See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/385737747

D3 framework: An evidence-based data-driven design framework for new

product service development

Article in Computers in Industry · November 2024

DOI: 10.1016/j.compind.2024.104206

CITATIONS READS

0 32

2 authors, including:

Boyeun Lee
University of Exeter Business School
18 PUBLICATIONS 95 CITATIONS

SEE PROFILE

All content following this page was uploaded by Boyeun Lee on 12 November 2024.

The user has requested enhancement of the downloaded file.

 Computers in Industry 164 (2025) 104206

Contents lists available at ScienceDirect

Computers in Industry
journal homepage: www.sciencedirect.com/journal/computers-in-industry

D3 framework: An evidence-based data-driven design framework for new

product service development
Boyeun Lee *, Saeema Ahmed-Kristensen
INDEX, Business School, Faculty of Environment, Science, Economics, University of Exeter,Tintagel House, 92 Albert Embankment, London SE1 7TY, UK

A R T I C L E I N F O A B S T R A C T

Keywords: Despite growing interest in the use of data for product and service development, a comprehensive understanding
Data-driven design of how data is employed in the context of new product, service and product–service system development is
Data-driven innovation lacking. With the aim of deepening understanding of data as a critical resource for generating value through new
New product development
products and services, we conducted a systematic literature review, conceptualised through a framework and
New service development
Systematic literature review
evaluated with a questionnaire survey. This study (1) identifies the relationships between methodologies and
various data-x design concepts, together with their contributions; (2) investigates the types of data captured and
utilised across the product/service development process; (3) identifies data-driven design (DDD) activities and
the types of data for each activity and (4) develops and validates an evidence-based framework of DDD for new
product/service development processes. This study is distinct from previous work as our theoretical foundation
identifies seven DDD activities alongside the types of data captured and utilised throughout the new product,
service or product–service system development. The key findings highlight the relationship between commonly
used concepts for using data in product/service development (i.e., data-driven, -enabled, -centric, -aware,
-informed, and design analytics) and their methodological differences. The findings show that whereas data is
currently captured predominantly from the in-use phase of a product/service, it is mainly used to support
concept development. This paper contributes by developing a DDD framework, which helps practitioners un­
derstand how data and machine learning approaches can be used for product/service development. The
evidence-based framework also contributes to the body of knowledge on data-x design and the understanding of
the role of data in product/service development.

1. Introduction data-driven design in the development of new products and services has
become an increasingly significant research area (Zhang et al., 2017).
Industry 4.0 promises a shift towards more connected products and The term ‘data-driven design’ (hereinafter, DDD) encompasses
services, which will involve transitioning into systems of systems, methods, approaches and processes using data to assist the development
requiring new business processes and presenting new opportunities for of products and services (Lee and Ahmed-Kristensen, 2023). Prior to
value creation from data (Gann et al., 2014; Papalambros, 2015; Porter Domazet et al.’s (1995) introduction of the term, although studies uti­
and Heppelmann, 2014). In this context, data has been generated lised big data in the design process, they lacked clear terminology to
throughout the product’s life cycle (Tao et al., 2018; Lee, 2022), and describe this practice (Briard et al., 2021; Zheng et al., 2020). Analysis of
efficient value extraction from data and the effective management of big data in product development processes can lead to new service
data and design knowledge are considered technological enablers for concept ideas (Kim et al., 2016) or improved design efficiency whilst
companies to stay competitive and reduce the time to market connecting users, products, services and production processes (Bogers
(Chattopadhyay et al., 2017; Lee et al., 2018; Speed et al., 2019). Studies et al., 2018; Lee et al., 2019a), and result in product innovation (Kusiak,
have described new product development as a primary driver for 2006; Lee et al., 2019b), and value creation (Lee et al., 2022a; Lee,
competitive advantage for many corporations (Alam, 2006; Kleinsch­ 2022).
midt and Cooper, 1991); thus, understanding how to effectively utilise Despite the increased innovation opportunities due to the greater
big data and establish appropriate methodologies and processes for availability of data, only a few papers conceptualise an integrative and

* Corresponding author.
E-mail address: [email protected] (B. Lee).

https://doi.org/10.1016/j.compind.2024.104206
Received 7 September 2023; Received in revised form 17 September 2024; Accepted 1 November 2024
0166-3615/© 2024 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
B. Lee and S. Ahmed-Kristensen Computers in Industry 164 (2025) 104206

holistic approach to understanding how data or analytics are used for 2.2. Research Methods
new product/service development (Bogers et al., 2016). Most systematic
literature review (SLR) papers focus on a specific phase (Bertoni, 2020; An SLR is a critical research methodology that enables the synthesis
Feng et al., 2020; Machchhar et al., 2022; Ortega et al., 2022), a of an existing body of knowledge to generate new knowledge on a
particular discipline (e.g. human–computer interaction (HCI) or engi­ broader scale than is possible with empirical studies and identify new
neering (Hou and Jiao, 2020)) or developing novel design activities future research agendas (Denyer and Tranfield, 2009; Taroun, 2014).
using data in a particular case (e.g., customising based on anthropo­ We conducted an SLR of the research articles about DDD published from
metric data; Ellena et al., 2018). Although valuable, these studies do not 2012 to 2022. The research approach included four stages, as outlined in
fully encompass all the DDD activities within the product/service Fig. 1: planning the review, selecting and reviewing the papers, syn­
development context, which makes it difficult to understand how data thesising the results and evaluating the findings. We modified the
can contribute to new product/service development. Recognition of the methodology adopted for the first three stages of the SLR from several
significance of DDD activities has increased across various fields, as academic sources (Denyer and Tranfield, 2009; Mostafa et al., 2016;
evidenced by the emergence of special interest groups (e.g., Tranfield et al., 2003) to accomplish the research aim and objectives.
data-informed design in Design Society, data as human-centred design in The author team conducted the entire process of selection and exami­
the CHI community). Scholars (Bertoni, 2020; Lin et al., 2016; Zheng nation of the papers and conceptualised the framework. The final stage
et al., 2019) have also stressed the need for the systematic development was to evaluate the conceptualisation of the SLR – that is, the DDD
of DDD models. Thus, it is timely to draw together these learnings to framework. Each of these stages are detailed in the next sections.
identify key DDD activities for researchers and practitioners (engineers,
designers, data analysts and product and service developers). 2.2.1. Stage 1: planning and searching for the articles
We aim to provide a comprehensive study for DDD in the context of a The first stage involved planning and searching the literature. The
new product, service, or product–service system (PSS) development planning phase consisted of establishing the purpose and protocol of the
process. We make several contributions to both theory and practice literature review (Denyer and Tranfield, 2009; Tranfield et al., 2003).
through a structured systematisation of the distinct perspectives and The protocol was outlined and organised to follow the SLR steps with the
research streams of DDD. aim of addressing the objectives as outlined in Section 2.1. We used the
The paper is organised as follows: Section 2 describes our adopted following electronic databases: ACM Digital Library, Web of Science,
research methodology, Section 3 presents and discusses results coming Scopus, ProQuest and IEEE Library, which offer comprehensive
from the literature and Section 4 provides some conclusions and future coverage of the topics under study. We searched on the keywords
research directions. ‘data-driven design‘, ‘data-enabled design’ and ‘data-led design’ in the
title and author keywords. To gain a better understanding of the key
2. Research Methodology concepts related to DDD, we added other relevant articles cited in the
articles reviewed, following the principle of snowballing (Choong et al.,
2.1. Research Aim and Objectives 2014). We then utilised the keywords found in these articles (e.g.,
‘data-informed design’, ‘data-aware design’, ‘data-centric design’,
This study aims to conceptualise data-driven design through a sys­ ‘design analytics’) to create additional search strings using Boolean
tematic investigation on how data and machine learning are utilised in connectors (AND, OR, and NOT). Due to the vastness of the topic and the
the context of new product, service or product–service system devel­ focus of the research, search terms included data-x design keywords and
opment processes (hereinafter referred to as new product development did not include ‘big data’ ‘machine learning’ and ‘data mining’ limiting
[NPD] processes). The primary objectives of the study are as follows: to papers published between 2012 and 2022.

