Leading in the Digital Age: Digital Leadership Capabilities, Organisational Innovation Climate, and AI Adoption Intention Among SMEs in Nigeria

Abstract

Although small and medium enterprises (SMEs) anchor employment and output across Sub-Saharan Africa, their uptake of artificial intelligence (AI) lags global benchmarks, and prevailing explanations dwell on capital, infrastructure, and institutional voids while overlooking the leadership competencies that determine whether available resources are mobilised at all. Addressing this gap, the present study asks how the digital leadership capabilities of SME owner-managers shape their intention to adopt AI in Nigeria, and through what organisational mechanisms and under what boundary conditions this influence operates. Anchored in the Diffusion of Innovation Theory and the Tigre–Henriques–Curado model of digital leadership, a cross-sectional survey was administered to owner-managers of registered SMEs drawn from six states; a sample of 390 was derived from a population of 23,290 firms using the Taro Yamane formula with proportionate allocation, and 306 valid responses were retained. Partial Least Squares Structural Equation Modelling (WarpPLS 8.0) was applied after confirming reliability (Cronbach’s $\alpha$: 0.69–0.84; composite reliability: 0.83–0.88), convergent validity (AVE: 0.56–0.67), and common method bias control. Strategic ($\beta$ = 0.298), interpersonal ($\beta$ = 0.245), and personal attribute ($\beta$ = 0.129) capabilities each significantly raised AI adoption intention. In contrast, delivery-related capabilities ($\beta$ = 0.090, p = 0.057) did not, indicating that pre-adoption intention is governed by cognitive-strategic and relational competencies rather than execution skills. Organisational innovation climate partially transmitted the effects of strategic and interpersonal capabilities, and firm size amplified the interpersonal pathway in medium-sized firms. The study contributes a leadership-centred account of AI adoption in an under-researched African setting and, by estimating mediation and moderation within a single framework, clarifies both why and when digital leadership translates into AI readiness, yielding capability-specific guidance for owner-managers and SME support policy.

1. Introduction

In a rapidly evolving, digitally interconnected global economy, small and medium enterprises (SMEs) must explore and exploit emerging technological opportunities to remain competitive and ensure long-term survival. Artificial intelligence (AI) presents transformative potential in this regard, enabling SMEs to improve operational efficiency, strengthen competitive positioning, and respond more adaptively to changing market demands [1]. Integration of AI technologies supports the dynamic capabilities of SMEs, facilitating swift responses to environmental uncertainty, the redesign of business processes, and the sustainability of long-term competitiveness [2,3]. Globally, evidence suggests that 91% of SMEs using AI report direct revenue gains, underscoring the strategic imperative of AI adoption for enterprise growth [4].

The intention of SMEs to adopt AI technologies is shaped by a complex interplay of organisational, technological, situational, and individual factors [2,5]. Among these, leadership has emerged as a central driver, influencing strategic direction while cultivating environments that support data-driven decision-making, innovation, and the development of essential employee competencies for digital transformation [6,7]. The adoption of AI within SMEs is thus closely related to the presence of strong digital leadership capabilities: leaders who possess the competencies to understand AI technologies, communicate their benefits, and integrate them effectively into organisational processes are better positioned to initiate adoption [8,9].

Despite growing scholarly attention to the intersection of digital leadership and technology adoption, research focusing specifically on Nigerian SMEs remains scarce. Nigeria, Africa’s largest economy, hosts more than 41 million SMEs that collectively contribute approximately 48% of national GDP and account for 84% of total employment [10]. However, existing accounts of weak AI adoption among African SMEs overwhelmingly focus on external and resource-based constraints such as infrastructural deficits, financing gaps, scarce digital skills, and institutional gaps [7,11,12,13]. Far less attention has been paid to an internal, agentic explanation: why firms possessing comparable basic resources nevertheless differ markedly in whether they initiate AI adoption at all. We argue that this divergence is rooted substantially in digital leadership, the strategic, relational, and adaptive competencies of owner-managers who must first recognise AI’s relevance, frame its value, and marshal organisational commitment before any technical or financial enabler can take effect. In this way, the digital leadership deficit, rather than capital or infrastructure alone, constitutes a core bottleneck that restricts AI adoption in much of the African SME sector. This reframing both motivates the present study and structures its research design, which isolates leadership capability as the focal antecedent while accounting for the resource and structural factors that prior work has emphasised.

Furthermore, existing research has predominantly examined direct relationships between leadership and technology adoption, with limited attention to the mechanisms through which digital leadership capabilities translate into adoption intention. Individual leadership traits do not act on collective firm decisions in isolation; their value must be converted into organisational capacity through shared norms and collaborative routines [14]. The organisational innovation climate is the shared perception that an organisation encourages and supports innovative behaviour and therefore may serve as a critical mediating mechanism, since leaders who foster innovation-supportive environments amplify the translation of their capabilities into organisational technology adoption decisions [15,16]. Additionally, firm size may moderate these relationships, as larger SMEs typically have greater resources, more formalised structures, and stronger absorptive capacity than their smaller counterparts [17].

