Digital transformation in construction supply chains requires the coordinated development of dynamic capabilities. While prior research identifies digital sensing, seizing and transforming as critical enablers of strategic renewal, limited attention has been paid to how these capabilities are perceived and aligned across hierarchical levels. This study aims to examine how dynamic capabilities for digital transformation are interpreted, prioritised and linked by managers, digital change agents and operational staff in the construction industry.
Digital dynamic capabilities were categorised into sensing, seizing and transforming dimensions. Perceptual data were collected from 74 respondents across three hierarchical groups within a construction firm operating in German and Austrian markets. The Decision-Making Trial and Evaluation Laboratory (DEMATEL) method was applied to model perceived interdependencies and cause–effect relationships between capabilities. Statistical tests were conducted to analyse differences in perceived capability manifestations within and across hierarchical levels.
The analysis shows that substantial perceptual divergence exists across hierarchical levels. Senior managers frame digital capabilities primarily as strategic enablers of competitiveness and long-term positioning, digital change agents adopt an intermediary implementation-oriented perspective and operational staff evaluate capabilities through their immediate impact on workflows and routines.
This study advances dynamic capability research by empirically modelling non-linear interdependencies between digital capabilities in a construction context. It extends micro foundational perspectives by demonstrating how hierarchical position shapes cognitive construction and prioritisation of dynamic capabilities. Methodologically, the study introduces DEMATEL as a systematic approach to modelling perceived causal structures in construction management research.
1. Introduction
The construction industry has been frequently criticised in recent years for its lack of adoption of new technologies for digital transformation (Adeniyi et al., 2024; Gharaibeh et al., 2024). Scholars link the under-utilisation of technology in construction to the scattered and fragmented project delivery processes, in particular between contractors and subcontractors, in combination with organisational, technological and managerial challenges along the supply chain (Gadde and Dubois, 2010; Yevu et al., 2021). Consequently, the construction industry seems to rely on proven and established business models that are built on their past successes rather than risking substantive changes through digital transformation (Mastio et al., 2021; Sull, 1999).
According to McKinsey, however, digital transformation in the construction industry shows huge potential (Barbosa et al., 2017). Yevu et al. (2021) point out that “new and emerging technologies such as blockchain, smart contracts and Internet of Things (IoT) are being explored for potential applications in construction supply chain and procurement processes (p.1). Studies show that the use of novel technologies would increase productivity in existing construction processes, for example, in the built environment projects (Papadonikolaki et al., 2022), prefabricated construction (Kwok and Chang, 2024), or through blockchain (Yang et al., 2020) and digital project-driven supply chains (Bhattacharya and Chatterjee, 2022).
Existing research points out that digital transformation in the construction industry depends on having or building the relevant organisational digital capabilities across the organisation (Ekanayake et al., 2021). Building on Warner and Wäger (2019), we understand digital capabilities as an extension of Teece et al. (1997) dynamic capabilities theory to digital transformation contexts, comprising digital sensing, digital seizing and digital transforming capabilities that enable organisations to identify, mobilise and reconfigure resources for digital renewal. As such, the task of construction management is therefore not only to engage in digital capabilities building to drive digital transformation, but also to formulate a digital strategy that aligns digital capabilities between organisational hierarchies.
Current research, however, points out that managers often fail to formulate the digital strategies for digital transformation (Chanias et al., 2019). More specifically, management in charge of transforming organisations are confronted with both technological and operational challenges stemming from the introduction of novel digital technologies due to a lack of digital capabilities or a lack of alignment between digital capabilities across organisational units and hierarchies (Bresciani et al., 2021; Fernandez-Vidal et al., 2022; Kunisch et al., 2022; Singh and Hess, 2017).
In this paper, we argue that existing literature has paid less attention to how digital capabilities are perceived, prioritised and causally linked across hierarchical levels. As a result, we know relatively little about digital capabilities across hierarchical organisational levels and whether senior managers, digital change agents and operational staff understand digital sensing, seizing and transforming capabilities in aligned or divergent ways. In particular, we claim (a) that little attention has been paid to what specific digital capabilities are required in the construction industry and (b) whether digital capabilities are aligned in the organisation hierarchy, that is, how digital capabilities are used and perceived between managers, operational staff and IT employees.
In response, this study contributes to the field’s empirical knowledge base by examining the managers, operational staff and IT-employees (digital agents) perceptions of the relevance and interdependencies of digital capabilities. The aim of this study is to examine how digital dynamic capabilities for construction supply chain transformation are perceived, prioritised and causally linked across hierarchical levels. To address this aim, we raise the following research questions:
Are capabilities for digital transformation in construction supply chains perceived as relevant?
Which interdependencies between these capabilities are perceived?
How do perceptions differ across managerial hierarchies?
Which manifestations of dynamic capabilities for digital transformations do managers perceive in their respective industries?
To address these research questions, we adopt a three-stage research design. Firstly, we identify and structure digital dynamic capabilities for construction supply chains through a systematic literature-based content analysis. Secondly, we collect perceptual data across three hierarchical groups – senior managers, digital change agents and operational staff-within a construction firm operating in the German and Austrian markets. Thirdly, we apply the Decision-Making Trial and Evaluation Laboratory (DEMATEL) method to analyse perceived interdependencies and causal relationships between capabilities across hierarchical levels, not only which capabilities are perceived as relevant, but also how they are understood to interact within a complex system of sensing, seizing and transforming.
The contribution of this study is fourfold. Firstly, it advances dynamic capability research by examining digital sensing, seizing and transforming as a perceived interdependent system rather than as sequential constructs, thereby empirically supporting more non-linear interpretations of capability orchestration. Secondly, it extends research on digital transformation in construction by identifying which specific digital capabilities are considered strategically central and how their causal importance differs across hierarchical levels. Thirdly, the study contributes to the micro foundations of dynamic capabilities by demonstrating that hierarchical position shapes how capabilities are cognitively constructed, prioritised and evaluated, thereby influencing the coherence of transformation efforts. Fourthly, methodologically, this study applies DEMATEL as a systematic approach to modelling perceptual cause–effect structures in construction management research, thereby offering a replicable analytical tool for studying complex transformation systems in project-based environments.
The remainder of this paper is structured as follows. Section 2 develops the conceptual foundation by reviewing digital transformation in construction supply chains and positioning dynamic capabilities as a lens for analysing strategic transition. Section 3 outlines the research design, sampling strategy and analytical procedures, including the DEMATEL methodology. Section 4 presents the empirical results, first analysing perceived capability interdependencies and subsequently examining differences in perceived capability manifestations across hierarchical groups. Section 5 discusses the findings in light of dynamic capability theory and strategic management research, before Section 6 concludes with implications, limitations and avenues for future research.
2. Digital transformation in construction supply chains
The construction industry plays an important role in global economic development, contributing between 6% and 10% of gross domestic product across major economies and generating annual revenues exceeding US$10tn (Chang and Antwi-Afari, 2025; El-Sayegh et al., 2020). Despite its economic relevance, the sector continues to be characterised by low productivity growth, fragmented supply chains, short-term project logic and comparatively limited investment in research and development (Bygballe and Ingemansson, 2014; Dixit et al., 2017; Oesterreich and Teuteberg, 2016).
Construction supply chains can be considered as inherently complex as projects involve multiple tiers of contractors and subcontractors operating under temporary contractual arrangements (Pham et al., 2023; Taroun, 2014). These arrangements not only result in decentralised coordination structures and high interdependence, but the associated structural fragmentation also constrains organisational learning and slows innovation diffusion (Ahmadisheykhsarmast and Sonmez, 2020; Gadde and Dubois, 2010). At the same time, the industry is increasingly exposed to digital technologies associated with Industry 4.0, including building information modelling (BIM), IoT-enabled sensors, digital twins, robotics, blockchain and artificial intelligence (Bolpagni et al., 2022; Bosch-Sijtsema et al., 2021; Ghosh et al., 2024). These technologies promise improvements in productivity, transparency, safety and coordination. However, despite their potential, adoption remains uneven and frequently superficial (Bolpagni et al., 2024; Gharaibeh et al., 2024). As a consequence, we argue that digital transformation in construction therefore represents a strategic transition that requires coordinated (dynamic) capability development across organisational levels and supply chain actors.