1. Identify the relationships between concepts of data-x design from the 2.2.2. Stage 2: screening
literature and the associated types of data use, their contributions, A search of the five electronic databases revealed 450 documents for
and methodologies for each of these. screening, as illustrated in Fig. 1. In the screening step, we removed 152
2. Identify the types of data captured and utilised to support product/ duplicate articles, resulting in a set of 298 papers. We established
service development in each phase of the new product process. research inclusion and exclusion criteria to ensure accuracy and
3. Identify the DDD activities and corresponding data types employed comprehensiveness of the review. We screened research articles by title,
for each activity for the development of products and services, abstract and keywords, and we retained all articles that satisfied one of
4. Develop and validate an evidence-based framework of DDD activities the following criteria: (1) papers that discuss how to use big data or
in NPD. machine learning in the new product, service, or product–service system
development process or (2) papers identified through the snowballing
To address these aims, we undertook a two-phase study consisting of approach. We restricted our search to English full-text peer-reviewed
an SLR, which led to the conceptualisation and evaluation of a frame­ papers, such as journal articles and conference proceedings. Grey liter­
work. First, we systematically reviewed DDD literature and distilled it ature (i.e., studies produced outside traditional publishing and distri­
into various data-x design concepts contributing to an understanding of bution channels) and non-open resources were excluded. Then, we
how these concepts are related, including the types of data, their con­ screened the papers by title, keywords and abstract, through which 187
tributions and methodologies for new product/service development. further papers were deemed irrelevant and removed. A further 43 papers
Second, we identified and categorised the DDD activities and the pat­ were excluded after we read the full papers and found they related to the
terns of data usage in each phase of the product/service development use of data but not within the context of new product/service develop­
process that emerged from the literature. We propose a framework ment. We added 72 studies through the snowballing method, resulting in
consisting of seven DDD activities with example cases illustrating how a set of 140 articles that met this study’s inclusion and exclusion criteria
various data types are utilized in the product/service development and were used for data analysis.
process, providing an overview of the types of activities possible with
DDD. We validated and refined the DDD framework through workshops 2.2.3. Stage 3: Descriptive and Thematic Analysis
with participants from industry and academia. We used descriptive and thematic analysis to analyse the papers. The
descriptive analysis involved analysing the selected articles, from which
an initial coding scheme was developed. The thematic analysis involved
three sections reviewing (1) the commonalities and differences between

2
B. Lee and S. Ahmed-Kristensen Computers in Industry 164 (2025) 104206

Fig. 1. Research design and stages for literature review and questionnaire survey in this study.

various data-x design concepts, (2) when and what types of data were
Table 1
captured and utilised and (3) DDD activities. We used the full set of 140
Coding scheme for descriptive analysis of this study.
papers for the descriptive analysis. For the thematic analysis, we
included only empirical studies conducted in a real-world setting (n = Coding schemes Description Sub-codes

98). This was critical to enable separation from research working solely Research type ( The research approach Validation research; evaluation
on the use of data/AI with a purely theoretical approach from work Wiering et al., used in the papers research; solution proposals;
2005) philosophical papers; opinion
conducted for a real-world context, thereby ensuring the framework is of
papers; experience papers.
relevance to industry whilst being academically grounded. Research setting The environment in which Artificial setting; in situ/wild
The process of conducting an SLR, which can be subjective, requires research is carried out setting.
detailed information on the coding and screening decisions made. We Discipline The first author’s research Engineering; design;
developed a coding scheme to enable the synthesis and documenting of discipline management; computer science;
service engineering/service
the final 98 articles. Next, we describe the methodology adopted in
science; miscellaneous
developing the coding scheme and its interreliability. Brown et al. Application area Area where research Consumer products;
(2003) propose the basic categories to capture essential information in activities is observed transportation; manufacturing;
the literature: (1) methodological and substantive features, (2) in­ healthcare; agriculture; urban/
city; miscellaneous; not specified
dicators of study quality and (3) intervention categories and relevant
Research The results and Method/tool; computational
outcome measures specific to the study and driven by the research outcomes conclusions arrived at as approach; framework/model/
questions. We coded the papers based on the first and last categories: the result of research process; guidelines; future
methodological and substantive features, intervention categories and activities research agenda; design artefact;
relevant outcome measures. In addition, the coding scheme included the miscellaneous

publication year, author names, article title, journal in which the article
was published, abstract and keywords. NPD phase in which data/AI was utilised, DDD activities and impact (see
We generated coding schemes in two ways: top-down, beginning Table 2 for definitions).
with existing categories, or bottom-up, emerging from the initial data To evaluate the reliability of the coding scheme, the coders randomly
set. We used establishing codes where possible (i.e., top-down selected 10 % of the studies, and each coded the studies independently
approach), and we only created codes (bottom-up approach) where using the coding scheme to determine interrater reliability. This process
none were found – which is to be expected for fields such as DDD. For resulted in interrater agreement of the following Cohen’s kappa for each
instance, the majority of methodological and substantive feature cate­ coding scheme:
gories were generated with existing codes (Table 1). Additionally, we
categorised and defined the stages of data/AI utilised based on Ulrich • 0.93 for NPD phase in which data captured,
and Eppinger’s (2015) NPD process. We adjusted the remaining coding • 0.90 for data source and context,
scheme, using a bottom-up approach, and updated it to incorporate new • 0.91 for NPD phase in which AI/data utilised,
categories until no new categories emerged. We identified five coding • 0.96 for DDD activity
schemes, each with codes that are mutually exclusive and collectively • 0.96 for impact.
exhaustive: data source and context, NPD phase in which data captured,

3
B. Lee and S. Ahmed-Kristensen Computers in Industry 164 (2025) 104206

Table 2 questionnaire in a face-to-face setting. The questionnaire consisted of

Coding scheme for thematic analysis of this study. two open questions and two multiple-choice questions individually
Coding schemes Definition Sub-codes (Table 3).
The feedback collected from the questionnaire in the first checked
Data Source and What data is about and User generated; Documents;
Context originality of data Sensor generated; synthetic that we had not omitted other DDD activities. The feedback from the
data; Multiple sources. first two workshops predominantly focused on the terminology of the
NPD Phase in The phases in which data is Pre-NPD stage, Planning, activities and led to alternative names (based on their suggestions) in the
which Data captured Concept Development, System revised framework. These were tested in the third and fourth workshops,
Captured Level/Detail Design, Testing &
Refinement, Production Ramp-
where respondents confirmed the suitability of the alternative names.
up, In use, Independent of (Further outlined in Section 3.2.3.) This research obtained ethical
Product Life cycle. approval by the Ethics Committee at the University of Exeter Business
NPD Phase in The phases in which DDD Pre-NPD stage, Planning, School (Ethics approval number 759840).
which AI/ Data activities are observed, and Concept Development, System
Utilised data/machine learning is Level/Detail Design, Testing &
utilised, to achieve a certain Refinement, Production Ramp- 3. Results and discussion
goal up, In use.
DDD Activities A goal achieved through Support business strategy; 3.1. Descriptive analysis
utilising data/machine Change user behaviour; Plan
learning in design process product families, portfolio, life
cycle; Understand users/
The descriptive analysis as illustrated in Fig. 2 depicts the number of
context better; Support publications and publication types. There was a steady increase in
sensemaking; Manage publications overtime. These are published in various places, with the
knowledge; Generate product/ top five journals being Journal of Mechanical Design (n = 9), CIRP Journal
service design ideas;
of Manufacturing Science and Technology (CIRP-JMST) (n = 7), Journal of
Customise/ Optimise products
and services; Improve design Computers and Industrial Engineering (n = 6), Journal of Advanced Engi­
reliability; Assess/predict/ neering Informatics (n = 5) and International Journal of Production
improve the system Research (n = 4). The top conferences were the International Conference
performance; Aid decision on Engineering Design (n = 12), the International Design Conference (n
making
= 7) and the Designing Interactive Systems Conference (n = 5). Most
Impact An effect or influence made The NPD process in practice;
through the proposed Product/service; Platform/ journal papers were engineering focused, whereas conference papers
research outcome system; Business strategy. included design, engineering and interactive systems.
Table 4 summarises the application areas, disciplines, research types
and outcomes, with the number of articles, and we discuss them further
According to Landis and Koch’s (1977) criteria, this reliability level
here. Application area refers to where (which sector) the proposed
is considered ‘almost-perfect agreement’. The main disagreement we
research outcomes were deployed, with Consumer Products (n = 38),
observed was around the NPD phases with the coding of the in-use
Transportation (n = 29) and Manufacturing (n = 21) identified as the
phase, as it is aligned with other phases (i.e., data collected in this
top three application areas, followed by Healthcare, Agriculture and
phase may be used in another phase). For example, whereas one coder
Urban/City (29 papers did not specify the application areas). Discipline
coded in-use phase, the other coded testing & refinement or system level
refers to the first author’s research field, with the top three being En­
design. To address differences that arose, coders met and discussed ar­
gineering, Design, and Computer Science. More specifically, 61 % of
ticles, resolved uncertainty and gained consensus in the coding by
publications were from Engineering, encompassing Systems Engineer­
revisiting the specific article.
ing, Aerospace Engineering, Mechanical Engineering and Engineering
We then summarised the articles using the descriptive and thematic
Design, with multiple application areas and research outcomes. A total
analyses. The descriptive analysis explained the research context, scope
of 17 % of studies were from Design, including Industrial Design and
and methodologies, and the thematic analysis synthesised the key
Service design, 10 % of publications from Computer Science and
findings from the literature resulting in the DDD framework.
Human-Computer Interaction and 8 % from Management.
In terms of type of research, most of the studies were either valida­
2.2.4. Stage 4: evaluating the framework through questionnaire survey
tion research (n = 92), philosophical papers (n = 20), solution proposals
We next evaluated the framework in workshops with industry
(n = 14), experience papers (10) and evaluation research (n = 4). Only
practitioners and scholars. The questionnaire was designed to assess (1)
3 % of publications were conducted in situ; 97 % of studies were per­
whether any additional DDD activities were present that were not
formed in artificial settings. As for research outcomes, 52 studies pro­
captured in the framework and (2) the perceived sensibility of each
duced a framework/model/process, 32 tested method/tools, 18
identified DDD activity. The evaluation categories align with the
established computational algorithms and 9 design artefacts (design
research assessment framework suggested by Ahmed (2001), extending
scholars using ‘research through design’). In total, 66 papers aimed to
the Kirkpatrick, (1959): The evaluation criteria include checking for any
improve activities to create value propositions (design and engineering),
missing categories, confirming the recognition and usage of identified
49 studies on an organisation’s tangible and intangible offerings, and 5
activities within organisations and assessing the clarity and under­
studies focused on a company’s plan for achieving their vision.
standing of the terminology used to describe the activities. We held four
The descriptive analysis revealed that the majority of publications
workshops from May to September 2023. In total, 34 responses were
collected, including 27 people from industry and 7 people from
academia or government (Lee and Ahmed-Kristensen, 2024). The 12 Table 3
respondents from the first workshop were from across the foundation Questions for evaluation of the framework.
industries (including cement, glass, ceramics, paper and metals and bulk Q1 Are there any other ways using data or machine learning in your company?
chemicals, whereas 22 respondents from the remaining three workshops Q2 From the framework, which of these do you use in your organisation? (Tick all
were from across industries (n = 17) and academia (n = 5). The work­ that apply)
Q3 Are there any of these words you find difficult to understand or you don’t agree
shop participants’ roles varied within organisations, including CTO, with? (Tick all that apply)
Business development manager, brand manager, and cloud developer. Q4 How would you change the name of DDD activity if you don’t think it delivers
We presented the framework to the participants and then distributed the its definition properly to make sense to you?