Accordingly, this study pursues four primary objectives: (1) to empirically examine the direct effects of four dimensions of digital leadership capabilities, strategic, delivery-related, interpersonal, and personal attributes on AI adoption intention among Nigerian SMEs; (2) to investigate the mediating role of the organisational innovation climate in the relationship between digital leadership capabilities and AI adoption intention; (3) to evaluate the moderating effect of firm size on the associations between the dimensions of digital leadership capability and AI adoption intention; and (4) to generate evidence-based recommendations for policymakers and SME support agencies seeking to accelerate AI adoption within Nigeria’s SME sector. Situating the study within this framework embeds it in a broader research agenda: AI increasingly functions not only as a driver of operational efficiency but also as a catalyst for the green and sustainable transformation of firms operating under resource constraints, with digital leadership shaping the initial conditions and trajectory of this transformation [18].

This study offers three principal contributions. Theoretically, it extends Diffusion of Innovation Theory and the Tigre–Henriques-Curado digital leadership model to the context of small and medium-sized enterprises (SMEs) in developing economies. It conceptualises the organisational innovation climate as the transmission mechanism that links individual leader competencies to collective adoption intentions, thereby elucidating why leadership capabilities require an organisational conduit to influence firm-level decision-making. Empirically, it delivers one of the first capability-disaggregated PLS-SEM examinations of digital leadership and artificial intelligence (AI) adoption among African SMEs. By estimating mediation and moderation effects within a unified structural model, it simultaneously identifies both the underlying causal pathway and the firm size-contingent boundary of the effect, rather than analysing these dimensions in isolation. Practically, it translates these results into segment-specific and capability-specific recommendations for owner-managers, as well as for the design of differentiated SME support and training policies.

2. Literature Review and Hypothesis Development

2.1. AI Adoption Intention: Conceptual Foundations

Artificial intelligence (AI) comprises computational systems capable of perceiving, interpreting, acting upon, and incrementally learning from data with varying degrees of autonomy [19,20]. Within organisational settings, AI is instantiated through a range of digital artefacts, including intelligent agents, conversational chatbots, expert systems, natural language processing applications, and predictive analytics platforms that collectively support the automation, optimisation, and enhancement of business processes and decision-making [21]. The economic significance of AI is considerable: the global AI market is estimated to expand from USD 233.46 billion in 2024 to USD 1771.62 billion by 2032, corresponding to a compound annual growth rate (CAGR) of 29.2% [4].

Adoption intention denotes the degree to which an individual or organisation is both willing and prepared to engage with a novel technology [22]. Drawing on the Theory of Planned Behaviour and the Technology Acceptance Model, prior research has consistently demonstrated that intention is a robust predictor of actual adoption behaviour [22,23,24,25]. In the context of small and medium-sized enterprises (SMEs), the intention to adopt artificial intelligence (AI) encompasses both current usage and anticipated future deployment, thereby capturing a gradual and iterative integration trajectory [26,27].

In developing economies such as Nigeria, however, adoption remains significantly constrained by financial, technological, knowledge-related, cultural, and governance barriers [5,7,12]. Empirical evidence from Africa indicates that, although SMEs in South-West Nigeria exhibit relatively high levels of AI awareness, actual adoption rates are low. In this setting, managerial cognition and leadership support, rather than infrastructural capacity, emerge as the principal determinants of AI adoption [28]. Pan-African reviews similarly affirm that organisational and human factors shape AI adoption to a degree comparable to, if not exceeding, that of technological resources and capabilities [13].

Systematic reviews further reveal heterogeneous evidence regarding the specific leadership competencies that influence AI adoption; operational leadership capabilities appear more strongly associated with post-adoption implementation and routinisation than with the initial decision to adopt [29]. Moreover, the extant literature is heavily skewed towards European and North American SMEs, with comparatively limited empirical attention to African contexts [30]. This geographical imbalance highlights a critical need for context-sensitive, region-specific investigations of AI adoption dynamics.

2.2. Digital Leadership and Digital Leadership Capabilities

Digital leadership has emerged from the concept of e-leadership and is concerned with exercising leadership effectively in digitalised environments while concurrently steering organisational transformation [31]. It encompasses the active involvement of organisational members in digitalisation processes, the development of requisite digital and socio-technical competencies, and the cultivation of organisational cultures oriented towards continuous digital learning [32]. Contemporary digital leaders are characterised by their capacity to promote inclusive organisational cultures, to install and uphold ethical digital governance structures, and to empower employees, thereby extending their role beyond the mere adoption of technological tools towards the orchestration of comprehensive strategic transformation [33,34].