In response, this study adopts the theory of dynamic capabilities to analyse such strategic transitions. Dynamic capability theory allows us for explaining how firms renew, coordinate and reconfigure resource bases when established routines and supply chain arrangements are challenged by technological change (Dobrovnik et al., 2025). In this study, we follow Warner and Wäger (2019), who extend Teece et al. (1997) dynamic capabilities theory to digital transformation by distinguishing digital dynamic capabilities between (a) digital sensing (e.g. digital scouting and scenario planning), (b) digital seizing (e.g. rapid prototyping, portfolio balancing and strategic agility) and (c) digital transforming (e.g. redesigning internal structures and improving digital maturity). Accordingly, we define digital supply chain transformation as the strategic reconfiguration of supply chain processes, relationships and decision-making through digital technologies and data-enabled coordination to improve integration, responsiveness and value creation across supply chain actors.
While much of the empirical literature implicitly assumes a sequential process from sensing to seizing to transforming, more recent research suggests that dynamic capabilities operate in a non-linear and interactive manner and interdependencies and feedback loops frequently shape transformation pathways (Castka et al., 2024; Leemann et al., 2021). In the construction industry, this complexity may be amplified due to decentralised governance, project-based structures and strong professional subcultures (Opoku et al., 2021; Sumanarathna et al., 2023). Research shows that capability development does not occur in isolation but is embedded within layered organisational hierarchies and supply chain relationships (Mason, 2017; McNamara et al., 2024).
However, existing research largely treats dynamic capabilities as organisational-level constructs with the implicit assumption that these capabilities are collectively understood and coherently enacted across the firm (Ekanayake et al., 2021; Herold et al., 2024). Furthermore, our argument is that digital transformation initiatives often fail not because technologies are unavailable, but because organisational actors interpret their relevance, urgency and interdependencies differently (Chanias et al., 2019; Dobrovnik et al., 2025; Fernandez-Vidal et al., 2022).
Scholas found that expectations about technological futures not only shape resource allocation, legitimacy and experimentation (Boone et al., 2025; Van Lente et al., 2013), but in particular in industries that lag in technological maturity such as construction, these expectation dynamics may vary substantially across managerial hierarchies (Bosch-Sijtsema et al., 2021). More specifically, senior managers may conceptualise digital capabilities at a strategic and abstract level, with digital change agents focusing on implementation feasibility and project coordination, while operational staff may evaluate digital initiatives through their immediate impact on workflows and routines. Thus, if dynamic capabilities are perceived differently across these groups, capability “orchestration” becomes fragmented, that is, instead of functioning as an integrated system, sensing, seizing and transforming activities may be cognitively decoupled (Ghaffarianhoseini et al., 2017).
So far, existing literature on dynamic capabilities has paid limited attention to these intra-organisational perception differences. In other words, empirical research rarely investigates whether different hierarchical groups attribute similar relevance, causal importance and maturity levels to dynamic capabilities. This gap is particularly relevant in construction, where transformation depends on coordinated action across heterogeneous actors. In response, this study examines the perceived digital transformation across managerial hierarchies.
3. Methodology
The overall research approach was designed to connect the theoretical framing, construct development and empirical analysis. Dynamic capability theory provides the conceptual lens for understanding digital supply chain transformation as capability renewal, while Warner and Wäger (2019) digital sensing, seizing and transforming capabilities provide the construct structure for operationalising this lens. These constructs are translated into survey-based capability items and analysed through DEMATEL to examine perceived interdependencies, complemented by statistical tests to assess differences in perceived capability manifestations across hierarchical groups
3.1 Research approach
To ensure a high degree of specificity in the results, the study was contextualised across four dimensions:
our research focused on the competitive environment in German and Austrian markets;
it focused on the construction industry;
predominantly analysed digital transformation leveraging disruptive digital technologies such as blockchain, digital twins and artificial intelligence; and
was discipline-specific by analysing digitalisation within the domains of supply chain, logistics and operations management.
Within this contextual framework, the study uses a three-stage research design:
In the initial stage, firm-level capabilities for digital transformation were identified through a semi-structured, concept-centered literature analysis. In accordance with previous research, these capabilities were then classified into three main categories, each being composed of three subcategories.
In the second stage, a sampling strategy and data collection procedure were defined to systematically gather perceptual data across three defined management levels. This information served as the basis for conducting two independent analyses in steps.
On the one hand, the DEMATEL method was applied to assess how these competencies and their interrelations are perceived by employees across managerial hierarchies (A). DEMTAL can be identified as an effective method for gathering group knowledge to derive structural model (Lee et al., 2024; Wu and Lee, 2007) and to identify and visualise cause–effect chain components of a complex system (Si et al., 2018). In addition, building on the same sample, for each competency and sub-competency it was analysed which actual levels of competency were attributed by the respondents to their own organisation as well as to direct competitors within their industry. This information was systematically evaluated using parametric and non-parametric statistical tests (B). The sampling strategy and data collection procedure as well as the analysis are further outlined below.
3.2 Sampling strategy and data collection procedure
Our empirical analysis is based on an embedded single case design (Yin, 2018). We studied a partially vertically-integrated firm in the construction industry with a focus on the central European market. It acts as a general contractor as well as a total contractor and therefore covers a substantial part of the value chain. Focusing on a characteristic company allows for a more nuanced and targeted examination of different management levels within this company. Hence, one can gain a deeper understanding of the studied phenomenon (Gustafsson, 2017). In this regard, however, it must be critically acclaimed that the implementation as an embedded single case design also limits the generalisability of the results.
Within the case company, we targeted experts of three distint subgroups, namely, senior managers (e.g. board members and senior executives), digital change agents (e.g. innovation managers and project managers) responsible for implementing digital transformation projects and operational staff (e.g. foremen and technicians). For this purpose, we can broadly distinguish between a probability sampling and non-probability sampling (NPS) approach (Taherdoost, 2016). Because our research requires specialised expertise and depends on respondents possessing significant perspectives on pertinent ideas (Campbell et al., 2020), NPS consindered to be adequate. Therefore, a combination of purposive and chain-referral sampling was used to identify suitable experts. Purposive sampling ensured that participants were selected according to their hierarchical position and involvement in strategic, implementation-oriented or operational aspects of digital transformation, while chain-referral sampling helped identify additional knowledgeable respondents within these groups. Previous research suggests that 4–15 experts are appropriate for DEMATEL if their selection is criteria-based and they have relevant domain knowledge (Jafari-Sadeghi et al., 2022; Quayson et al., 2023; Raj and Sah, 2019). We defined a related quota of 24 responses for each group to also allow for subsequent statistical analyses. The use of predefined quotas across senior managers, digital change agents and operational staff further supported comparison across hierarchical levels.
A digital survey was developed to systematically collect the data. The survey comprised three sections. In the initial segment, each respondent was required to assess the influence of each studied capability on every other capability, using a five-point Likert scale ranging from 0 (no influence) to 4 (very high). In the second part, each respondent was tasked with evaluating the perceived capability manifestations inside their own organisation and for their direct competitors in the construction industry. In the third segment, socio-demographic data was collected. Pre-tests – which were conducted online, through interviews and using pen-and-paper methods – confirmed that the initial phase of the study presented a cognitively demanding task. This increases the risk of experiencing insufficient effort responding and survey fatigue (DeSimone et al., 2018; Huang et al., 2012; Schoenherr et al., 2015). Based on interviews with pre-test participants, we identified limited background information and excessive room for interpretation with respect to the studied capabilities as their main areas of concern. This was particularly the case for lower-level employees. We addressed these issues by systematically framing the research project using verbal and written briefings. Furthermore, we systematically included generic examples for each capability as well as explanatory pictograms and visual aids throughout the survey. Together with representatives of the company – who did not participate in the survey themselves – potential experts for each group were identified based on their corporate roles, internal functional codes, leadership responsibilities and affiliation with the company. Data was gathered between December 2023 and February 2024. The data collection procedure was suspended once the quota was reached for each group. A total of 98 individuals were invited to participate in the study, and 74 (75.5% response rate) fully analysable data sets were generated. Table 1 provides an overview of the respondents.