4
B. Lee and S. Ahmed-Kristensen Computers in Industry 164 (2025) 104206

Fig. 2. Publication trend by year.

Table 4
Metadata (with the number of articles).
Categories Application Area Discipline Research Type Research Outcomes Impact

Sub-codes Consumer products 38 Engineering 85 Validation research 94 Model/ 52 NPD process/ practice 66
framework
Transportation 29 Design 24 Evaluation research 4 Method/ 32 Product/ service 49
tool
Manufacturing 21 Computer science 14 Solution proposal 12 Computational algorithms 18 Business strategy 5
Healthcare 11 Management 11 Philosophical papers 20 Guidelines 10 Platform/ system 3
Agriculture 2 Service science 3 Opinion papers 0 Design artefacts 9 Not specified 17
Urban/City 2 Miscellaneous 3 Experience papers 10 Research agenda 6 ​ ​
Miscellaneous 8 ​ ​ ​ ​ Miscellaneous 13 ​ ​
Not specified 29 ​ ​ ​ ​ ​ ​ ​ ​

are conceptual and exploratory, with few (3 %) reporting an application utilisation.

in a real-world context. Thus, more contributions are needed that vali­ Fig. 3 illustrates the expanding field of DDD and related concepts
date new methods in a real-world context. Over half of the studies were from the selected literature over the 10-year review period. These con­
conducted by engineering scholars developing, validating and evalu­ cepts are mapped out according to the use of data (sole use of big data or
ating frameworks, models and methods through case studies. The combined use of big and thick data) and the design outcome (tangible
research field is more mature in tangible products than intangible ser­ design outcomes or intangible strategy). The concepts of big data and
vices. The studies evaluated research impact predominantly from tech­ thick data are often interpreted differently; herein, we have adopted
nical perspectives, rather than strategic or innovation perspectives. Hong et al., (2022) definitions: ‘Big data’ refers to massive data sets that
can be computationally analysed to reveal associations, patterns and
3.2. Thematic Analysis trends, and ‘thick data’ refers to qualitative information collected and
analysed to develop a holistic understanding of a phenomenon or
3.2.1. Research Objective 1. Concepts related to data-driven design practice, including considering the relevant context, how people
To understand how different concepts are being used, we analysed construct meanings and why they act in certain ways.
140 papers and found 16 that contained definitions of DDD or related The term ‘data-driven design’ originally pertained to big data
concepts. Table 5 defines the six unique concepts used in literature. focusing on tangible design outcomes, as shown in the bottom left-hand
‘Data-driven design’, first coined by Domazet et al. (1995), is the most side of the fourth quadrant. However, as the field of DDD developed,
commonly used term in our data set, but the studies lack a unified studies covering the use of big and thick data for both tangible and
definition. For instance, Kusiak (2009) limits DDD to the utilisation of intangible outcomes emerged. ‘Data-aware design’ is positioned in the
data in generating a large number of design alternatives. King et al. same quadrant because it aims to effectively use data to relate design
(2017) and Quinones-Gomez (2021) also limit DDD as using data for work to business goals (Churchill, 2012; King et al., 2017). In the context
design decisions. In contrast, Briard et al. (2021) broadly define the term of the balanced use of big and thick data, note the considerable number
‘the usage of data during the design process’, and other scholars (Kim of studies on ‘data-enabled design’ mapped on the right-hand side of the
et al., 2017; Lim et al., 2018) stress exploiting data and valuable infor­ fourth quadrant – for example, employing personal health data, activity
mation in the design process, such as the informatics-based design data and interview data for proposing intangible strategies (Noortman
method, which emphasises data collection and valuable information et al., 2022; Versteegden et al., 2022). We also found ‘data-enabled
enabled by emerging technologies. ‘Data-enabled design’ was the sec­ design’ studies related to more tangible design outcomes, such as per­
ond most used term, reflecting an exploratory design practice of utilising sonalising dementia care service and improving the interaction between
qualitative and quantitative data (IoT devices) to determine design di­ healthcare professionals and parents (Bogers et al., 2016; 2018; Wang
rections (van Kollenburg and Bogers, 2019). Only a few papers used the et al., 2021). Alongside ‘data-enabled design’, other concepts (data-a­
terms ‘data-informed design’ (Diels et al., 2022; King et al., 2017), ware design, data-informed design, data-centric design and design an­
‘design analytics’ (King et al., 2017; Lewis and van Horn, 2013), alytic) also use big and thick data focussing on user experiences.
‘data-centric design’ (Ortega et al., 2022) and ‘data-aware design’ (King To summarise, studies show a variety of data-x design concepts.
et al., 2017), which shared similarities in the purpose and type of data Initially, ‘data-driven design’ primarily focused on using big data as the

5
B. Lee and S. Ahmed-Kristensen Computers in Industry 164 (2025) 104206

Table 5 decision making, whereas ‘data-driven design’ encompasses this activity

Example papers related to the selected concepts of design and data integration. and more. ‘Data-driven design’, ‘data-aware design’ and ‘data-enabled
Concepts Authors Definitions design’ in the top right-hand quadrant (Fig. 3) have a more compre­
hensive problem space than ‘data-informed design’, ‘data-centric
Data -driven (Kusiak, 2009) An enabler for innovation methods
design enhancing creativity by generating design’ and ‘design analytics’, which are about finding design solutions
large number of design alternatives by supporting design decision making. In addition, our analysis dem­
(Kim et al., 2017) An opportunity granted by the onstrates that the term ‘data-driven design’ is utilised broadly to
increased diffusion of cyber-physical describe the following in NPD process: the substantial use of big data,
systems and concurrent reduction of
technological barriers for data
the balanced use of big and thick data and the use of machine learning.
collection and communication Thus, DDD can be defined as ‘a design approach to utilise big data or data
(Lim et al., 2018) The informatics-based design method, science algorithms alongside thick data for product and service innovation’.
which emphasises ‘creating valuable In relating the different concepts, we found that ‘data-driven design’ was
information’
an umbrella term, and each concept was not mutually exclusive but
(Bertoni, 2020) The use of any kind of data science
algorithm that supports a specific rather had overlapping elements.
phase of the product development
process 3.2.2. Research Objective 2. When and what types of data are captured and
(Lee and Methods, approaches and processes utilised
Ahmed-Kristensen, using data to assist the development of
2023) products and services
This section aims to explain where and what data is captured and
(Feng et al., 2020) The process of mining the relevance utilised across the NPD process and DDD activities. Herein, we use the
and hidden pattern of things through term ‘NPD’ in a broad sense, not limited solely to products and services.
modelling and analysis of this data to We include studies that employ data in the context of business strategy
assist product design
development, but only if the data is generated in the context of product/
(Gorkovenko et al., The approach of utilising data from
2020) devices in the wild for design purposes, service development. Fig. 4 depicts where data is captured and utilised
facilitated by the use of sensors and in the NPD process as synthesised from the identified studies. Data in­
functional data, as well as various dependent of the product life cycle (e.g., patent, open source, synthetic
qualitative sources, such as feedback data) is situated out of the cycle. The colour legend implies the stage
from users
where data is captured (left), and the captured data are fed back to the
(Quiñones-Gómez, The data being used determines design
2021) decisions development cycle (right). In this study, the seven stages of the NPD
(Briard et al., 2021) The usage of data during the design process are modified based on Ulrich and Eppinger’s (2015) NPD pro­
process cess by adding two stages: (1) Pre-NPD stage is identified as any business
Data-enabled (van Kollenburg and A situated design practice of using
development activities closely related to NPD (e.g., developing finan­
design Bogers, 2019) qualitative and quantitative data as a
creative material for designing cial, resource, technology, product, supply, and business process strat­
intelligent ecosystems egy), and (2) in-use stage refers to when a product is launched, deployed
Data- (King et al., 2017) A practice in which design and implemented by end users, whilst businesses continue to improve
informed professionals utilise data as one of the their products and services through data collected from sensors (Lee,
design inputs into their decision-making
2022; Lee et al., 2018). Fig. 4 shows that data is captured mainly from
process
(Hou and Jiao, 2020) A method based on inverse thinking the in-use stage, followed by the concept development stage, indepen­
about problem solving dent of the product life cycle (i.e., it can be utilised across multiple
(Diels et al., 2022) The process of a design team using data NPDs), and production ramp-up stage. In contrast, little or no data is
as one of many references, including
collected from the planning, pre-NPD, testing and refinement stages.
design intuition, creative expression
and qualitative feedback, to inform
Captured data was utilised in various stages of the NPD process, with
their decisions more than half of the data used during concept development.
Design (van Horn et al., 2012) The processes and tools supporting the The DDD papers predominantly used document(s) as a data type,
analytics transformation of design-related data followed by ’sensor(s)-generated data’, ’sensor-generated data with
to suit design decision-making
documents’, ’user-generated data’ and ’sensor(s), user-generated data
processes
(King et al., 2013) The process of inspecting, cleaning, with document’ (Table 6). The types of documents varied, including
transforming and modelling data to internal organisation data (e.g., supply-chain data, point of sales data,
extract knowledge, which could be requirement specification document, use scenario), service provider
valuable to generate and evaluate new
data (maintenance cost), experts’ domain knowledge, market research–/
design solutions
Data-centric (Ortega et al., 2022) An approach to obtain valuable
designer-generated data (e.g., observation, questionnaire, interview),
design insights from the mix of quantitative patent data set and open-source data (e.g., Google images, Wikipedia).
and qualitative, structured and Most documents were used during concept development to generate
unstructured data to inform and ideas and support decision making and sensemaking. The second most
support design decisions
common data type was sensor-generated data, which consisted of (1)
Data-aware (King et al., 2017) An approach that enables designers to
design be more systematic and more product and space usage (e.g., log data, pressure, timestamp, acceler­
ambitious about design and product ometers, GPS, sound); (2) product performance (e.g., CPU utilisation,
development process, helps shed light battery capacity, RAM, storage) and (3) environment (e.g., weather).
on how to determine different user Over 50 % of the research papers that solely used sensor data combined
types and needs and expectations and
effectively relates the design work to
multiple sensors.
the business goals and metrics Fig. 5 illustrates the types of data mapped against the stages of the
NPD process and whether they come from a single source or combined.
In these figures, one colour block implies one study; thus, the total
sole resource for tangible design outcomes. However, these approaches number of blocks is 98. We identified four types of single data sources:
have since been diversified, giving rise to terms such as ‘data-enabled documents (n = 32), sensor(s) collected (n = 19), user-generated (n =
design’, each with its own unique approach. For example, our review 13) and synthetic data (n = 4), accounting for 69 % of studies. 31 % of
indicates that ‘data-informed design’ is primarily employed to support studies combined multiple types of data, such as document(s) with