Digital leadership capabilities are increasingly conceptualised as multidimensional constructs. Abbu et al. [35] distinguish between character-related capabilities (e.g., integrity, ethical approaches to AI, and transparency) and competency-related capabilities (e.g., digital literacy, continuous skill acquisition, and knowledge sharing). Complementing this view, Chen [36] identifies technology literacy, data-driven decision-making, innovative thinking, and collaborative orientation as key dimensions of digital leadership. Similarly, Hokmabadi et al. [37] underscore the importance of strategic foresight, change management, and digital innovation in shaping effective digital leadership.

Synthesising and extending these prior conceptualisations, Tigre et al. [31] propose a four-dimensional framework of digital leadership capabilities comprising: strategic capabilities (vision, change management, innovation, agility, and calculated risk-taking); delivery-related capabilities (analytical thinking, technological proficiency, team performance, and results orientation); interpersonal capabilities (relationship building, communication, coaching, and fostering psychological safety); and personal attributes (adaptability, lifelong learning, ethical behaviour, and empathy). Empirical evidence further indicates that leaders’ digital capabilities constitute a central antecedent of internal AI implementation, as executives’ digital awareness has a direct and significant influence on organisational AI adoption [8,9].

2.3. Strategic Capabilities and AI Adoption Intention

Strategic capabilities denote a leader’s capacity to plan and orchestrate digital transformation by articulating a clear vision, establishing digital alignment, and effectively executing implementation processes [38,39]. This dimension encompasses data management to support evidence-based decision-making, the continuous updating and optimisation of digital tools, as well as systematic risk and change management practices [40].

The accumulation of empirical evidence indicates a positive impact of strategic capabilities on technology adoption. Hossain et al. [9] demonstrated that aligning technological initiatives with overall strategic objectives facilitates the adoption of artificial intelligence rather than operational competence, whereas Yu et al. [8] found that executives’ digital expertise accelerates the organisational deployment and use of AI. In a structural equation modelling study (PLS-SEM) involving 245 employees in Pakistan, Mahmood et al. [41] showed that strategic digital leadership amplifies the positive effects of AI on organisational performance. Furthermore, Suljic [42] argued that strategic leadership in AI-enabled transformation processes is essential for both multinational corporations and small and medium enterprises (SMEs).

These findings are strongly underpinned by Diffusion of Innovation (DOI) theory. In particular, Rogers’ concept of “relative advantage” provides a theoretical rationale for why visionary leaders not only recognise the potential of AI but also actively champion and communicate its anticipated benefits throughout the organisation [43]. Thus:

Hypothesis 1 ($H_1$).

Strategic capabilities have a significant positive influence on the intention to adopt AI among SMEs in Nigeria.

2.4. Delivery-Related Capabilities and AI Adoption Intention

Delivery-related capabilities comprise the practical competencies required to convert digital strategies into realised outcomes, including analytical reasoning, technological proficiency, team performance, collaborative capacity, and outcome orientation [31]. Their relationship with adoption intention is theoretically ambiguous. On the one hand, Sony et al. [44] contended that operational competence facilitates the implementation of Industry 4.0 technologies. On the other hand, “strategy-before-execution” perspectives posit that firm-level performance improvements are primarily driven by top-level strategic design; in the absence of adequate strategic cognition, enhancements in execution alone generate limited benefits [45]. Given that adoption intention is a cognitive, pre-implementation rather than operational competence, the needed skills may be more consequential in the post-adoption phase [46]. Consistent with this view, Arroyabe et al. [47], using data from 12,108 EU SMEs, reported that digital maturity and innovation capability, rather than operational competence, constituted the principal enablers of AI adoption. To formally examine this proposition, we hypothesise:

Hypothesis 2 ($H_2$).

Delivery-related capabilities have a significant positive influence on the intention to adopt AI among SMEs in Nigeria.

2.5. Interpersonal Capabilities and AI Adoption Intention

Interpersonal capabilities such as communication, team leadership, digital collaboration, network development, coaching, and the cultivation of psychologically safe work environments constitute critical antecedents for generating collective commitment to technological transformation [31,40]. Empirical evidence indicates that collaboration, trust, transparency, and ethically grounded communication are key drivers of employee engagement in organisations implementing AI-based systems [48,49]. Yang et al. [50] provide evidence that digital leadership positively affects employee creativity, primarily through the mediating mechanism of knowledge sharing. Similarly, Alghamdi [51], in a study of 158 leaders in the United Kingdom, identified robust positive associations between leaders’ interpersonal digital competencies and favourable attitudes towards AI. Within Diffusion of Innovations (DOI) theory, the notion of the “social system” underscores the centrality of interpersonal influence processes in the adoption of innovations [43]. Hence:

Hypothesis 3 ($H_3$).

Interpersonal capabilities have a significant positive influence on the intention of adopting AI among SMEs in Nigeria.