Profile of the respondents
| Total sample | Managers | Digital change agents | Operational staff | ||
|---|---|---|---|---|---|
| (n = 74) | (n = 24) | (n = 26) | (n = 24) | ||
| Respondent characteristics | Category | Count | |||
| Years of expertise | ≥03 and <10 | 22 | 2 | 12 | 8 |
| ≥10 and <20 | 21 | 4 | 8 | 9 | |
| ≥20 and <30 | 16 | 7 | 4 | 5 | |
| ≥30 and <40 | 10 | 6 | 2 | 2 | |
| ≥40 and <50 | 5 | 5 | 0 | 0 | |
| Formal education | Professional training | 11 | 1 | 1 | 9 |
| Highschool diploma | 19 | 5 | 4 | 10 | |
| University | 44 | 18 | 21 | 5 | |
| Function/role | Commercial expert | 8 | 2 | 6 | |
| Technical expert | 7 | 1 | 6 | ||
| Foreman | 5 | 5 | |||
| Site supervisor | 3 | 3 | |||
| Digital transformation expert | 9 | 9 | |||
| Product owner | 1 | 1 | |||
| Innovation manager | 3 | 3 | |||
| Project manager | 12 | 8 | 4 | ||
| Senior project manager | 2 | 2 | |||
| Business manager | 1 | 1 | |||
| Head of department | 13 | 13 | |||
| Managing director | 3 | 3 | |||
| Senior executive | 7 | 7 |
4. Data analysis and results
DEMATEL applications in construction research are commonly presented as a sequence of steps, including the construction of an average direct-relation matrix, normalisation, derivation of the total relation matrix, calculation of prominence and net effects, threshold selection and visualisation through cause–effect diagrams (Lee et al., 2024). While these steps provide a useful orientation for readers less familiar with DEMATEL, the analysis follows the specific procedure outlined by Agi and Jha (2022) and Lee et al. (2013), which is consistent with our multi-group research design and the comparison of perceived capability interdependencies across hierarchical levels.
In an initial step, the collected perceptual information was examined using DEMATEL. The study followed a stepwise approach described by Agi and Jha (2022) and applied a normalisation technique described by Lee et al. (2013). The steps depicted in Figure 1 were applied separately for each subsample (managers, digital change agents and operational staff).
The flowchart begins with Step 1: define sample and subsamples. Step 2 gathers expert opinion as matrices A subscript k. Step 3 calculates average matrix A. Step 4 calculates the normalized initial direct-relation matrix B. Step 5 derives the total relation matrix T. Step 6 calculates the sums of rows R subscript i and columns C subscript j of matrix T. Step 7 sets the threshold value alpha. Step 8 builds the cause-and-effect relationship diagram. Step 9 asks whether the diagram is acceptable. If the answer is no, the flow returns to Step 7. If the answer is yes, the flow continues to Step 10, which derives final cause-and-effect relationships.Step-wise DEMATEL analysis adapted from Agi and Jha (2022)
The flowchart begins with Step 1: define sample and subsamples. Step 2 gathers expert opinion as matrices A subscript k. Step 3 calculates average matrix A. Step 4 calculates the normalized initial direct-relation matrix B. Step 5 derives the total relation matrix T. Step 6 calculates the sums of rows R subscript i and columns C subscript j of matrix T. Step 7 sets the threshold value alpha. Step 8 builds the cause-and-effect relationship diagram. Step 9 asks whether the diagram is acceptable. If the answer is no, the flow returns to Step 7. If the answer is yes, the flow continues to Step 10, which derives final cause-and-effect relationships.Step-wise DEMATEL analysis adapted from Agi and Jha (2022)
Step 7 is particularly important. It facilitates the identification of the most relevant relations in the total relation matrix by defining a threshold value . Selecting a rather small value would result in the exclusion of very few relationships and therefore generate complex diagrams. On the other hand, applying a high value for leads to the omission of numerous potentially interesting interconnections. Considering the moderate number of variables in this study, we opted to calculate sample-specific threshold values that represent the mean of all relationships included in the total relation matrix (T).
4.1 Analysis of perceived relationships between capabilities
By applying DEMATEL as described above, we computed the total relation matrix (). Each cell () in Table 2 represents the effect that a capability in a row () has a receiving capability in a column (). Bold values indicate effects above the threshold value . Shaded areas highlight effects within the same capability category. Despite strong differences across management levels, patterns can be observed. For example, C2 (digital scenario planning), C3 (digital mindset crafting) and improving digital maturity (C9) are affected by most other capabilities in each sample. By contrast, capability C4 (rapid prototyping) does not appear to be significantly influenced throughout, except for C5 (balancing digital portfolios) in the sample of operational staff.
Total relation matrix T for each sample and individual capabilities
| C1 | C2 | C3 | C4 | C5 | C6 | C7 | C8 | C9 | Ri | ||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Respondent group | Capability | Digital scouting | Digital scenario planning | Digital mindset crafting | Rapid prototyping | Balancing digital portfolios | Strategic agility | Navigating innovation ecosystems | Redesigning internal structures | Improving digital maturity | Sum |
| Managers (α = 0.76) | |||||||||||
| C1 | Digital scouting | 0.6881 | 0.9147 | 0.8733 | 0.6772 | 0.6893 | 0.8758 | 0.7438 | 0.8384 | 0.8538 | 7.1546 |
| C2 | Digital scenario planning | 0.8088 | 0.8012 | 0.9006 | 0.7122 | 0.7476 | 0.9156 | 0.7307 | 0.8823 | 0.8605 | 7.3595 |
| C3 | Digital mindset crafting | 0.8572 | 0.9659 | 0.8142 | 0.7535 | 0.7518 | 0.9474 | 0.7889 | 0.9225 | 0.9383 | 7.7396 |
| C4 | Rapid prototyping | 0.6633 | 0.7466 | 0.7396 | 0.5156 | 0.5927 | 0.7589 | 0.6398 | 0.7572 | 0.7686 | 6.1823 |
| C5 | Balancing digital portfolios | 0.7171 | 0.8427 | 0.7770 | 0.6177 | 0.5736 | 0.8316 | 0.6588 | 0.8108 | 0.7888 | 6.6180 |
| C6 | Strategic agility | 0.7462 | 0.8558 | 0.8026 | 0.6637 | 0.6906 | 0.7350 | 0.6800 | 0.8565 | 0.8080 | 6.8383 |
| C7 | Navigating innovation ecosystems | 0.7453 | 0.8143 | 0.7854 | 0.6085 | 0.6275 | 0.7750 | 0.5706 | 0.7457 | 0.7617 | 6.4340 |
| C8 | Redesigning internal structures | 0.6097 | 0.6968 | 0.6822 | 0.5295 | 0.5829 | 0.7175 | 0.5667 | 0.5998 | 0.6914 | 5.6765 |
| C9 | Improving digital maturity | 0.8063 | 0.9024 | 0.9027 | 0.7008 | 0.7233 | 0.8859 | 0.7485 | 0.8857 | 0.7648 | 7.3204 |
| Ci | Sum | 6.6420 | 7.5405 | 7.2776 | 5.7788 | 5.9792 | 7.4428 | 6.1278 | 7.2989 | 7.2357 | |
| Digital change agents (α = 1.11) | |||||||||||