6
B. Lee and S. Ahmed-Kristensen Computers in Industry 164 (2025) 104206

Fig. 3. Expanding field of DDD and relevant concepts, 2012–2022.

Fig. 4. Where data is captured and utilised in the NPD process.

sensor(s)-generated data (n = 15), sensor(s)- and user-generated data Kim et al., 2016). Customer ratings and feedback on products and ser­
with document(s) (n = 8), sensor(s)- and user-generated data (n = 5) or, vices were the most common type of user-generated data. Qualitative
documents with user-generated data (n = 2). Fig. 6 big data of this type was primarily analysed using text mining or opinion
Sensor(s)-generated data combined with documents was the third mining to identify the concepts and feature specifications of products
most common type of data, often a combination of a product’s perfor­ and services during the concept development phase. In contrast, syn­
mance or usage data with user interviews, demographics, experts’ thetic data (e.g., crash simulation data, data extracted from simulation
domain knowledge or product data. The studies employing multiple software) was used during the concept development, detail design and in
data sources highlight the integration of big and thick data to improve the in-use stage.
design practice. We observed this combined use of multiple data mainly A key contribution of this analysis is that two-thirds of studies on
in the in-use phase, particularly for understanding user context and DDD rely solely on a single data source, and the combined use of mul­
generating products and service design ideas. This phenomenon in­ tiple data sources is predominantly observed at the early stages of new
dicates that the use of big data alone is insufficient for these activities. product/service development. This finding highlights the ongoing need
Only few papers utilised data from domain experts with other data to complement big data with thick data, in particular when under­
sources despite arguments that the integration of domain knowledge standing contextual needs and user needs and behaviour. With this shift,
with big data is critical in improving detection accuracy, timeliness and an evolution of research methods has occurred, from qualitative to
transparency or in proposing new innovative ideas (Briard et al., 2021; quantitative or mixed approaches, which has an implication for the skills

7
B. Lee and S. Ahmed-Kristensen Computers in Industry 164 (2025) 104206

Table 6
Data source categories and description.
Data Source Description

Documents Information generated by the organisations, third parties or

experts or freely available data in the form of documents.
Examples: simulation data set (Du and Zhu, 2018); market
survey data (Wang et al., 2016); historic data on marketing,
transaction records (Georgiou et al., 2016; Bang and Selva,
2016); technical reports and product performance data (Lim,
2016); demographic data, users’ head shape data (Stavrakos
et al., 2016; Tan et al., 2020); open data source (Ahmed and
Fuge, 2018); patent data set (Venkataraman et al., 2017; Yoon
et al., 2017; Wodehouse et al., 2018)
Sensor collected ’Sensor generated data’ can be categorised into (1) product and
data space usage, (2) product performance and (3) environment.
Examples: data about energy generation and consumption (
Bourgeois et al., 2014); motion sensors (attach to several
objects, e.g., refrigerators, bikes, chairs) (Lovei et al., 2020);
pressure data of the users breathing (Zheng et al. 2019);
operating data (e.g., CPU utilisation, battery remaining, RAM,
storage) on smartphones (Zhang et al., 2017, 2019); user
behaviour data on running machine (Hollauer et al., 2018);
animal behaviour data (Lee et al., 2022b)
User generated Any data generated by users, including ratings and comments
data on products and services, video recordings, individual daily
reports and manual inputs of product use. Examples: bike riding Fig. 6. Types of data captured.
data (Wang et al., 2021); online reviews, customer feedback (
Chiu and Lin, 2018; Jin et al., 2016; Zhang et al., 2017; Chen
et al., 2019); Social media data (Tuarob and Tucker, 2015) sets and training of future designers.
Synthetic data Information that is artificially generated rather than produced
by real-world events. Examples: Okoh et al. (2017); Xiong et al. 3.2.3. Research Objective 3. DDD activities
(2019) In this section, we report the DDD activities identified from the SLR
Multiple sources The combination of different data sources or different data sets
that were conceptualised as part of the framework. We analysed the set
Examples: sensor(s)- and user-generated data (Bogers et al.
2018; Kim et al. 2016 Wang et al. 2021; Zheng et al. 2019); of 98 documents to identify how data and machine learning were
sensor(s) and experts opinions (Bertoni et al. 2017); sensor(s) exploited in the product/service development process context. Initially,
and document (Khoshkangini et al. 2020; van den Heuvel et al., we identified 10 activities, then categorised them into seven DDD ac­
2020; Zheng et al. 2020); sensor(s), users, and documents (
tivities: visioning, discovering, translating, generating, optimising, sus­
Bogers et al. 2016; Bogers et al. 2018; Versteegden et al. 2022);
sensor(s), users and experts opinions (Noortman et al. 2022); taining and validating. During the validation portion of our research, the
sensor(s), users, documents and expert opinions (Wang et al. questionnaire survey described in Section 2.2.4 showed that the names
2019; Zheng et al. 2019) of four of these categories (Visioning Translating, Optimising and Sus­
taining) were considered unclear. Through a process of collecting and
iterating on alternative names and further validation, we ultimately
identified the final set of seven DDD activities: Planning, Discovering,
Defining, Generating, Customising, Maintaining and Validating. The
follow subsections explain each activity, summarised with sub-activities
and examples in Fig. 7.

3.2.3.1. Planning. Planning includes two subcategories of DDD activ­

ities, namely ‘shaping business strategy’ and ‘planning product families
and portfolio’. Planning is the use of data or machine learning that helps
shape organisations’ aspirations and goals, set the strategic level of
design or plan product families and portfolios. For this DDD activity,
customer preference or purchase data was the most common type of data
used to shape the business strategy. Data and machine learning played a
significant role in supporting companies in adopting servitisation or in
the planning of PSSs. Studies employed data sources such as patents and
customer preferences to predict optimal product families or upcoming
product trends informing product portfolios. Common methodologies
adopted included text mining, demand modelling and k-means clus­
tering algorithms.
3.2.3.1.1. Shape business strategy. DDD activities support building
business strategies and ecosystem with research outcome(s) contrib­
uting beyond the NPD process to extend to business activities and
strategy. Studies in this category included scaling from non-connected
products to smart products, to product systems (one to multiple con­
nected products) and ecosystems (system of systems)(Bogers et al.,
2018; Porter and Heppelmann, 2014; Zheng et al., 2019). For example,
Fig. 5. Single or multiple sources of data. Watanabe et al. (2020) introduce a conceptual framework to support
ecosystems of connected products, focussing on the interrelational and
interdependent nature of IoT systems. Similarly, Zambetti et al., (2019)

8
B. Lee and S. Ahmed-Kristensen Computers in Industry 164 (2025) 104206

Fig. 7. Scheme of DDD activities.