2.6. Personal Attributes and AI Adoption Intention

Personal attributes, including adaptability, a disposition towards lifelong learning, ethical grounding, initiative, empathy, and multi-perspective thinking, equip leaders to address transformation-related challenges effectively [52,53]. Empirical evidence indicates that openness, adaptability, and digital self-efficacy increase individuals’ willingness to adopt and utilise new technologies [51,54]. Consistently, Malik et al. [55] identified agility, openness, and participative orientation as critical characteristics for a successful digital transformation. Furthermore, the Diffusion of Innovations (DOI) construct of “innovativeness” conceptually corresponds to a combination of risk tolerance, openness, and adaptability [43]. Therefore:

Hypothesis 4 ($H_4$).

Personal attributes have a significant positive influence on the intention to adopt AI among SMEs in Nigeria.

2.7. The Mediating Role of Organisational Innovation Climate

The organisational innovation climate denotes the collectively shared perception that the organisation systematically encourages, supports, and rewards innovative behaviour [15,16]. Digital leadership capabilities constitute a key antecedent of this climate: leaders who exhibit a clear strategic vision, foster collaboration, and demonstrate adaptive thinking are more likely to establish environments that are conducive to innovation. A stronger innovation climate, in turn, attenuates perceived risks and enhances employees’ receptivity to artificial intelligence (AI)–based initiatives.

Empirical evidence supports this mechanism. Yansen and Yujie [56] show that transformative digital leadership promotes innovation in the product, process and organisational domains. Similarly, Sarkis and Pallotta [57] report that although leaders face resistance during AI adoption, the implementation process simultaneously nurtures organisational competencies that support innovation. Recent work further indicates that digital leadership enhances both organisational resilience and innovative capacity by cultivating supportive organisational cultures [58].

Taken together, these findings suggest that the innovation climate serves as a central mediating mechanism by which digital leadership translates into AI adoption intentions. Individual-level digital competence among leaders requires an organisational channel to translate into collective, firm-level strategic decisions [14]. Within this process, innovation climate represents the most immediate collective manifestation of leadership behaviours that encourage experimentation, tolerate ambiguity, and reward novelty. Accordingly:

Hypothesis 5 ($H_5$).

The organisational innovation climate mediates the relationship between the ability to lead digitally and the intention to adopt AI among SMEs in Nigeria.

2.8. The Moderating Role of Firm Size

Firm size is widely recognised as a critical contextual determinant in technology adoption processes. Larger SMEs generally possess more substantial financial resources, more formalised managerial structures, greater absorptive capacity, and deeper pools of human capital [17,59]. These attributes can enhance the effectiveness of leadership in driving technological change. Empirically, Badghish and Soomro [17], examining Saudi Arabian SMEs, identified a moderating effect of firm size on the relationship between AI adoption and organisational performance, with more pronounced performance gains observed among medium-sized enterprises. Similarly, Arroyabe et al. [47] documented heterogeneous AI adoption trajectories across firm size categories within EU-based SMEs. In the Nigerian context, micro-enterprises (i.e., firms with fewer than 10 employees) encounter severe resource constraints that may weaken the extent to which leadership capabilities are translated into intentions to adopt new technologies [10]. Therefore:

Hypothesis 6 ($H_6$).

Firm size moderates the relationship between digital leadership capabilities and the intention to adopt AI, so that the relationship is stronger for medium-sized enterprises than for micro and small enterprises.

2.9. Theoretical Framework

This study integrates two complementary perspectives. First, Diffusion of Innovation (DOI) Theory [43] explains how innovations spread through social systems through four elements: innovation, communication channels, time, and the social system. It identifies five perceived attributes influencing adoption (relative advantage, compatibility, complexity, trialability, observability), and categorises adopters by innovation. DOI provides a foundation for understanding how individual characteristics (personal attributes), social influence (interpersonal capabilities), and perceived innovation benefits (strategic capabilities) shape the intention of adoption. Second, the Tigre, Henriques, and Curado Model of Digital Leadership Capabilities [31] operationalises the four dimensions examined. The integration works as follows: the Tigre–Henriques–Curado model provides the antecedent structure (the four capability dimensions), while the DOI provides the explanatory logic for why each dimension affects the intention of adoption. Strategic capabilities map to relative advantage, interpersonal capabilities to social influence, personal attributes to individual innovation, and delivery-related capabilities to the management of perceived complexity at implementation. The organisational innovation climate is positioned as the mediator: the collective, climate-level realisation of leader behaviours that turns individual capability into firm-level decisions. Firm size enters as a contextual moderator. The conceptual model is shown in Figure 1 and summarised in

Table 1. Conceptual mapping of constructs, theoretical anchors, and expected relationships.