| C1 | Digital scouting | 0.9298 | 1.2410 | 1.1815 | 0.9864 | 1.1145 | 1.1842 | 1.0263 | 1.1380 | 1.2036 | 10.0053 |
| C2 | Digital scenario planning | 1.0279 | 1.1212 | 1.1940 | 0.9890 | 1.1441 | 1.2065 | 1.0261 | 1.1765 | 1.2236 | 10.1087 |
| C3 | Digital mindset crafting | 1.0920 | 1.2959 | 1.1465 | 1.0408 | 1.1929 | 1.2576 | 1.0895 | 1.2349 | 1.3081 | 10.6582 |
| C4 | Rapid prototyping | 0.8864 | 1.0771 | 1.0415 | 0.7936 | 0.9861 | 1.0572 | 0.9071 | 1.0214 | 1.0712 | 8.8416 |
| C5 | Balancing digital portfolios | 0.9950 | 1.1964 | 1.1333 | 0.9558 | 0.9930 | 1.1640 | 0.9984 | 1.1320 | 1.1851 | 9.7530 |
| C6 | Strategic agility | 0.9871 | 1.1770 | 1.1341 | 0.9848 | 1.1006 | 1.0521 | 0.9836 | 1.1440 | 1.1783 | 9.7416 |
| C7 | Navigating innovation ecosystems | 1.0853 | 1.2663 | 1.2218 | 1.0270 | 1.1449 | 1.2116 | 0.9576 | 1.1693 | 1.2534 | 10.3372 |
| C8 | Redesigning internal structures | 1.0269 | 1.2286 | 1.2073 | 1.0002 | 1.1393 | 1.2127 | 1.0252 | 1.0655 | 1.2367 | 10.1426 |
| C9 | Improving digital maturity | 1.0884 | 1.2868 | 1.2848 | 1.0307 | 1.1728 | 1.2512 | 1.0754 | 1.2173 | 1.1679 | 10.5753 |
| Ci | Sum | 9.1189 | 10.8903 | 10.5447 | 8.8084 | 9.9881 | 10.5970 | 9.0892 | 10.2990 | 10.8279 | |
| Operational staff (α = 0.81) | |||||||||||
| C1 | Digital scouting | 0.6233 | 0.8209 | 0.8379 | 0.6691 | 0.7159 | 0.7636 | 0.7889 | 0.6736 | 0.7878 | 6.6809 |
| C2 | Digital scenario planning | 0.7830 | 0.7644 | 0.9108 | 0.7041 | 0.7856 | 0.7989 | 0.8429 | 0.7309 | 0.8395 | 7.1601 |
| C3 | Digital mindset crafting | 0.7139 | 0.8074 | 0.7358 | 0.6586 | 0.7121 | 0.7513 | 0.7682 | 0.6807 | 0.7839 | 6.6119 |
| C4 | Rapid prototyping | 0.7737 | 0.8975 | 0.9411 | 0.6527 | 0.8219 | 0.8336 | 0.8867 | 0.7390 | 0.8775 | 7.4237 |
| C5 | Balancing digital portfolios | 0.8460 | 0.9754 | 1.0065 | 0.8177 | 0.7658 | 0.9162 | 0.9221 | 0.8196 | 0.9658 | 8.0350 |
| C6 | Strategic agility | 0.8195 | 0.9646 | 1.0013 | 0.7910 | 0.8530 | 0.7869 | 0.9205 | 0.8293 | 0.9344 | 7.9005 |
| C7 | Navigating innovation ecosystems | 0.7583 | 0.8624 | 0.8943 | 0.7123 | 0.7648 | 0.8027 | 0.7256 | 0.7203 | 0.8302 | 7.0709 |
| C8 | Redesigning internal structures | 0.7607 | 0.8592 | 0.9096 | 0.7213 | 0.7876 | 0.8326 | 0.8302 | 0.6508 | 0.8653 | 7.2172 |
| C9 | Improving digital maturity | 0.7508 | 0.8690 | 0.9164 | 0.7117 | 0.7738 | 0.8139 | 0.8332 | 0.7500 | 0.7460 | 7.1648 |
| Ci | Sum | 6.8292 | 7.8207 | 8.1538 | 6.4385 | 6.9805 | 7.2997 | 7.5182 | 6.5942 | 7.6303 | |
Based on the total relation matrix (), we first analysed the prominence and net effect of each individual capability for each sample. The prominence of a capability denotes the relevance of importance of a capability. It is calculated by aggregating the total effect of a capability () on each receiving capability () as well as the influence it receives from any other capability (cf. Equation (1)). A high prominence signifies the overall importance of a capability. Hence, it can be stipulated that high-prominence factors deserve increased attention as they allow for decision makers to prioritise capabilities likely to have the highest overall impact. The net effect refers to whether a studied capability is a cause or an effect. It is computed by subtracting the aggregate received influence of a capability from the exerted influence on other capabilities (cf. Equation (2)).
A positive net effect shows that the capability predominantly functions as a catalyst or cause within the system, impacting other factors. By contrast, a negative net effect indicates that the capability is a consequence, shaped by other influences. It allows decision makers to differentiate between cause capabilities that impact digital transformation and effect capabilities. The combination of prominence and net effect helps to identify critical leverage points within the studied system.
Table 3 represents the prominence and net effects for each capability studied. For managers, C3 (digital mindset crafting) has the highest prominence and C4 (rapid prototyping) has the least prominent. For digital change agents, C9 (improving digital maturity) is the most prominent variable. They also identify C4 (rapid prototyping) as the least prominent capability. Operational staff denotes the highest overall impact to C6 (strategic agility) and the least to C1 (digital scouting). Concerning net effects, managers identify every capability except C2 (digital scenario planning), C6 (strategic agility) and C8 (redesigning internal structures) as causes. This differs from the data for digital change agents who also perceive C5 (balancing digital portfolios) and C9 (improving digital maturity) as effects rather than causes. Even more mismatches can be seen for operational staff, who only identify C4 (rapid prototyping), C5 (balancing digital portfolios), C6 (strategic agility) and C8 (redesigning internal structures) as causes.
Prominence and net effect of capabilities
| Managers (MAN) | Digital change agents (DCA) | Operational staff (OPS) | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Capability | Ri + Ci | Ri-Ci | Identity | Rank | Ri + Ci | Ri-Ci | Identity | Rank | Ri + Ci | Ri-Ci | Identity | Rank | |
| C1 | Digital scouting | 13.797 | 0.513 | Cause | 5 | 19.124 | 0.886 | Cause | 8 | 13.510 | −0.148 | Effect | 9 |
| C2 | Digital scenario planning | 14.900 | −0.181 | Effect | 2 | 20.999 | −0.782 | Effect | 3 | 14.981 | −0.661 | Effect | 3 |
| C3 | Digital mindset crafting | 15.017 | 0.462 | Cause | 1 | 21.203 | 0.113 | Cause | 2 | 14.766 | −1.542 | Effect | 5 |
| C4 | Rapid prototyping | 11.961 | 0.404 | Cause | 9 | 17.650 | 0.033 | Cause | 9 | 13.862 | 0.985 | Cause | 7 |
| C5 | Balancing digital portfolios | 12.597 | 0.639 | Cause | 7 | 19.741 | −0.235 | Effect | 6 | 15.016 | 1.055 | Cause | 2 |
| C6 | Strategic agility | 14.281 | −0.604 | Effect | 4 | 20.339 | −0.855 | Effect | 5 | 15.200 | 0.601 | Cause | 1 |
| C7 | Navigating innovation ecosystems | 12.562 | 0.306 | Cause | 8 | 19.426 | 1.248 | Cause | 7 | 14.589 | −0.447 | Effect | 6 |
| C8 | Redesigning internal structures | 12.975 | −1.622 | Effect | 6 | 20.442 | −0.156 | Effect | 4 | 13.811 | 0.623 | Cause | 8 |
| C9 | Improving digital maturity | 14.556 | 0.085 | Cause | 3 | 21.403 | −0.253 | Effect | 1 | 14.795 | −0.466 | Effect | 4 |
Figure 2 visualises prominence and net effect for each group and categorises each capability into one of four segments: Strong causes (I) exhibit high prominence and have a positive net effect. Strong effects (II) are high-prominence capabilities exhibiting a negative net effect. Weak causes (III) reflect low prominence scores and positive net effects. Weak effects (IV) have a negative net effect and low prominence.