proposed framework supports product-based firms to move to serviti­ patients who have undergone bariatric surgery (Versteegden et al.,
sation by exploiting data availability. 2022).
3.2.3.1.2. Plan product families and portfolios. Despite the well-
established individual research area of product planning and data ana­ 3.2.4.2. Change user behaviour. Studies focussing on user behaviour
lytics, scholars argue that research is absent in DDD product planning. change utilised a data-enabled design approach, a situated design
Thus, research studies include new approaches and frameworks for data practice of using data as a creative material for designing intelligent
utilisation to plan product families and portfolios – for example, a low- ecosystems (van Kollenburg and Bogers, 2019). Examples include an
risk approach to product recommendation (Yoon et al., 2017), a sys­ intelligent ecosystem designed to support a healthy lifestyle change for
tematic approach for integrating usage data into the planning PSS bariatric care (Lovei et al., 2020) or post-surgery (Jansen et al., 2020;
(Kammerl et al., 2016) and a framework for demand trend mining al­ Lovei et al., 2020) and to support medication adherence (van den
gorithms for the predictive life cycle of a product (Ma et al., 2014). Heuvel et al., 2020). Other studies propose framework or guidelines to
design solutions to trigger sustainable behaviour, often referred to as
3.2.4. Discovering design for behaviour change (Bourgeois et al., 2014; Montecchi and
In the discovering activity, discovery of both the problem (often Becattini, 2021).
referred to as problem framing or task clarification within design liter­
ature) and the user occurs. To support the discovery of user patterns and 3.2.5. Defining
the meaning behind their behaviour, several studies employ design Defining studies aim to define the specifications (or requirements) of
probes to collect specific user-generated data. However, because most NPD, encompassing two subcategories: ‘support sensemaking’ and
studies focus on understanding the meaning behind user behaviour, few ‘manage knowledge’. These activities were typically undertaken using a
studies propose the use of machine learning algorithms alone in this single type of data, rather than combining multiple data, for example,
DDD activity, instead employing sensor-generated data or combining user information data (documents), customer feedback on products/
sensor-generated data with other thick data related to product and services (user generated) or sensor data from the product in use.
service usage to facilitate the understanding of user contexts. We iden­ Example studies included the analysis of online customer reviews using
tified two subcategories of discovery, as outlined next. methods such as opinion mining, web mining and network analysis to
formulate recommendations for product and service functions or using
3.2.4.1. Understand users and context better. DDD can enhance existing customer feedback along with knowledge discovery methods such as the
approaches to understanding users and context to overcome limitations knowledge sourcing framework or graph-based elicitation methods for
of conventional methods (observation, empathic design, and interview) data-driven knowledge management in a closed-loop for smart PSS.
and their subjective and time-consuming nature (Engel and Ebel, 2019;
Stavrakos and Ahmed-Kristensen, 2016; Zheng et al., 2019). Approaches 3.2.5.1. Support sensemaking. When defining requirements, a compre­
enable the capture of latent customer needs or intrinsic product re­ hensive understanding of end users at a large scale and of the complex
quirements (Lakoju et al., 2021) or the actual usage context (van den needs of stakeholders is necessary. Studies utilised data or machine
Heuvel et al., 2022). Stavrakos and Ahmed-Kristensen (2016) employ learning to support this activity, such as Wodehouse et al. (2018), who
anthropometric data sets to cluster users and inform the design of var­ develop a model for database interpretation for user-centred design, and
iants to maximise comfort, leading to customised products. A number of Jin et al. (2016). who utilise big consumer data to explore value and
studies demonstrate how data enables a better understanding of users requirements. Others propose approaches for the identification of key
and context in concept development and in-use phases for improving user experience factors (Yang et al., 2019), linking market understand­
product and service design (Zheng et al., 2019) – for example, ing to technology features (Wang et al., 2016) and exploring the design
employing PSS usage data to better understand customers (Hollauer space (Xiong et al., 2019). Studies in this vein propose various algo­
et al., 2018) and combining quantitative and qualitative data for an rithms for data-driven function formulation for product/service design
improved understanding of co-responsibilities in the households of (Zhang et al., 2017; Chen et al., 2019).

9
B. Lee and S. Ahmed-Kristensen Computers in Industry 164 (2025) 104206

3.2.5.2. Manage knowledge. DDD is used to support knowledge man­ objective data-driven approach to selecting a final concept design
agement in the NPD process. Research outcomes include data manage­ (Georgiou et al., 2016). In addition to validating concepts or ap­
ment approaches for requirement elicitations, such as graph-based proaches, Kushiro et al. (2014) and Bertoni et al. (2017) also focus on
requirement elicitations (Wang. et al., 2019, 2021), an ontology-based validating requirements.
framework (Maleki et al., 2018) and an interactive method for sup­ To summarise, the thematic analysis shows that DDD research ac­
porting designers with high-level knowledge discovery in tivities can be synthesised into seven DDD activities within the NPD
multi-objective design spaces (Bang and Selva, 2016). Others aim to context, providing a rich overview of the field. These DDD activities
support mapping product features and attributes to consumer percep­ impact NPD processes, systems, products or services, product platforms
tions, ultimately reducing potential bias and cognitive loads (Ghosh and business strategy. Each activity has its own distinct focus and
et al., 2017). approach. For example, ’discovering’ predominantly combines data
types to understand the meaning behind patterns, while ’maintaining’
3.2.6. Generating involves prediction and simulation models using product usage or per­
During the concept development phase, data or machine learning formance data. To facilitate an understanding of each of these activities,
plays a crucial role in generating product/service design ideas. The we have developed a framework of DDD activities, which we presented
majority of studies utilise documents, including patents, online reviews in the next section.
and Wikipedia and Google images, or they used a combination of sensor-
collected data (from vehicles) and documents. These studies demon­ 3.3. A Data-Driven Design framework
strate the effectiveness of network-based methodologies, semantic tools
and morphological matrices for generating product and service concept From the thematic analysis, we developed a framework of DDD ac­
ideas. Several scholars have explored the use of patent data to improve tivities within the NPD process context, as shown in Fig. 8. This
ideation (Venkataraman et al., 2017) and to identify potential unex­ framework represents how data or machine learning can benefit the
plored technologies for future products through a network-based development of new products/services, as determined by consolidating
methodology (Song et al., 2017). Kim et al. (2016) introduce concep­ disparate research efforts over a 10-year period from when the field first
tual frameworks of informatics-based service idea generation to support emerged into the identified activities. The evidence-based framework
the use data to generate concepts (Lim et al., 2018), and others have comprises seven DDD activities with selected enablers and each activity
combined approaches to understand perception of products and utilise is mapped across the relevant NPD phases. For instance, planning is
learning from data collected to evaluate or generate concepts (Mata frequently observed during pre-NPD, planning and in-use phases, and
et al., 2017). discovering, defining and generating are often observed in concept
development and in-use phases. Customising occurred mainly during
3.2.7. Customising system-level design, detail design and in-use phases, whereas main­
Customising is DDD activity that helps develop customised or opti­ taining occurred predominantly during the in-use phase. Validating is
mised products and services. Due to the limitation of the traditional positioned across DDD activities across all NPD phases, as the NPD
product optimisation method (Zhang et al., 2019), scholarly inquiry on process is a series of making design decisions. The framework does not
developing customised products and services used data or machine propose a step-by-step process; rather, it identifies all possible DDD
learning. These studies tended to introduce customisation processes or activities based upon prior research.
systems as validation research conducted in real-world settings. Several Fig. 9 illustrates the types of data types used for each DDD activity
studies demonstrate that data can support customising or optimising identified. This is extended from our earlier work (Lee and
products and services, including personalising dementia care services Ahmed-Kristensen, 2023) through the validation exercise as described in
through data visualisations in the digital platform (Wang et al., 2021) Section 2.2.4 and an extensive review. From Section 3.3.2, we identified
and a data-driven approach to customise products to maximise comfort that the majority of studies use a single type of data for all the activities
for individuals or user segments (Ellena et al., 2018; Stavrakos and except discovering. The aim of discovering is to understand contexts,
Ahmed-Kristensen, 2016). user experiences or the motivation behind certain behaviours, and big
data in this context can lack this richness; thus, different data sources are
3.2.8. Maintaining combined. In general, we observed the combined use of multiple data in
Predictive maintenance is an established research area, advanced the early stage of the NPD process for planning, discovering and
through the use of data to support systems, products or services in defining, illustrating that despite advances, big data alone is insufficient.
predicting or assessing performance, enabling, for example, parts to be
replaced ahead of a potential failure. Geiger and Sarakakis (2016) and 4. Conclusions
Shin et al. (2015), propose data modelling and analysis techniques to
investigate performance feature data and increase product performance. This article undertakes a systematic literature review of how data
Typically, researchers use product usage and performance data to pro­ and machine learning are utilised in the context of a new product (or
pose prediction or simulation models. For instance, Ströer et al. (2018) service or PSS) development. The review identifies data-x design con­
and Okoh et al. (2017) introduce simulation models to support condition cepts from the literature and the relationships between them and the
monitoring and to predict failure in early product development stages to associated methodologies and contribution; in addition, we identify the
ascertain whether detailed maintenance, repair or overhaul is needed. types of data utilised to support each phase of product/service devel­
opment and data-driven design (DDD) activities undertaken in the
3.2.9. Validating process. In addition, this study conceptualises and validates an evidence-
Validating supports effective decision making and increasing design based framework of DDD activities within the NPD process.
reliability whilst reducing uncertainty, complexity and risk. Data is We identified DDD as an overarching term that encompasses various
utilised to validate and support design decisions due to design un­ concepts of data-x design. This finding reflects that DDD has expanded
certainties in complex system engineering (Du and Zhu, 2018) or to from the use of big data for tangible design outcomes to combined use of
support early decision making (Bertoni et al., 2017). Typically, studies big and thick data for both tangible and intangible outcomes. We
used documents or documents with sensor data about product specifi­ describe the methodologies for each concept of the data-x design, which
cation, performance or marketing history. Several studies propose can researchers clearly identify their approaches and the related ap­
data-driven approaches for decision support tool, approaches and sys­ proaches and adopt appropriate methods.
tems, such as a performance target setting strategy (Lim, 2016) and an For NPD (including services and PSSs), the type and the range of data

10
B. Lee and S. Ahmed-Kristensen Computers in Industry 164 (2025) 104206

Fig. 8. Data-Driven Design Framework in the NPD process.