Construct (Role)	Theoretical Anchor	Expected Relationship
Strategic capabilities (IV)	DOI: relative advantage	Positive direct effect on AIAI ($H_1$); indirect via OIC ($H_5$)
Delivery-related capabilities (IV)	DOI: complexity (implementation)	Effect on AIAI hypothesised but theoretically contested ($H_2$)
Interpersonal capabilities (IV)	DOI: social system/influence	Positive direct effect on AIAI ($H_3$); indirect via OIC ($H_5$)
Personal attributes (IV)	DOI: innovativeness	Positive direct effect on AIAI ($H_4$)
Organisational innovation climate (Mediator)	Innovation climate theory	Transmits capability effects to AIAI ($H_5$)
Firm size (Moderator)	Resource-based view	Strengthens capability–AIAI paths in larger SMEs ($H_6$)
AI adoption intention (DV)	TPB/TAM, DOI	Proximal determinant of actual AI adoption

IV = independent variable; DV = dependent variable; AIAI = AI adoption intention; OIC = organisational innovation climate.

.

3. Materials and Methods

3.1. Research Design and Population

This study used a cross-sectional survey research design, which is appropriate to examine relationships between variables at a single time and is widely used in organisational behaviour and management research [60]. The target population comprised 23,290 registered SMEs in six Nigerian states (Lagos, Ogun, Oyo, Osun, Ekiti, and Ondo), as recorded in the Small and Medium Enterprises Development Agency of Nigeria (SMEDAN) database; therefore, the sampling frame and the population on which the sample size was based consist of enterprises formally classified as SMEs, not of the general business population. Although these states are geographically located in South-West Nigeria, they collectively represent the most commercially active and digitally advanced SME ecosystem in the country, housing the nation’s primary commercial hub (Lagos) and several major industrial centres. However, we acknowledge this as a source of potential regional bias: South-West firms may enjoy stronger digital infrastructure and market exposure than firms in other geopolitical zones, so the estimates are best read as reflective of Nigeria’s most digitally mature SME segment, with extension to other zones flagged as a limitation and a direction for future research.

3.2. Sample Size Determination and Sampling Technique

The sample size was determined using the Taro Yamane formula [61]:

$$n={\displaystyle \frac{N}{1+N{\left(e\right)}^2}}$$

(1)

where N = 23,290 (total population) and e = 0.05 (margin of error), yielding a required sample of 390 SMEs. The Bowley proportionate allocation formula distributed sampling across states proportional to each state’s SME population (

Table 2. Proportional distribution of the sample size across Nigerian states.

State	Number of SMEs	Proportional Sample
Lagos State	8396	141
Ogun State	2465	41
Oyo State	6131	103
Osun State	3007	50
Ekiti State	928	16
Ondo State	2363	39
Total	23,290	390

). Within each state, SMEs were selected using stratified random sampling in three categories of firm size: micro (1–9 employees), small (10–49 employees), and medium (50–199 employees).

3.3. Research Instrument

Data were collected using a structured questionnaire administered to SME owners and managers—individuals with decision-making authority over technology adoption. The questionnaire comprised four sections: (A) demographic characteristics; (B) digital leadership capability items adapted from Tigre et al. [31], measuring strategic capabilities (SC; 3 items), delivery-related capabilities (DRC; 6 items), interpersonal capabilities (IC; 4 items), and personal attributes (PA; 3 items); (C) organisational innovation climate items adapted from Scott and Bruce [15], measuring the degree to which organisations support innovative behaviour (OIC; 4 items); and (D) AI adoption intention items adapted from established technology adoption scales [23,27] (AIAI; 4 items). All items were measured on a five-point Likert scale (1 = strongly disagree to 5 = strongly agree).

Because the original instruments were developed in Western and East Asian organisational settings, the elements were contextually adapted to the Nigerian SME environment before being fielded. The items framed for large corporations referred to the realities of the owner-manager and small-firm context (e.g., the “executive team” was rendered as “you and your management”); (ii) two domain experts in information resources management and an SME practitioner reviewed each item for face and content validity, after which ambiguous or doubly-loaded items were reworded; and (iii) a pilot study with 40 SME managers (not part of the main sample) was conducted to confirm clarity, comprehensibility, and internal consistency. The pilot returned acceptable construct-level reliability (Cronbach’s $\alpha$ between 0.71 and 0.86) and informed the final sample size adequacy assessment based on the observed variance and model complexity; items that reduced reliability or were poorly understood were dropped, which accounts for the retained indicator counts reported in the measurement model. Of the 390 questionnaires distributed, 306 were valid, yielding an effective response rate of 78.5%.

3.4. Analytical Technique

Data were analysed using Partial Least Squares Structural Equation Modelling (PLS-SEM) with WarpPLS version 8.0 software [62]. PLS-SEM was selected for three methodological reasons: (1) its suitability for data that exhibit non-normal distributions, common in survey-based behavioural research [60]; (2) its effectiveness with moderate sample sizes relative to model complexity [63]; and (3) its ability to simultaneously estimate complex path models that include direct, mediating, and moderating relationships [59].