The diagram contains 3 cause-and-effect maps titled Management, Digital change agents, and Operational staff. Each map plots Prominence, shown as R plus C, on the horizontal axis, and Net effect, shown as R minus C, on the vertical axis. The upper half is labelled cause capacities and the lower half is labelled effect capacities. Each map is divided into 4 quadrants: weak causes, strong causes, weak effects, and strong effects. The plotted capacities include balancing digital portfolio, rapid prototyping, navigating innovation ecosystems, digital scouting, digital mindset crafting, improving digital maturity, digital transformation planning, strategic agility, and redesigning internal structures. In the Management map, digital scouting and digital mindset crafting appear in the cause area, while strategic agility and digital transformation planning appear in the effect area. In the Digital change agents map, navigating innovation ecosystems, digital scouting, rapid prototyping, and digital mindset crafting appear in the cause area, while improving digital maturity, digital transformation planning, strategic agility, redesigning internal structures, and balancing digital portfolio appear in the effect area. In the Operational staff map, rapid prototyping and balancing digital portfolio appear in the cause area, while digital scouting, improving digital maturity, strategic agility, digital transformation planning, digital mindset crafting, redesigning internal structures, and navigating innovation ecosystems appear lower on the map.Prominence and net effect of capabilities
The diagram contains 3 cause-and-effect maps titled Management, Digital change agents, and Operational staff. Each map plots Prominence, shown as R plus C, on the horizontal axis, and Net effect, shown as R minus C, on the vertical axis. The upper half is labelled cause capacities and the lower half is labelled effect capacities. Each map is divided into 4 quadrants: weak causes, strong causes, weak effects, and strong effects. The plotted capacities include balancing digital portfolio, rapid prototyping, navigating innovation ecosystems, digital scouting, digital mindset crafting, improving digital maturity, digital transformation planning, strategic agility, and redesigning internal structures. In the Management map, digital scouting and digital mindset crafting appear in the cause area, while strategic agility and digital transformation planning appear in the effect area. In the Digital change agents map, navigating innovation ecosystems, digital scouting, rapid prototyping, and digital mindset crafting appear in the cause area, while improving digital maturity, digital transformation planning, strategic agility, redesigning internal structures, and balancing digital portfolio appear in the effect area. In the Operational staff map, rapid prototyping and balancing digital portfolio appear in the cause area, while digital scouting, improving digital maturity, strategic agility, digital transformation planning, digital mindset crafting, redesigning internal structures, and navigating innovation ecosystems appear lower on the map.Prominence and net effect of capabilities
Our analysis indicates strong perceptual differences across different management levels. No overall pattern can be identified. This effect can also be seen at the construct level (cf. Table 4). Managers identify digital sensing and seizing as causes and digital transforming as an effect, whereas digital change agents identify sensing and transforming as causes. For operational staff, only digital seizing is seen as a cause.
Prominence and net effect of capabilities
| Managers (MAN) | Digital change agents (DCA) | Operational staff (OPS) | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Capability | Ri + Ci | Ri-Ci | Identity | Rank | Ri + Ci | Ri-Ci | Identity | Rank | Ri + Ci | Ri-Ci | Identity | Rank |
| Digital sensing | 14.5712 | 0.2645 | Cause | 1 | 20.4420 | 0.0728 | Cause | 1 | 14.4189 | −0.7836 | Effect | 2 |
| Digital seizing | 12.9465 | 0.1459 | Cause | 3 | 19.2432 | −0.3524 | Effect | 3 | 14.6927 | 0.8802 | Cause | 1 |
| Digital transforming | 13.3645 | −0.4105 | Effect | 2 | 20.4237 | 0.2796 | Cause | 2 | 14.3985 | −0.0966 | Effect | 3 |
By further analysing interaction effects at construct level (cf. Table 5 and Figure 3), we can see that for our management sample, digital sensing is the most prominent capability which not only has a significant within-category effect but also significantly affects digital seizing and digital transforming. Futhermore, digital sening is heavitly influenced by digital seizing and digital transforming. Digital change agents also perceive digital sensing as a strong enabler influencing each category. A similar effect can be seen for digital transforming. By contrast, data for operational staff neither shows a within-category effect for digital sensing, nor a significant influence on other categories. For this sample, digital sensing has the strongest effect, influencing two categories.
Aggregated total relation matrix T for each sample and capability categories
| Managers (α = 0.76) | Digital change agents (α = 1.11) | Operational staff (α = 0.81) | |||||||
|---|---|---|---|---|---|---|---|---|---|
| Capability category | Digital sensing | Digital seizing | Digital transforming | Digital sensing | Digital seizing | Digital transforming | Digital sensing | Digital seizing | Digital transforming |
| Digital sensing | 0.8471 | 0.7856 | 0.8399 | 1.1366 | 1.1240 | 1.1585 | 0.7775 | 0.7288 | 0.7663 |
| Digital seizing | 0.7656 | 0.6644 | 0.7520 | 1.0698 | 1.0097 | 1.0690 | 0.9140 | 0.8043 | 0.8772 |
| Digital transforming | 0.7717 | 0.6834 | 0.7039 | 1.1885 | 1.1323 | 1.1298 | 0.8423 | 0.7690 | 0.7724 |
The diagram contains three cause-and-effect maps titled Managers, Digital change agents, and Operational staff. Each map plots Prominence, shown as R plus C, on the horizontal axis, and Net effect, shown as R minus C, on the vertical axis. The upper half represents cause capabilities, and the lower half represents effect capabilities. In the Managers map, digital sensing, digital seizing, and digital transforming appear in the cause area, while digital transforming is also connected to a lower effect position. The links between capabilities are marked with values including 0.7206, 0.4466, 0.4000, 0.5713, and 0.7226. In the Digital change agents map, digital transforming appears in the upper cause area, digital seizing appears slightly above the centre line, and digital sensing appears in the lower effect area. The links are marked with values including 1.1939, 1.1988, 1.0657, 1.0509, 1.3124, and 1.4200. In the Operational staff map, digital sensing appears in the upper cause area, digital transforming appears near the centre line, and digital seizing appears in the lower effect area. Vertical links between the three capabilities are marked with values including 1.2554, 1.5545, 1.1179, and 0.6737.Interaction effects at the construct level for each sample
The diagram contains three cause-and-effect maps titled Managers, Digital change agents, and Operational staff. Each map plots Prominence, shown as R plus C, on the horizontal axis, and Net effect, shown as R minus C, on the vertical axis. The upper half represents cause capabilities, and the lower half represents effect capabilities. In the Managers map, digital sensing, digital seizing, and digital transforming appear in the cause area, while digital transforming is also connected to a lower effect position. The links between capabilities are marked with values including 0.7206, 0.4466, 0.4000, 0.5713, and 0.7226. In the Digital change agents map, digital transforming appears in the upper cause area, digital seizing appears slightly above the centre line, and digital sensing appears in the lower effect area. The links are marked with values including 1.1939, 1.1988, 1.0657, 1.0509, 1.3124, and 1.4200. In the Operational staff map, digital sensing appears in the upper cause area, digital transforming appears near the centre line, and digital seizing appears in the lower effect area. Vertical links between the three capabilities are marked with values including 1.2554, 1.5545, 1.1179, and 0.6737.Interaction effects at the construct level for each sample
4.2 Analysis of perceived capability manifestations
Based on each respondent’s self-assessment and industry assessment on a five-point Likert scale (0 – capability does not exist, 1 – low, 2 – moderate, 3 – high and 4 – very high) we analysed the perceived capability manifestations. The table presents the mean evaluations for each group of respondents at the capability level, construct level and for the total digital transformation capability. Scores for capability categories and the overall digital transformation capability were calculated as average values. Theoretically, we could have applied sample-specific weights based on a capability’s prominence determined in our DEMATEL analysis. However, we came to the conclusion that this would have had a distorting effect. Therefore, we opted for an unweighted approach. The data suggests that digital transformation skills are not particularly strong in the construction industry. The ratings for the company and industry examined are largely low to moderate.
