a comprehensive conceptual framework, based on recent literature and anecdotal evidence from the industry. It identifies three main streams: namely the consequences on designers, the consequences on design processes and the role of methods for data analytics. In turn, these three streams lead to implications at individual, organizational and managerial level, and several questions arise worthy of defining future research agendas. Moreover, the paper introduces relational diagrams depicting the interactions between the objects and the actors involved in the design process and suggests that what is occurring is by no means a simple evolution but a paradigmatic shift in the way artefacts are designed.
Full-text available
Ubiquitous and pervasive computing holds great potential in the domain of Product-Service Systems to introduce a model-driven paradigm for decision support. Data-driven design is often discussed as a critical enabler for developing simulation models that comprehensively explore the PSS design space for complex systems, linking of performances to customer and provider value. Emerging from the findings of two empirical studies conducted in collaboration with multinational manufacturing companies in the business-to-business market, this paper defines a data-driven framework to support engineering teams in exploring, early in the design process, the available design space for Product-Service Systems from a value perspective. Verification activities show that the framework and modeling approach is considered to fill a gap when it comes to stimulating value discussions across functions and organizational roles, as well as to grow a clearer picture of how different disciplines contribute to the creation of value for new solutions.
Full-text available
Advances in automation, communication, sensing and computation enable experimental scientific processes to generate data at increasingly great speeds and volumes. Research infrastructures are devised to take advantage of these data, providing advanced capabilities for acquisition, sharing, processing, and analysis; enabling advanced research and playing an ever-increasing role in the environmental and Earth science research domain. The ENVRI community identified several recurring requirements in the development of environmental research infrastructures such as i) duplication of efforts to solve similar problems; ii) lack of standards to harmonise and accelerate development, and bring about interoperability; iii) a large number of data models and data information systems within the domain, and iv) a steep learning curve for integration complex research infrastructure systems. To address these challenges, the ENVRI community has developed and refined the Environmental Research Infrastructures Reference Model (ENVRI Reference Model or ENVRI RM), a modelling framework encoding this knowledge. The proposed modelling framework encompasses a language and a notation to describe the research domain, its systems and the requirements and challenges faced when implementing those systems. By adopting ENVRI RM as an integrative approach, the environmental research community can secure interoperability between infrastructures, enable reuse, share resources, experiences and common language, reduce unnecessary duplication of effort, and speed up the understanding of research infrastructure systems. This chapter provides a short introduction to the ENVRI RM.
Full-text available
The paper presents a systematic literature review investigating definitions, uses, and application of data-driven design in the concept development process. The analysis shows a predominance of the use of text mining techniques on social media and online reviews to identify customers’ needs, not exploiting the opportunity granted by the increased accessibility of IoT in cyber-physical systems. The paper argues that such a gap limits the potential of capturing tacit customers’ needs and highlights the need to proactively plan and design for a transition toward data-driven design.
Full-text available
The use of end-to-end data mining methodologies such as CRISP-DM, KDD process, and SEMMA has grown substantially over the past decade. However, little is known as to how these methodologies are used in practice. In particular, the question of whether data mining methodologies are used ‘as-is’ or adapted for specific purposes, has not been thoroughly investigated. This article addresses this gap via a systematic literature review focused on the context in which data mining methodologies are used and the adaptations they undergo. The literature review covers 207 peer-reviewed and ‘grey’ publications. We find that data mining methodologies are primarily applied ‘as-is’. At the same time, we also identify various adaptations of data mining methodologies and we note that their number is growing rapidly. The dominant adaptations pattern is related to methodology adjustments at a granular level (modifications) followed by extensions of existing methodologies with additional elements. Further, we identify two recurrent purposes for adaptation: (1) adaptations to handle Big Data technologies, tools and environments (technological adaptations); and (2) adaptations for context-awareness and for integrating data mining solutions into business processes and IT systems (organizational adaptations). The study suggests that standard data mining methodologies do not pay sufficient attention to deployment issues, which play a prominent role when turning data mining models into software products that are integrated into the IT architectures and business processes of organizations. We conclude that refinements of existing methodologies aimed at combining data, technological, and organizational aspects, could help to mitigate these gaps.
Full-text available
An appropriate problem-solution-fit is essential to develop purposeful artificial intelligence (AI) applications. However, in domains with an unintuitive problem-solution-fit, such as project management (PM), organizations require methodological guidance. Hence, we propose a five-step method to develop organization-specific AI use cases: First, companies must consider the context factors technology, organization (in particular data and application domain), and environment. Second, companies must identify existing domain problems and AI solutions. Third, our method facilitates abstraction to understand the underlying nature of the identified problems and AI solutions. Fourth, our problem-solution-matrix assists companies to match AI functions with the domain context. Fifth, companies derive necessary implications for the subsequent use case implementation. To construct and evaluate our method, we followed the design science research paradigm complemented by situational method engineering and based on 14 interviews. Our method addresses a relevant practical problem and contributes to identifying purposeful AI use cases in unintuitive application domains.
Conference Paper
Full-text available
Due to the two mega trends Digitalization and Servitization, companies try to start a business with Smart Services. These are digital services that create value using data of a corresponding physical product, e.g. predictive maintenance. The business processes of traditional manufacturing companies have been optimized for doing business with physical products. In many aspects, designing and operating a Smart Service differs. While designing them, certain data security measures have to be included, for example. We propose a reference process based on established processes in literature that we specifically advanced for Smart Services. It focuses four main processes: planning, developing, performing, and billing. Using OMEGA as a modelling language, we included information about activities, resources (e.g. IT systems), required skills, potentially used methods, and detailed descriptions of the 126 steps. This information and two developed workshop tools facilitate the specification. The reference process has been validated with a German plant manufacturer.
Full-text available
A significant body of knowledge exists on inverse problems and extensive research has been conducted on data-driven design in the past decade. This paper provides a comprehensive review of the state-of-the-art methods and practice reported in the literature dealing with many different aspects of data-informed inverse design. By reviewing the origins and common practice of inverse problems in engineering design, the paper presents a closed-loop decision framework of product usage data-informed inverse design. Specifically reviewed areas of focus include data-informed inverse requirement analysis by user generated content, data-informed inverse conceptual design for product innovation, data-informed inverse embodiment design for product families and product platforming, data-informed inverse analysis and optimization in detailed design, along with prevailing techniques for product usage data collection and analytics. The paper also discusses the challenges of data-informed inverse design and the prospects for future research.
We examine the relationship between data analytics capabilities and innovation using detailed firm-level data. To measure innovation, we first utilize a survey to capture two types of firm practices, process improvement and new technology development for 331 firms. We then use patent data to further analyze new technology development for a broader sample of more than 2,000 publicly traded firms. We find that data analytics capabilities are more likely to be present and are more valuable in firms that are oriented around process improvement and that create new technologies by combining a diverse set of existing technologies than they are in firms that are focused on generating entirely new technologies. These results are consistent with the theory that data analytics are complementary to certain types of innovation because they enable firms to expand the search space of existing knowledge to combine into new technologies, as well as the theoretical arguments that data analytics support incremental process improvements. Data analytics appears less effective for developing entirely new technologies or creating combinations involving a few areas of knowledge, innovative approaches where there is either limited data or limited value in integrating diverse knowledge. Overall, our results suggest that firms that have historically focused on specific types of innovation—process innovation and innovation by diverse recombination—may receive the most benefits from using data analytics. This paper was accepted by Chris Forman, information systems.