The analysis followed a rigorous two-stage approach [59,60]. Stage one assessed the measurement model through indicator reliability (factor loadings ≥ 0.70), internal consistency reliability (Cronbach’s alpha ≥ 0.70; composite reliability ≥ 0.70), convergence validity (AVE ≥ 0.50), and discriminant validity (Fornell–Larcker criterion, HTMT < 0.90 and cross loads). Stage two assessed the structural model through model fit indices (APC, ARS, AARS, AFVIF, GoF), path coefficient analysis, mediation testing via the Sobel test and bootstrapped indirect effects, and moderation testing through interaction term analysis. The effect sizes ($f^2$) were evaluated using Cohen’s [64] benchmarks: 0.02 (small), 0.15 (medium), and 0.35 (large).

3.5. Bias Assessment and Causal Considerations

Because the data are self-reported and collected from a single source at a time point, three potential threats were addressed. First, common method bias (CMB) was evaluated procedurally and statistically. Procedurally, respondent anonymity was ensured, predictor and criterion items were separated within the instrument, and item wording was simplified to reduce evaluation apprehension. Statistically, Kock’s full collinearity test was applied. All variance inflation factors in the full collinearity assessment were below the conservative 3.3 threshold (and well below the 5.0 ceiling reported in

Table 3. Model fit and quality indices.

Index	Full Name	Threshold	Value	Decision
APC	Average Path Coefficient	p < 0.05	0.190 (p < 0.001)	Accepted
ARS	Average R-Squared	p < 0.05	0.418 (p < 0.001)	Accepted
AARS	Avg. Adjusted R-Squared	p < 0.05	0.410 (p < 0.001)	Accepted
AFVIF	Avg. Full Collinearity VIF	≤5.0	2.277	Accepted
GoF	Tenenhaus Goodness-of-Fit	>0.36	0.505	Accepted
SPR	Sympson’s Paradox Ratio	≥0.70	0.857	Accepted
RSCR	R-Sq. Contribution Ratio	≥0.90	0.981	Accepted
SSR	Statistical Suppression Ratio	≥0.70	1.000	Accepted

GoF thresholds: small > 0.10; medium > 0.25; large > 0.36 [62].

, AFVIF = 2.277), indicating that common method bias is unlikely to have materially distorted the estimates [62]. Second, non-response bias was examined by comparing early and late respondents (the last quartile of returns serving as a proxy for non-respondents) on key demographic and construct means; independent-sample comparisons revealed no statistically significant differences, suggesting that nonresponse is not a serious concern. Third, the design cannot rule out reverse or reciprocal causality. For example, firms already favourably inclined towards AI may cultivate stronger innovation climates, and adoption-oriented firms may attract or develop more digitally capable leaders. Although the hypothesised directions are grounded in theory (leadership and climate as antecedents of intention), the cross-sectional design precludes a definitive causal ordering, a limitation that longitudinal or experimental designs should address.

Table 3 presents the fit indices of the structural model, all of which satisfy established thresholds, indicating strong model quality and explanatory power.

4. Results

4.1. Demographic Profile of Respondents

Table 4. Demographic characteristics of respondents (n = 306).

Variable	Frequency	Percentage (%)
Gender
Male	202	66.0
Female	104	34.0
Age
Less than 18 years	12	3.9
18–28 years	160	52.3
29–39 years	85	27.8
40–50 years	42	13.7
51–60 years	5	1.6
61 years and above	2	0.7
Educational Qualification
PhD	7	2.3
MSc	29	9.5
BSc/BA	180	58.8
HND	53	17.3
OND	37	12.1
Years of Business Operation
0–5 years	165	53.9
6–10 years	89	29.1
11–15 years	34	11.1
16–20 years	6	2.0
21+ years	12	3.9
Firm Size
Micro (1–9 employees)	144	47.1
Small (10–49 employees)	118	38.6
Medium (50–199 employees)	44	14.3

presents the demographic characteristics of the 306 respondents. The majority (66%) were male; 52.3% were 18–28 years old, indicating a predominantly young respondent profile. Most (58.8%) had at least a bachelor’s degree, and more than half (53.9%) had operated their business for five years or less. Regarding firm size, 47.1% of the respondents were micro-enterprises (19 employees), 38.6% were small companies (10–49 employees), and 14.3% were medium companies (50–199 employees).

Figure 2 provides a visual summary of the demographic distribution.

4.2. Measurement Model Assessment

4.2.1. Internal Consistency Reliability

Table 5. Internal consistency reliability results.