To statistically analyse our data set, we first compared the evaluations within each sample. Because our data is not normally distributed and therefore does not allow for computing a paired t-test, we applied the Wilcoxon signed-rank test as an alternative, non-parametric statistical instrument to compare the two related subsamples for each management level. In each case, our null hypothesis () defines that the median difference between pairs of observations is 0 and therefore assumes that no significant difference exists between the two measurements. The alternative hypothesis assumes that the median difference is not equal to 0. In this case, a significant difference between the subsamples is determined. We have to stress that the Wilcoxon signed-rank test does not compare mean values. We nonetheless included this information in Table 6 to gain a basic understanding of the underlying data basis. Significant p-values are highlighted in bold. For individual capabilities, in our management sample, significant differences between industry and self-assessment can be observed across the board. On the other hand, digital change agents and operational staff only perceive significant differences for select capabilities. C7 (navigating innovation ecosystems) is the only capability where a significant effect is observed in each case. At the construct level, we see a similar pattern for digital sensing. With respect to the unweighted aggregate evaluation of the total digital transformation capability, significant differences between self-assessment and industry assessment are discovered for managers, digital change agents and operational staff.
Comparison of self-assessment and industry assessment
| Managers | Digital change agents | Operational staff | |||||||
|---|---|---|---|---|---|---|---|---|---|
| Self | Industry | Wilcoxon paired | Self | Industry | Wilcoxon paired | Self | Industry | Wilcoxon paired | |
| Capability | Mean | Mean | p-value | Mean | Mean | p-value | Mean | Mean | p-value |
| C1 Digital scouting | 2.083 | 1.375 | 0.003 | 2.346 | 2.115 | 0.223 | 1.542 | 2.083 | 0.023 |
| C2 Digital scenario planning | 1.833 | 1.292 | 0.011 | 1.962 | 1.692 | 0.143 | 1.958 | 2.083 | 0.484 |
| C3 Digital mindset crafting | 1.875 | 1.167 | 0.006 | 2.038 | 1.692 | 0.059 | 1.375 | 2.042 | 0.004 |
| C4 Rapid prototyping | 1.458 | 1.125 | 0.040 | 1.346 | 1.308 | 1.000 | 1.417 | 1.500 | 0.714 |
| C5 Balancing digital portfolios | 1.875 | 1.125 | 0.001 | 1.654 | 1.346 | 0.001 | 1.792 | 1.958 | 0.276 |
| C6 Strategic agility | 2.000 | 1.375 | 0.002 | 1.192 | 1.269 | 0.805 | 1.792 | 2.125 | 0.040 |
| C7 Navigating innovation ecosystems | 1.750 | 1.292 | 0.009 | 2.192 | 1.731 | 0.026 | 1.250 | 1.958 | 0.000 |
| C8 Redesigning internal structures | 1.458 | 1.083 | 0.042 | 1.346 | 1.346 | 1.000 | 1.625 | 1.542 | 0.627 |
| C9 Improving digital maturity | 1.792 | 1.208 | 0.003 | 1.962 | 1.731 | 0.167 | 2.167 | 2.000 | 0.072 |
| Capability category | |||||||||
| Digital sensing | 1.931 | 1.278 | 0.000 | 2.115 | 1.833 | 0.039 | 1.625 | 2.069 | 0.004 |
| Digital seizing | 1.778 | 1.208 | 0.000 | 1.397 | 1.308 | 0.129 | 1.667 | 1.861 | 0.221 |
| Digital transforming | 1.667 | 1.194 | 0.002 | 1.833 | 1.603 | 0.066 | 1.681 | 1.833 | 0.111 |
| Total capability | 1.792 | 1.227 | 0.000 | 1.782 | 1.581 | 0.018 | 1.657 | 1.921 | 0.015 |
The data suggest that there are systematic differences between the groups studied. The boxplots depicted in Figure 4 – representing the mean evaluations of the total digital transformation capability – suggest that the groups assess the capabilities of their own company comparatively similarly, while cross-sample deviations can be observed for the assessment of competitors. To further investigate if there are differences between the samples, we applied the Kruskal–Wallis test as a non-parametric alternative to one-way ANOVA, which is not applicable given that the data is not distributed normally. It is commonly used to study multiple independent samples to determine if there are differences in their means. Our null hypothesis () is that there are no differences in means. By contrast, the alternative hypothesis (H1) suggests that differences exist between the groups. There are only two prerequisites to consider for the Kruskal–Wallis test. (1) There have to be more than two independent samples. (2) The y-variable must be at least ordinally scaled. Our data meets both conditions. Next, we apply the Dunn test as a post hoc test to find out between which specific samples potential differences exist. We also use the Hommel adjustment procedure for p-values to correct for multiple hypotheses testing, as it is more powerful than the Bonferroni method used in other studies.
The chart contains three box plots titled Managers, Digital change agents, and Operational staff. Each panel has a vertical scale from 0 to 4. Managers have a higher self-assessment range than industry assessment. The managers’ self-assessment median is about 1.8, while the industry assessment median is about 1.2. Digital change agents also have higher self-assessment than industry assessment. Their self-assessment median is about 1.8, and their industry assessment median is about 1.6. Operational staff have higher industry assessment than self-assessment. Their self-assessment median is about 1.7, while their industry assessment median is about 1.9. The legend identifies self-assessment and industry assessment.Comparison of digital capability self-assessment and industry assessment across management levels
The chart contains three box plots titled Managers, Digital change agents, and Operational staff. Each panel has a vertical scale from 0 to 4. Managers have a higher self-assessment range than industry assessment. The managers’ self-assessment median is about 1.8, while the industry assessment median is about 1.2. Digital change agents also have higher self-assessment than industry assessment. Their self-assessment median is about 1.8, and their industry assessment median is about 1.6. Operational staff have higher industry assessment than self-assessment. Their self-assessment median is about 1.7, while their industry assessment median is about 1.9. The legend identifies self-assessment and industry assessment.Comparison of digital capability self-assessment and industry assessment across management levels
As depicted in Table 7, with respect to the respondents’ self-assessment of individual capabilities, significant differences can primarily between digital change agents and operational staff, that is, for C1 (digital scouting), C3 (digital mindset crafting) and C7 (navigating innovation ecosystems). In addition, for C6 (strategic agility) the assessment of managers and digital change agents differs. At the construct level, there only seems to be a measurable difference for digital sensing when comparing managers and digital change agents. For all other comparisons, we reject H1.