Fig. 9. Data types used for each DDD activity (modified from Lee and Ahmed-Kristensen, 2023). One block represents one paper; 98 papers are mapped across the
DDD activities.

sources varied. Studies captured the data predominantly from the in-use changing nature of the DDD will propose new subjects, concepts and
phase (a product/service/PSS in operation) and fed the data back to the approaches. However, this period enables a clear conceptualisation of a
concept development phase. We observed that over two-thirds of the relatively nascent field of DDD as it emerged within NPD processes and
studies used a single data type, and documents were the most common key related concepts. Another limitation, due to the vastness of the topic,
type of data, followed by sensor data. As research efforts in DDD have is the selection of articles, which mainly include those explicitly indi­
increased, it is timely to draw together this learning to identify key DDD cating DDD and relevant concepts, which inevitably omits some relevant
activities. We consolidated research efforts over the review period into articles and the review approach limits a thorough qualitative per­
seven DDD activities and conceptualised them into a framework for DDD spectives. Despite these limitations, this study is distinctive from pre­
activities, which we validated with industry and academic data (Fig. 8). vious papers as our theoretical foundation is to situate the types of data
The seven DDD activities and enablers we identified were: planning, capture and utilisation to provide a comprehensive understanding of
discovering, defining, generating, customising, maintaining and vali­ data- or machine learning–driven design activities within the new
dating. For the discovering activities, more studies relied on the com­ product/service development process and provide a conceptualisation
bined use of multiple data than with a single type of data (Fig. 9), thus of these activities.
highlighting how discovery of the problem or user needs remains an ill- This paper makes three contributions. First, we provide a compre­
defined task with multiple sources of data needed. hensive definition of DDD and show how disparate concepts are inter­
Like any review paper, this study is not free of limitations. First, the related, thus enabling practitioners and scholars to compare and select
research period (2012–2022) is limited; in the future, we anticipate the DDD and related concepts with a fuller understanding of design

11
B. Lee and S. Ahmed-Kristensen Computers in Industry 164 (2025) 104206

outcomes and data needs. Second, through the overview of design ac­ Bertoni, A., Larsson, T., Larsson, J., and Elfsberg, J. (2017) Mining Data to Design Value:
A Demonstrator in Early Design. In International Conference on Engineering Design.
tivities and the use of different types of data across new product/service
Vancouver, Canada.
development processes, practitioners can be challenged to rethink data Bogers, S., Deckers, E., van Kollenburg, J., Frens, J., Hummels, C., Rutjes, H., 2018.
utilisation to create value. Our evidence-based framework enables in­ A showcase of data-enabled design explorations. CHI – Comput. Hum. Interact. Montr.
dustry practitioners to develop appropriate DDD activities to product/ https://doi.org/10.1145/3170427.3186543.
Bogers, S., Frens, J., van Kollenburg, J., Deckers, E., Hummels, C., 2016. Connected baby
service development in their business context based on the framework bottle: a design case study towards a framework for data-enabled design. DIS. Brisb.,
and provides academics with a research platform to build on. As a Aust. https://doi.org/10.1145/2901790.2901855.
methodological contribution, we define a novel coding scheme for the­ Bogers, S., van Kollenburg, J., Frens, J., Djajadiningrat, T., 2016. Data-enabled-design: a
situated exploration of rich interactions. DIS. Brisb., Aust. https://doi.org/10.1145/
matic analysis of DDD, which differentiates our research from existing 2908805.2913015.
SLR papers of DDD. Finally, our research supports scholars in navigating Bourgeois, J., van Der Linden, J., Kortuem, G., Price, B., Rimmer, C., 2014. Conversations
a multidisciplinary research landscape, shedding light on current with my washing machine: an in-the-wild study of demand shifting with self-
generated energy. UbiComp 2014 – Proceedings of the 2014 ACM International Joint
knowledge and conceptualisations of DDD for increased cross- Conference on Pervasive and Ubiquitous Computing. Association for Computing
fertilisation of research. With the changing nature of products and ser­ Machinery, Inc. https://doi.org/10.1145/2632048.2632106.
vices, for researchers, engineer designers, data analysts and product/ Briard, T., Jean, C., Aoussat, A., Véron, P., Le Cardinal, J., Wartzack, S., 2021. Data-
driven design challenges in the early stages of the product development process. Int.
service developers, a comprehensive understanding of how data can be Conf. Eng. Des. Gothenbg., Swed. https://doi.org/10.1017/pds.2021.85.
used is crucial to understand the type of activities that are possible and Brown, S., Upchurch, S., Acton, G., 2003. A framework for developing a coding scheme
how they impact products and services. for meta-analysis. West. J. Nurs. Res.
Chattopadhyay, P., Mondal, S., Bhattacharya, C., Mukhopadhyay, A., Ray, A., 2017.
Finally, we propose an agenda for future research in DDD domain.
Dynamic data-driven design of lean premixed combustors for thermoacoustically
First, researchers should conduct research in real-world settings and stable operations. J. Mech. Des., Trans. ASME 139 (11).
validate new methods within a broader context. Second, diverse Chen, D., Zhang, D., Tao, F., Liu, A., 2019. Analysis of customer reviews for product
research applications of DDD, including intangible services, are required service system design based on cloud computing. Procedia CIRP 83, 522–527.
Chiu, C., Lin, Z., 2018. Utilizing text mining and Kansei Engineering to support data-
to demonstrate potential in these areas, beyond physical products. In driven design automation at conceptual design stage. Adv. Eng. Inform. 38,
terms of research impact, many studies are advancing technical per­ 826–839.
spectives, and an opportunity exists to foster innovation beyond the Choong, K., Galgani, F., Dunn, A., Tsafnat, G., 2014. Automatic evidence retrieval for
systematic reviews. J. Med. Internet Res. 16 (10).
technical. The surge of research on artificial intelligence approaches is a Churchill, F., 2012. From data divination to data-aware design. Interactions 19 (5),
rapidly changing landscape for new product service development, and 10–13.
we expect far more future research, interesting discussions, and a larger Denyer, D., Tranfield, D., 2009. Producing a systematic review. In: Buchanan, D.,
Bryman, A. (Eds.), The Sage Handbook of Organizational Research Methods. Sage,
impact in this topic area. London.
Diels, C., Stylidis, K., Safa, F., Charwick, C., Haramoto, H., Shu, Y., Wu, J., and Harrow,
D. (2022) Towards a Data-Informed-Design Framework for Autonomous Vehicle
CRediT authorship contribution statement
Design. In International Conference on Design and Semantics of Form and Movement.
Hongkong, China.
Boyeun Lee: Writing – review & editing, Writing – original draft, Domazet, S., Choong, N., Sng, D., Ho, C., Lu, Y., 1995. Active data-driven design using
Visualization, Validation, Methodology, Investigation, Formal analysis, dynamic product models. CIRP Ann. – Manuf. Technol. 44 (1), 109–112.
Du, X., Zhu, F., 2018. A new data-driven design methodology for mechanical systems
Data curation, Conceptualization. Saeema Ahmed-Kristensen: Writing with high dimensional design variables. Adv. Eng. Softw. 117, 18–28.
– review & editing, Writing – original draft, Supervision, Methodology, Ellena, T., Mustafa, H., Subic, A., Pang, Y., 2018. A design framework for the mass
Funding acquisition, Validation, Conceptualization. customisation of custom-fit bicycle helmet models. Int. J. Ind. Ergon. 64, 122–133.
Engel, C., Ebel, P., 2019. Data-driven service innovation: a systematic literature review
and development of a research agenda. The 27th European Conference on
Information Systems (ECIS). Stockholm and Upsala, Sweden.
Declaration of Competing Interest Feng, Y., Zhao, Y., Zheng, H., Li, Z., Tan, J., 2020. Data-driven product design toward
intelligent manufacturing: A review. Int. J. Adv. Robot. Syst. 1–18. https://doi.org/
10.1177/1729881420911257.
The authors declare that they have no known competing financial Gann, D., Parmer, R., Cohn, D., 2014. The new patterns of innovation. Harv. Bus. Rev.
interests or personal relationships that could have appeared to influence 1–10.
the work reported in this paper. Geiger, C., Sarakakis, G., 2016. Data driven design for reliability. In 2016. Annu. Reliab.
Maintainab. Symp. . (RAMS). https://doi.org/10.1109/RAMS.2016.7448023.
Georgiou, A., Haritos, G., Fowler, M., Imani, Y., 2016. Attribute and technology value
Acknowledgments mapping for conceptual product design phase. In: In Proceedings of the Institution of
Mechanical Engineers, Part C: Journal of Mechanical Engineering Science, 230. SAGE
Publications Ltd.
This work was supported by the Engineering and Physical Sciences Ghosh, D., Olewnik, A., Lewis, K., Kim, J., Lakshaman, A., 2017. Cyber-Empathic Design:
Research Council (grant number EP/T022566/1). DIGIT Lab is a Next A data-driven framework for product. J. Mech. Des. 139 (9), 1–12.
Gorkovenko, K., Burnett, D., Thorp, J., Richards, D., Murray-Rust, D., 2020. Exploring
Stage Digital Economy Centre. We thank all the participants of industry
the future of data-driven product design. CHI- Comput. Hum. Interact. Virtual.
who undertook workshops to validate the framework and the reviewers https://doi.org/10.1145/3313831.3376560.
for their valuable comments. Hollauer, C., Shalumov, B., Wilberg, J., Omer, M., 2018. Graph databases for exploiting
use phase data in product-service-system development: a methodology to support
implementation. Int. Des. Conf. https://doi.org/10.21278/idc.2018.0399.
Data availability Hong, A., Baker, L., Curiel, R., Duminy, J., Buswala, B., Guan, C., and Ravindranath, D.
(2022) Reconciling big data and thick data to advance the new urban science and
No data was used for the research described in the article. smart city governance. doi:10.1080/07352166.2021.2021085.
Hou, L., Jiao, R., 2020. Data-informed inverse design by product usage information: a
review, framework and outlook. J. Intell. Manuf. 31, 529–552.
References Jansen, J., Niemantsverdriet, K., Burghoorn, A., Lovei, P., Neutelings, I., Deckers, E.,
Nienhuijs, S., 2020. Design for co-responsibility: connecting patients, partners, and
professionals in bariatric lifestyle changes. DIS. https://doi.org/10.1145/
Ahmed, S. (2001) Understanding the use and reuse of experience in engineering design. doi
3357236.3395469.
:10.17863/CAM.96653.
Jin, J., Liu, Y., Ji, P., Liu, H., 2016. Understanding big consumer opinion data for market-
Ahmed, F., Fuge, M., 2018. Creative exploration using topic-based bisociative networks.
driven product design. Int. J. Prod. Res. 54 (10), 3019–3041.
Des. Sci. 4 (12)).
Kammerl, D., Novak, G., Hollauer, C., and Mörtl, M. (2016) Integrating usage data into
Alam, I., 2006. Removing the fuzziness from the fuzzy front-end of service innovations
the planning of Product-Service Systems. In IEEE International Conference on
through customer interactions. Ind. Mark. Manag. 35, 468–480.
Industrial Engineering and Engineering Management (Vol. 2016-December). IEEE
Bang, H., Selva, D., 2016. IFEED: interactive feature extraction for engineering design.
Computer Society.
ASME Des. Eng. Tech. Conf.
Bertoni, A., 2020. Data-driven design in concept development: systematic review and
missed opportunities. Int. Des. Conf. Virtual. https://doi.org/10.1017/dsd.2020.4.