Construct	Indicator	Loading	$\mathit{\alpha }$	CR
AI Adoption Intention (AIAI)	AIAI1	0.815	0.76	0.85
	AIAI2	0.749
	AIAI6	0.757
	AIAI7	0.719
Personal Attributes (PA)	PA7	0.761	0.69	0.83
	PA8	0.758
	PA9	0.842
Strategic Capabilities (SC)	SC2	0.819	0.76	0.86
	SC3	0.841
	SC5	0.800
Interpersonal Capabilities (IC)	IC11	0.778	0.80	0.87
	IC12	0.780
	IC13	0.836
	IC14	0.762
Delivery-Related Capabilities (DRC)	DRC15	0.752	0.84	0.88
	DRC16	0.725
	DRC17	0.774
	DRC18	0.771
	DRC19	0.734
	DRC20	0.720
Org. Innovation Climate (OIC)	OIC1	0.804	0.78	0.86
	OIC2	0.791
	OIC3	0.768
	OIC4	0.742

α = Cronbach’s alpha; CR = composite reliability. All loadings exceed 0.70.

presents the reliability results. All constructs demonstrated satisfactory internal consistency based on established thresholds: Cronbach’s alpha values ≥ 0.70 (with personal attributes at 0.69, acceptable in exploratory contexts [63]), composite reliability values ≥ 0.70, and all factor loadings exceeding 0.70 [60].

4.2.2. Convergent and Discriminant Validity

Convergent validity was confirmed as all AVE values exceeded 0.50 (

Table 6. Convergent validity results (average variance extracted).

Construct	AVE
AI Adoption Intention	0.58
Personal Attributes	0.62
Strategic Capabilities	0.67
Interpersonal Capabilities	0.62
Delivery-Related Capabilities	0.56
Org. Innovation Climate	0.60

), indicating that each construct explains more than half of the variance in its indicators [65].

Discriminant validity was assessed using the Fornell–Larcker criterion (

Table 7. Fornell–Larcker criterion results.

	AIAI	PA	SC	IC	DRC	OIC
AIAI	(0.761)
PA	0.476	(0.788)
SC	0.573	0.615	(0.820)
IC	0.548	0.500	0.569	(0.789)
DRC	0.542	0.625	0.661	0.753	(0.746)
OIC	0.564	0.512	0.621	0.583	0.549	(0.775)

Diagonal values in parentheses = $\sqrt{\mathrm{AVE}}$; off-diagonal = inter-construct correlations.

), confirming that the $\sqrt{\mathrm{AVE}}$ of each construct exceeds its correlations with all other constructs [65]. The HTMT analysis (

Table 8. Heterotrait–monotrait (HTMT) ratio results.

	AIAI	PA	SC	IC	DRC	OIC
AIAI	—
PA	0.661	—
SC	0.756	0.850	—
IC	0.703	0.675	0.733	—
DRC	0.677	0.822	0.828	0.875	—
OIC	0.718	0.694	0.789	0.741	0.683	—

All HTMT values fall below the 0.90 threshold.

) further confirmed discriminant validity, with all values below 0.90 [66].

4.3. Structural Model Assessment

4.3.1. Direct Effects: Path Coefficients and Hypothesis Testing

Table 9. Direct effects: path coefficients, t-statistics, effect sizes, and hypothesis testing.

Path	$\mathit{\beta }$	t	p	${\mathit{f}}^2$	Decision
$H_1$: SC → AIAI	0.298	5.459	<0.001	0.171	Supported
$H_2$: DRC → AIAI	0.090	1.589	0.057	0.049	Not Supported
$H_3$: IC → AIAI	0.245	4.453	<0.001	0.134	Supported
$H_4$: PA → AIAI	0.129	2.304	0.011	0.062	Supported
SC → OIC	0.383	7.143	<0.001	0.238	Significant
OIC → AIAI	0.198	3.612	<0.001	0.112	Significant

$f^2$ effect sizes: 0.02 = small; 0.15 = medium; 0.35 = large [64]. R² (AIAI) = 0.418; R² (OIC) = 0.385; Q² (AIAI) = 0.412; Q² (OIC) = 0.379.

presents the analysis of the path coefficients, and Figure 3 illustrates the results of the structural model.

Beyond statistical significance, the relative importance of predictors is informative. The ranking of the standardised coefficients and effect sizes places strategic capabilities first ($\beta$ = 0.298, $f^2$ = 0.171, medium), interpersonal capabilities second ($\beta$ = 0.245, $f^2$ = 0.134, small-to-medium), and personal attributes third ($\beta$ = 0.129, $f^2$ = 0.062, small), with delivery-related capabilities not reaching significance. The four dimensions of leadership and the innovation climate path jointly explain a substantial share of variance in adoption intention (R² = 0.418), which in practical terms means that the abilities of the owner-manager and the climate factors that are developed through training account for a large part of why some firms intend to adopt AI and others do not. The gap between the strategic and delivery coefficients (0.298 versus 0.090) is itself practically consequential: it implies that an SME-support programme that raises strategic-cognitive capability is likely to move adoption intention several times more than one of equal intensity targeting operational execution.

To probe the non-significant delivery-related effect, two robustness checks were performed. First, the model was re-estimated with delivery-related capabilities entered alone (without the competing strategic and interpersonal paths); the coefficient rose modestly but remained non-significant ($\beta$ = 0.121, p = 0.071), indicating the null result is not merely an indication of multicollinearity among capability dimensions (full-collinearity VIFs were all below 3.3). Second, splitting the sample by adoption stage showed that the delivery effect was uniformly weak between firms with lower and higher readiness. The most plausible interpretation is therefore substantive rather than methodological: at the pre-adoption intention stage, execution capacity is latent rather than activated, so its influence is muted until firms move into implementation.