Statistical evaluation of capability self-assessment and industry assessment across management levels
| Self-assessment | Industry assessment | |||||||
|---|---|---|---|---|---|---|---|---|
| Group 1 | MAN | MAN | DCA | MAN | MAN | DCA | ||
| Capability dimension | Statistical analysis | Group 2 | DCA | OPS | OPS | DCA | OPS | OPS |
| Capability | ||||||||
| C1 Digital scouting | Kruskal–Wallis RST | p-value | 0.0040 | 0.0023 | ||||
| Dunn test (Hommel) | Adj. p-value | 0.3150 | 0.0539 | 0.0033 | 0.0053 | 0.0046 | 0.9170 | |
| C2 Digital scenario planning | Kruskal–Wallis RST | p-value | 0.7494 | 0.0008 | ||||
| Dunn test (Hommel) | Adj. p-value | 0.8330 | 0.8330 | 0.8330 | 0.0529 | 0.1060 | 0.0005 | |
| C3 Digital mindset crafting | Kruskal–Wallis RST | p-value | 0.0421 | 0.0002 | ||||
| Dunn test (Hommel) | Adj. p-value | 0.6140 | 0.1240 | 0.0485 | 0.0388 | 0.0001 | 0.0622 | |
| C4 Rapid prototyping | Kruskal–Wallis RST | p-value | 0.8303 | 0.4184 | ||||
| Dunn test (Hommel) | Adj. p-value | 0.7700 | 0.7700 | 0.7700 | 0.5320 | 0.5320 | 0.5320 | |
| C5 Balancing digital portfolios | Kruskal–Wallis RST | p-value | 0.8114 | 0.0004 | ||||
| Dunn test (Hommel) | Adj. p-value | 0.7710 | 0.7710 | 0.7710 | 0.2210 | 0.0004 | 0.0145 | |
| C6 Strategic agility | Kruskal–Wallis RST | p-value | 0.0114 | 0.0006 | ||||
| Dunn test (Hommel) | Adj. p-value | 0.0121 | 0.4510 | 0.0701 | 0.6440 | 0.0045 | 0.0010 | |
| C7 Navigating innovation ecosystems | Kruskal–Wallis RST | p-value | 0.0007 | 0.0075 | ||||
| Dunn test (Hommel) | Adj. p-value | 0.0744 | 0.0744 | 0.0004 | 0.1270 | 0.0056 | 0.1880 | |
| C8 Redesigning internal structures | Kruskal–Wallis RST | p-value | 0.3186 | 0.0281 | ||||
| Dunn test (Hommel) | Adj. p-value | 0.7950 | 0.5010 | 0.3750 | 0.1800 | 0.0226 | 0.1800 | |
| C9 Improving digital maturity | Kruskal–Wallis RST | p-value | 0.2254 | 0.0004 | ||||
| Dunn test (Hommel) | Adj. p-value | 0.4030 | 0.2530 | 0.4030 | 0.0158 | 0.0003 | 0.1840 | |
| Capability category | ||||||||
| Digital sensing | Kruskal–Wallis RST | p-value | 0.0204 | 0.0000 | ||||
| Dunn test (Hommel) | Adj. p-value | 0.4020 | 0.1250 | 0.0186 | 0.0033 | 0.0000 | 0.1570 | |
| Digital seizing | Kruskal–Wallis RST | p-value | 0.1062 | 0.0032 | ||||
| Dunn test (Hommel) | Adj. p-value | 0.1290 | 0.6190 | 0.2580 | 0.4720 | 0.0039 | 0.0208 | |
| Digital transforming | Kruskal–Wallis RST | p-value | 0.4790 | 0.0004 | ||||
| Dunn test (Hommel) | Adj. p-value | 0.6620 | 0.8980 | 0.5410 | 0.0329 | 0.0002 | 0.1000 | |
| Total capability | ||||||||
| Kruskal–Wallis RST | p-value | 0.4354 | 0.0000 | |||||
| Dunn test (Hommel) | Adj. p-value | 0.8500 | 0.6350 | 0.4760 | 0.0087 | 0.0000 | 0.0476 | |
The respective evaluations for competitors in the construction industry reveal bigger discrepancies. The Kruskal–Wallis test indicates significant differences. Distinct differences can be seen among all comparison groups based on the Dunn test. They are especially evident between managers and operational staff. For this group, with the exception of C4 (rapid prototyping), a difference can be tested for all individual abilities as well as at the construct level and for the overall ability.
5. Discussion
Our study shows that there are clear differences in perception in the company surveyed, both in terms of the relationships and interactions between the capabilities examined and in the assessment of their manifestations. Prior studies have shown that digital transformation in construction is shaped by technological, organisational, strategic, workforce-related, policy and sustainability-oriented conditions, rather than by technology adoption alone (Oke et al., 2023b; Oke et al., 2023a; Zhang et al., 2023; Zhang et al., 2024). Other studies further highlight that digital transformation is influenced by how organisational actors frame technological change (Ernstsen et al., 2021) and by the organisational capabilities, implementation mechanisms, technological readiness and managerial resources required to apply Construction 4.0 and AI-enabled technologies in practice (Zhang et al., 2025; Zhang and Wang, 2026).
However, this study found that while (digital) sensing, seizing and transforming capabilities are certainly understood as a complex system, most studies often suggest a simplified understanding and often assume a discrete and linear relationship between sensing, seizing and transforming. Our work does not support this thesis. Although we understand that in many situations it makes sense or is even necessary to simplify to reduce the world’s inherent complexity at the model level, our research showed how interconnected digital capabilities are perceived. The analyses indicate that dynamic capabilities for digital transformation are perceived as a highly complex system with numerous interaction effects and interdependencies. This is in line with other recent work acknowledging the need for a more nuanced interpretation (Bechtel et al., 2023; Castka et al., 2024; Leemann et al., 2021).
We derive two key questions from the results. (A) How can the differences in perception be assessed in terms of their validity? (B) Why do the observed differences in perception exist? Because our study only examines perceptions, but no control or benchmarking instrument is available for validation and no comparable studies exist, the first question cannot be answered meaningfully. This means that the results must be interpreted with caution and no direct conclusions should be drawn regarding the actual relevance of the factors investigated. For example, the fact that managers perceive C1 (digital scouting), C2 (digital mindset crafting) and C9 (improving digital maturity) as strong enablers does not allow the conclusion to be drawn that companies in the construction industry must primarily invest in these skills to successfully master the digital transformation. This would also not make sense based on a synthesised and comprehensive analysis that includes all management levels examined.
5.1 Divergence in perception across capabilities
We therefore see the added value of this study primarily in highlighting the differences. These variances may be attributed to several factors, including the respondents’ distinct roles, responsibilities and experiences within an organisation. Firstly, the analysis revealed a managerial versus an operational focus as managers tend to adopt a broader, strategic perspective as digital transformation is primarily interpreted as a mechanism to enhance competitiveness, long-term efficiency and innovation capacity. From this perspective, capabilities such as digital scouting or digital mindset crafting are framed as enablers of strategic positioning. Operational staff, in contrast, interpret digital transformation through the lens of day-to-day practice, that is, capabilities are evaluated according to their immediate effect on workflows, routines and task execution. This divergence reflects a classical strategy execution tension: while top management formulates digital capabilities as instruments of competitive advantage, their operational enactment is filtered through feasibility, usability and workload implications. In a digital transformation context, such misalignment between strategic intent and operational interpretation has been shown to undermine implementation consistency (Chanias et al., 2019; Fernandez-Vidal et al., 2022). The findings suggest that dynamic capabilities in construction are not only strategic constructs, but also execution-dependent practices whose perceived relevance shifts across hierarchical levels, thereby potentially fragmenting the sensing-seizing-transforming cycle (Leemann et al., 2021).
Secondly, the results show different perceptions between opportunities and risks as senior managers frequently interpret digital transformation in terms of opportunity, such as investments in new technologies, portfolio balancing and ecosystem navigation are associated with growth potential and competitive advantage (Mikl et al., 2020; Rader, 2019). This orientation may explain why managers in our sample assess their own organisation’s digital maturity more favourably than competitors. In contrast, operational staff displays greater sensitivity to risks as new technologies are associated with learning curves, process disruption, increased workload and uncertainty regarding job stability. This divergence indicates that dynamic capabilities are framed differently along an opportunity-risk spectrum depending on hierarchical position, often emphasising strategic upside and legitimacy-building in managerial narratives (Bosch-Sijtsema et al., 2021; Herold et al., 2021). Frontline actors, however, may experience transformation through exposure to operational friction and ambiguity, leading to an asymmetry that can generate strategic optimism at the top, but also to rather cautious pragmatism below or even to overestimation of organisational readiness (Gfrerer et al., 2021; Singh and Hess, 2017).