12
B. Lee and S. Ahmed-Kristensen Computers in Industry 164 (2025) 104206

Khoshkangini, R., Mashhadi, P.S., Berck, P., Shahbandi, S.G., Pashami, S., Nowaczyk, S., design material. Comput. Hum. Interact. https://doi.org/10.1145/
Niklasson, T., 2020. Early prediction of quality issues in automotive modern 3491101.3516403.
industry. Information 11 (7), 354. Papalambros, P., 2015. Design science: why, what and how. Des. Sci. 1 (1)), 1–38.
Kim, M., Lim, C., Lee, C., Kim, K., Choi, S., Park, Y., 2016. Data-driven approach to new Porter, M., Heppelmann, J., 2014. How Smart, Connected Products Are Transforming
service concept design. Int. Conf. Explor. Serv. Sci. https://doi.org/10.1007/978-3- Competition. Harv. Bus. Rev. 23.
319-32689-4_37. Quiñones-Gómez, J., 2021. Creativity forward: a framework that integrates data analysis
Kim, H., Liu, Y., Wang, C., Wang, Y., 2017. Special issue: data-driven design. J. Mech. techniques to foster creativity within the creative process in user experience
Des., Trans. ASME 139 (11). contexts. Creativity. Studies 14 (1), 51–73.
King, R., Churchill, E., Tan, C., 2017. Designing with Data: Improving the User Shin, J., Kiritsis, D., Xirouchakis, P., 2015. Design modification supporting method based
Experience with A/B Testing, 1st ed. O’Reilly. on product usage data in closed-loop PLM. Int. J. Comput. Integr. Manuf. 28 (6),
King, I., Lyu, M., and Yang, H. (2013) Online Learning for Big Data Analytics. Tutorial 551–568.
presentation at IEEE Big Data. Song, B., Srinivasan, V., Luo, J., 2017. Patent stimuli search and its influence on ideation
Kirkpatrick, D., 1959. Techniques for evaluation training programs. J. Am. Soc. Train. outcomes. Des. Sci. 3 (e25), 1–25.
Dir. 13, 21–26. Speed, C., Lee, B., and Hands, D. (2019) The Little Book of Creating Value through
Kleinschmidt, E., Cooper, R., 1991. The impact of product innovativeness on Design in the IoT. Edited by Coulton, C. Lancaster University.
performance. J. Prod. Innov. Manag. 8 (4), 240–251. Stavrakos, S., Ahmed-Kristensen, S., 2016. Methods of 3D data applications to inform
Kushiro, N., Matsuda, S., Torikai, R., Takahara, K., 2014. A system design method based design decisions for physical comfort. Work 55 (2), 321–334.
on interaction between logic and data sets. IEEE Int. Conf. Data Min. Workshop. Stavrakos, S., Ahmed-Kristensen, S., Goldman, T., 2016. Using archetypes to create user
Kusiak, A., 2006. Data mining: manufacturing and service applications. Int. J. Prod. Res. panels for usability studies: streamlining focus groups and user studies. Appl. Ergon.
44 (18–19), 4175–4191. 56, 108–116.
Kusiak, A., 2009. Innovation: A data-driven approach. Int. J. Prod. Econ. 122 (1), Ströer, F., Sivasothy, P., Faißt, K., Apostolov, H., Eickhoff, T., Bechev, D., Bulun, G.,
440–448. Seewig, J., Eigner, M., Sauer, B., 2018. Combined development and test of product-
Lakoju, M., Ajienka, N., Khanesar, M.A., Burnap, P., Branson, D., 2021. Unsupervised service systems in early product development stages for customized, availability-
learning for product use activity recognition: an exploratory study of a ‘chatty oriented business models in the capital goods industry. Procedia CIRP 72, 714–719.
device. Sensors 21, 1–23. Tan, X., Chen, W., Cao, J., Ahmed-Kristensen, S., 2020. Identify critical data during
Landis, J., Koch, G., 1977. The measurement of observer agreement for categorical data, product customisation-a case study of orthoses fabrication. Int. Des. Conf. https://
33 (1), 159–174. doi.org/10.1017/dsd.2020.105.
Lee, B. (2022) Understanding New Product Development and Value Creation for the Internet Tao, F., Qi, Q., Liu, A., Kusiak, A., 2018. Data-driven smart manufacturing. J. Manuf.
of Things. Lancaster University, Lancaster doi:10.17635/lancaster/thesis/1646. Syst. 48, 157–169.
Lee, B., Ahmed-Kristensen, S., 2023. Four patterns of data-driven design activities in new Taroun, A., 2014. Towards a better modelling and assessment of construction risk:
product development. Proc. Des. Soc. 3, 1925–1934. Insights from a literature review. Int. J. Proj. Manag. 32 (1), 101–115.
Lee, B., Ahmed-Kristensen, S., 2024. D3IKIT: data-driven design innovation kit. Proc. Des. Tranfield, D., Denyer, D., Smart, P., 2003. Towards a methodology for developing
Soc. 4, 2109–2118. evidence-informed management knowledge by means of systematic review. Br. J.
Lee, B., Cooper, R., Hands, D., 2018. Are Traditional NPD processes relevant to IoT Manag. 14, 207–222.
product and service development activities: a critical examination. DRS Conf. Proc. Tuarob, S., Tucker, S., 2015. Quantifying product favorability and extracting notable
Limerick, Irel. product features using large scale social media data. J. Comput. Inf. Sci. Eng. 15 (3).
Lee, B., Cooper, R., Hands, D., Coulton, P., 2018. [Re]-imagining vision and values: Ulrich, K., Eppinger, S., 2015. Product Design and Development. (Vol, 6th edn. McGraw-
Design as a driver for value creation in the internet of things. IET Conf. Publ. 2018. Hill, Boston, MA.
Lee, B., Cooper, R., Hands, D., Coulton, P., 2019a. Design Drivers: A critical enabler to van den Heuvel, R., Driesse, E., Dekker, M., Calota, M., 2020. Understanding Routines
meditate value over the NPD process within Internet of Things. 4D Conference Around Medicine Intake through a Data-Enabled Design approach. UbiComp 2020.
Proceedings: Meanings of Design in the Next Era. Osaka, Japan. ACM. https://doi.org/10.1145/3410530.3414408.
Lee, B., Cooper, R., Hands, D., and Coulton, P. (2019b) Value creation for IoT: Challenges van den Heuvel, R., van Bussel, T., Lallemand, C., 2022. Habilyzer: a user-driven open-
and opportunities within the design and development process. IET Conference ended sensor kit for office workers. CHI- Comput. Hum. Interact.
Publications 2019. van Horn, D., Olewnik, A., Lewis, K., 2012. Design analytics: Capturing, understanding,
Lee, B., Cooper, R., Hands, D., Coulton, P., 2022a. Continuous cycles of data-enabled and meeting customer needs using big data. Proc. ASME Des. Eng. Tech. Conf. 7,
design: reimagining the IoT development process. AIEDAM. https://doi.org/ 863–875.
10.1017/S0890060421000299. van Kollenburg, J., and Bogers, S. (2019) Data-enabled design: a situated design approach
Lee, B., Hands, D., Cooper, R., Coulton, P., 2022b. Emergent NPD process and that uses data as creative material when designing for intelligent ecosystems. Eindhoven
development risks for IoT: an exploratory case study in agri-tech. Int. J. Bus. Syst. University of Technology.
Res. 16 (2), 183–198. Venkataraman, S., Song, B., Luo, J., Subburaj, K., Rajesh, M., Blessing, L., and Wood, K.
Lewis, K., and van Horn, D. (2013) Design Analytics in Consumer Product Design: A (2017) Investigating Effects of Stimuli on Ideation Outcomes. In International
Simulated Study. In ASME 2013 International Design Engineering Technical Conferences Conference on Engineering Design, Vancouver, Canada (Vol. 8). Vancouver, Canada.
and Computers and Information in Engineering Conference. Portland. Versteegden, D., van Himbeeck, M., Burghoorn, A., Lovei, P., Deckers, E., Jansen, J.,
Lim, D., 2016. Inverse DEA with frontier changes for new product target setting. Eur. J. Nienhuijs, S., 2022. The value of tracking data on the behavior of patients who have