Thirdly, differences in perceptions are also likely linked to varying levels of involvement in strategic decision-making. Managers are typically engaged in defining digital strategy and resource allocation, which may reinforce commitment and confidence in transformation initiatives (Canhoto et al., 2021; Geske et al., 2025). Operational staff, however, often experience digital initiatives as externally imposed as perceived limited participation in early planning phases may reduce ownership and increase scepticism, in particular towards capability-building efforts. Prior research demonstrates that digital transformation success depends on shared understanding and cross-level alignment (Jin et al., 2025; Poláková-Kersten et al., 2023; Weisz et al., 2025). The results indicate that when involvement is uneven, capability systems risk cognitive decoupling, where sensing and transforming are understood strategically, but are not collectively enacted.
Fourthly, the analysis showed the perceived differences may further stem from unequal access to training and strategic information, as managers often receive broader exposure to transformation rationales and industry benchmarking. Operational employees may encounter new tools without equivalent contextual framing or skill development, for example, when training structures lag behind implementation, operational staff in our study perceive competitors as more advanced. This pattern highlights that dynamic capabilities require systematic skill-building and diffusion across hierarchical levels, confirming that capability development is cumulative and learning-dependent (Bechtel et al., 2023; Magliocca et al., 2023; Warner and Wäger, 2019). The findings suggest that uneven training intensity may produce perceived gaps that may weaken organisational coherence, particularly in industries such as construction, where professional subcultures and project-based routines already challenge integration.
5.2 Divergence in perception across hierarchical levels
The findings also have implications for the perceived divergence among hierarchies. The findings show that senior managers often take a high-level, strategic view of the organisation, focusing on overarching goals, competitive positioning and investment portfolios related to digital transformation. As such, senior management may perceive the company’s digital capabilities through the lens of these investments, assuming that allocated resources and formulated strategies translate into sustained competitive advantage. This perspective tends to privilege formal initiatives, strategic roadmaps and long-term transformation programs as indicators of capability development. The findings show that this perspective may also distance senior managers from the operational frictions that accompany digital implementation. As a result, their assessment of digital maturity may be shaped more by strategic intent and planned trajectories than by the practical realities of adoption and integration. Prior research suggests that executive actors frequently interpret transformation progress in relation to strategic ambition and narrative coherence rather than everyday execution (Fernandez-Vidal et al., 2022; Kunisch et al., 2022). This may foster a degree of strategic overconfidence, in particular in industries where digital investments signal modernisation and legitimacy (Bosch-Sijtsema et al., 2021).
We also found that digital change agents occupy an intermediate position between strategic formulation and operational execution. While they are directly involved in the design and coordination of digital transformation initiatives, they also possess a relatively nuanced understanding of implementation challenges, that is, their perceptions reflect both awareness of strategic objectives and exposure to practical constraints such as resistance, technical integration issues and adoption speed. This “hybrid” positioning appears to be two-sided: while digital change agents recognise progress and strategic direction, they are also confronted with the complexity of orchestrating sensing, seizing and transforming activities across departments. Research on digital transformation emphasises the critical role of boundary-spanning actors who translate strategic intent into operational practice, but at the same time, such actors are often exposed to tensions arising from competing priorities and resource constraints (Azzouz and Papadonikolaki, 2020; Fleischer and Carstens, 2022). The results suggest that digital change agents function as cognitive intermediaries within the organisation’s dynamic capability architecture, thereby mitigating (but not resolving) perceived divergences between top management and frontline actors.
The analysis also showed that operational staff engaging with digital transformation at the level of daily routines, tools and task execution are shaped by a direct interaction with digital systems and by the immediate consequences for productivity, workload and coordination. From this perspective, digital capabilities are less abstract constructs and more tangible artefacts embedded in workflows. In contrast to the senior managers or digital change agents, operational employees are typically not involved in strategic planning processes and digital transformation may therefore be experienced as externally imposed rather than collectively shaped. Prior research on technology adoption highlights that frontline perceptions are strongly influenced by usability, clarity of purpose and perceived fairness of change processes (Alkaabi et al., 2025; Montargot and Ben Lahouel, 2018). In project-based industries such as construction, where routines are tightly coupled to time and cost pressures, even minor inefficiencies can significantly influence evaluation. The findings indicate that operational actors ground dynamic capabilities in lived experience, thereby providing a reality check that can either strengthen or destabilise organisational transformation trajectories depending on the quality of alignment.
5. Conclusion
This study examined how dynamic capabilities for digital supply chain transformation are perceived across hierarchical levels and how the relationships between these capabilities are understood in the construction industry. Using DEMATEL, we modelled perceived cause–effect relationships between digital sensing, seizing and transforming capabilities across senior managers, digital change agents and operational staff.
The findings show that these capabilities are perceived as interdependent rather than as discrete or strictly sequential dimensions of digital transformation. At the same time, the results reveal systematic divergence in how capabilities are prioritised, interpreted and evaluated across hierarchical groups. Senior managers tend to frame digital capabilities in strategic and opportunity-oriented terms, digital change agents occupy an intermediary position shaped by both strategic exposure and implementation realities, while operational staff evaluate digital capabilities through their immediate practical implications for workflows, routines and task execution. These findings suggest that digital supply chain transformation in construction depends not only on the availability of digital technologies or formal transformation strategies, but also on the extent to which digital capabilities are understood and aligned across hierarchical levels.
This study contributes to the literature in four ways. Firstly, we advance dynamic capability research by examining digital sensing, seizing and transforming as an interdependent system rather than as discrete or sequential capability dimensions. By modelling perceived causal relationships, the findings reinforce recent calls to understand dynamic capabilities as mutually reinforcing and feedback-driven processes. Secondly, we contribute to research on digital transformation in construction supply chains by identifying which digital capabilities are perceived as strategically central and how their causal importance differs across hierarchical levels. Thirdly, we extend microfoundational perspectives on dynamic capabilities by showing that hierarchical position shapes how capabilities are cognitively framed, prioritised and evaluated. This suggests that dynamic capabilities are not only organisational-level constructs, but also distributed interpretive structures whose effectiveness depends on cross-level alignment. Fourthly, by applying DEMATEL as a systematic approach for modelling perceived cause–effect structures in construction management research, thereby offering a replicable analytical tool for studying complex transformation systems in project-based environments.
For practitioners in the construction industry, the findings highlight the strategic importance of alignment across hierarchical levels. Digital transformation initiatives should not be evaluated solely in terms of technological investment or formal strategy formulation. Attention must also be directed towards how digital capabilities are interpreted, enacted and experienced throughout the organisation. Senior managers should actively create feedback mechanisms that surface operational implementation realities, thereby strengthening the sensing function within the organisation itself. Digital change agents should be supported as boundary-spanning actors who translate strategic intent into operationally viable practices. At the operational level, systematic training and contextual communication are essential to embed capabilities beyond symbolic adoption. In industries where digital transformation is often pursued through project-based experimentation, sustained capability diffusion across hierarchical layers becomes a critical determinant of long-term innovation performance.
However, the results have to be viewed in the light of their limitations. This study relies on an embedded single case design within one construction firm in the German and Austrian markets, which limits generalisability, however, the findings may be transferable to the wider construction industry due to shared sector characteristics. The findings reflect perceptual assessments rather than objective performance measures, and the DEMATEL approach captures cognitively constructed cause–effect structures that may be influenced by respondent bias. Future research could extend this work through comparative multi-case designs across institutional contexts, integrate perceptual modelling with objective performance indicators and examine longitudinal shifts in capability alignment as digital initiatives mature. Given the fragmented nature of construction supply chains, further research should also explore inter-organisational capability alignment across contractors and subcontractors to better understand how digital transformation scales beyond the firm level.
Declaration of generative AI and AI-assisted technologies in the manuscript preparation process
During the preparation of this manuscript, the author(s) used ChatGPT (OpenAI) to support language refinement, structural suggestions and clarity of expression. All outputs were critically reviewed and edited by the author(s), who take full responsibility for the content of the publication. While AI may have helped shape sentences, it had no role in shaping the arguments, thus any remaining errors, questionable judgments or moments of academic overconfidence are entirely the fault of the authors.