Oper. Res. 254 (2), 510–516. undergone bariatric surgery: explorative study. JMIR 6 (5), 1–8.
Lim, C., Kim, M., Heo, J., Kim, K., 2018. Design of informatics-based services in Wang, G., Albayrak, A., Kortuem, G., der Cammen, T., 2021. A digital platform for
manufacturing industries: case studies using large vehicle-related databases. J. Intell. facilitating personalized dementia care in nursing homes: formative evaluation
Manuf. 29, 497–508. study. JMIR Form. Res. 5 (5).
Lovei, P., Niemantsverdriet, K., van Dijk, R., Burghoorn, A., Jansen, J., Neutelings, I., Wang, Z., Chen, C., Zheng, P., Li, X., Khoo, L., 2019. A novel data-driven graph-based
Deckers, E., Nienhuijs, S., 2020. Together in shape: a co-responsibility system to requirement elicitation framework in the smart product-service system context. Adv.
support bariatric lifestyle changes. DIS. https://doi.org/10.1145/3393914.3397094. Eng. Inform. 42, 1–11.
Ma, J., Kwak, M., Kim, H., 2014. Demand trend mining for predictive life cycle design. Wang, Z., Chen, C., Zheng, P., Li, X., Khoo, L., 2021. A graph-based context-aware
J. Clean. Prod. 68, 189–199. requirement elicitation approach in smart product-service systems. Int. J. Prod. Res.
Machchhar, R., Toller, C., Bertoni, A., Bertoni, M., 2022. Data-driven value creation in 59 (2), 635–651.
Smart Product-Service System design: state-of-the-art and research directions. Wang, M., Sha, Z., Huang, Y., Fu, Y., Chen, W., 2016. Forecasting technological impacts
Comput. Ind. 137, 1–21. on customers’ co-consideration behaviors: a data-driven network analysis approach.
Maleki, E., Belkadi, F., Boli, N., van der Zwaag, B., Alexopoylos, K., Koukas, S., Marin- ASME Des. Eng. Tech. Conf.
Perianu, M., Bernard, A., Mourtzis, D., 2018. Ontology-based framework enabling Watanabe, K., Okuma, T., Takenaka, T., 2020. Evolutionary design framework for Smart
smart product-service systems: application of sensing systems for machine health PSS: service engineering approach. Adv. Eng. Inform. 45, 1–11.
monitoring. Internet Things J. 5 (6), 4496–4505. Wiering, R., Maiden, N., Mead, N., Rolland, C., 2005. Requirements engineering paper
Mata, M., Ahmed-Kristensen, S., Brockhoff, P., Yanagisawa, H., 2017. Investigating the classification and evaluation criteria: a proposal and a discussion. Requir. Eng. 11,
influence of product perception and geometric features. Res. Eng. Des. 28 (3), 102–107.
357–379. Wodehouse, A., Vasantha, G., Corney, J., Jagadeesan, A., MacLachlan, R., 2018.
Montecchi, T., Becattini, N., 2021. A modelling framework for data-driven design for Realising the affective potential of patents: a new model of database interpretation
sustainable behaviour in human-machine interactions. Int. Conf. Eng. Des. Gothenbg., for user-centred design. J. Eng. Des. 29 (8–9), 484–511.
Swed. https://doi.org/10.1017/pds.2021.16. Xiong, Y., Duong, P., Wang, D., Park, S., Ge, Q., Raghavan, N., Rosen, D., 2019. Data-
Mostafa, S., Chileshe, N., Abdelhamid, T., 2016. Lean and agile integration within offsite driven design space exploration and exploitation for design for additive
construction using discrete event simulation A systematic literature review. Lean manufacturing. J. Mech. Des., Trans. ASME 141 (11).
Agil. Integr. 16 (4), 483–525. Yang, B., Liu, Y., Liang, Y., Tang, M., 2019. Exploiting user experience from online
Noortman, R., Lovei, P., Funk, M., Deckers, E., Wensveen, S., Eggen, B., 2022. Breaking customer reviews in product design. Int. J. Inf. Manag. 46, 173–186.
up data-enabled design: expanding and scaling up for the clinical context. AIEDAM Yoon, J., Seo, W., Coh, B., Song, I., Lee, J., 2017. Identifying product opportunities using
36 (19)), 1–13. collaborative filtering-based patent analysis. Comput. Ind. Eng. 107, 376–387.
Okoh, C., Roy, R., Mehnen, J., 2017. Predictive maintenance modelling for through-life Zambetti, M., Pinto, R., Pezzotta, G., 2019. Data lifecycle and technology-based
engineering services. Procedia CIRP 59, 196–201. opportunities in new product service system offering towards a multidimensional
Ortega, A., van Kollenburg, J., Shen, Y., Murray-Rust, D., Jimenez Garcia, J., Meijer, W., framework. In: In Procedia CIRP, 83. Elsevier B.V.
Kumar Chaudhary, P., Bourgeois, J., Nedić, D., 2022. SIG on data as human-centered

13
B. Lee and S. Ahmed-Kristensen Computers in Industry 164 (2025) 104206

Zhang, L., Chu, X., Chen, H., Yan, B., 2019. A data-driven approach for the optimisation Prof. Saeema Ahmed-Kristensen is an Associate Pro-vice
of product specifications. Int. J. Prod. Res. 57 (3), 703–721. Chancellor of Research and Impact (Faculty of Environne­
Zhang, Z., Liu, L., Wei, W., Tao, F., Li, T., Liu, A., 2017. A systematic function ment, Science & Economy), Chair of Engineering Design and
recommendation process for data-driven product and service design. J. Mech. Des., Innovation and Director of UKRI Research Centre DIGITLab,
Trans. ASME 139 (11). bringing experience from Royal College of Art (Head of Design
Zheng, P., Chen, C., Shang, S., 2019. Towards an automatic engineering change Products) and Imperial College London, Cambridge Engineer­
management in smart product-service systems – a DSM-based learning approach. ing and Denmark. Saeema’s research focuses upon improve­
Adv. Eng. Inform. 39, 203–213. ments of both products (including product service systems) and
Zheng, P., Liu, Y., Tao, F., Wang, Z., Chen, C., 2019. Smart product-service systems processes (creative, product development and innovation)
solution design via hybrid crowd sensing approach. IEEE Access 7, 128463–128473. through developing a scientific understanding of the processes,
Zheng, P., Wang, Z., Chen, C., Khoo, L., 2019. A survey of smart product-service systems: contributing to Design cognition, knowledge and data struc­
Key aspects, challenges and future perspectives. Adv. Eng. Inform. 42, 1–19. turing, Data driven design, Quantifying User experiences. She
Zheng, P., Xu, X., Chen, C.-H., 2020. A data-driven cyber-physical approach for works closely with a range of industries; complex product,
personalised smart, connected product co-development in a cloud-based aerospace; oil; consumer; medical, and adopts a multidisciplinary approach including
environment. J. Intell. Manuf. 31, 3–18. psychology, computing and design. She has authored over 100 Publications across these
domains.

Dr. Boyeun Lee is an Assistant Professor in Design and Inno­

vation at the University of Exeter Business School. Before her
current role, she held the position of the Post-Doctoral
Research Fellow for Digitlab at the Univesrity of Exeter, and
the Post-Doctoral Research Associate for the Beyond Imagina­
tion at Lancaster University. She completeld her PhD in Design
at Lancaster University in 2022 as a part of EPSRC’s PETRAS
IoT Hub, with the support of two prestigious scholarships: one
from Lancaster University and EPSRC, as well as the RADMA
Doctoral Student Award. Her research primarily centres
around how design contributes to digital innovation, particu­
larly utilising big data and AI within the context of product
service development.

14

View publication stats