4.3.2. Mediation Analysis

The mediating role of the organisational innovation climate ($H_5$) was tested using bootstrapped indirect effects and the Sobel test.

Table 10. Mediation analysis results: indirect effects through organisational innovation climate.

Indirect Path	Indirect $\mathit{\beta }$	95% CI	p	Mediation Type
SC → OIC → AIAI	0.076	[0.031, 0.128]	0.003	Partial
IC → OIC → AIAI	0.048	[0.009, 0.096]	0.024	Partial
PA → OIC → AIAI	0.032	[−0.008, 0.074]	0.112	None
DRC → OIC → AIAI	0.021	[−0.015, 0.059]	0.234	None

CI = confidence interval (bias-corrected bootstrapping, 5000 resamples). Partial mediation: both direct and indirect effects are significant.

presents the results.

The results indicate that the organisational innovation climate partially mediates the relationship between strategic capabilities and AI adoption intention (indirect $\beta$ = 0.076, p = 0.003) and between interpersonal capabilities and AI adoption intention (indirect $\beta$ = 0.048, p = 0.024). The total effect of strategic capabilities on the intention to adopt AI is thus 0.374 (direct 0.298 + indirect 0.076), reinforcing its position as the strongest predictor. Hypothesis $H_5$ is partially supported.

4.3.3. Moderation Analysis

The moderating effect of firm size ($H_6$) was tested by analysing the interaction term (

Table 11. Moderation analysis: firm size as moderator.

Interaction Path	${\mathit{\beta }}_{\mathbf{interaction}}$	p	Decision
SC × Firm Size → AIAI	0.067	0.124	Not Significant
DRC × Firm Size → AIAI	0.038	0.268	Not Significant
IC × Firm Size → AIAI	0.109	0.029	Significant
PA × Firm Size → AIAI	0.054	0.178	Not Significant

). The company size was coded as a categorical variable (1 = micro, 2 = small, 3 = medium) and introduced as a moderator for each digital leadership capability path.

The size of the company significantly moderates the relationship between interpersonal capabilities and the intention to adopt AI (${\beta }_{\mathrm{interaction}}$ = 0.109, p = 0.029), with the effect of interpersonal capabilities being stronger in medium-sized enterprises than in micro-sized enterprises. Hypothesis $H_6$ is partially supported.

5. Discussion

This study developed and tested a comprehensive model that examined how four dimensions of digital leadership capabilities, strategic, delivery-related, interpersonal, and personal attributes, influence the intention to adopt AI among Nigerian SMEs, with the organisational innovation climate as a mediator and the size of the firm as a contextual moderator. The findings advance both theoretical understanding and practical knowledge regarding the leadership antecedents of AI adoption in SMEs in developing economies.

5.1. Direct Effects of Digital Leadership Capabilities

The strongest predictor of AI adoption intention is strategic capabilities ($\beta$ = 0.298, p < 0.001, $f^2$ = 0.171), a finding consistent with Hossain et al. [9], who demonstrated through a dynamic managerial capability framework that leaders who align technological initiatives with strategic goals significantly drive AI adoption. This finding is further corroborated by Yu et al. [8], who established that the digital backgrounds of executives facilitate the adoption of corporate AI, and by Mahmood et al. [41], who found that strategic digital leadership amplifies the positive effects of AI on organisational performance in Pakistan. The medium effect size ($f^2$ = 0.171) indicates that strategic vision, change-management competence, and innovation orientation provide substantive cognitive and organisational foundations for SME leaders to recognise AI’s potential and commit to its adoption.

Interpersonal skills emerge as the second-strongest predictor ($\beta$ = 0.245, p < 0.001, $f^2$ = 0.134), consistent with Kumari et al. [48], who argued that effective digital-era leadership requires collaboration and shared influence to foster innovation, and with Brunner et al. [46], who demonstrated strong correlations between relational leadership competencies and digital transformation outcomes. Yang et al. [50] further showed that digital leadership enhances creativity through knowledge sharing, a mechanism consistent with the interpersonal pathway to AI adoption observed in this study. Leaders who build networks, foster psychological safety, and promote team collaboration create organisational environments conducive to the adoption of novel technologies.

Personal attributes demonstrate a positive though more modest influence ($\beta$ = 0.129, p = 0.011, $f^2$ = 0.062), which corroborates Mikalef et al. [54], who found that openness, adaptability, and digital self-efficacy enhance the readiness of leaders to technology, and Alghamdi [51], who demonstrated that personal digital literacy positively influences attitudes towards AI adoption. The size of the small-to-medium effect suggests that personal attributes serve as facilitating conditions rather than the primary drivers of adoption intention.

The insignificant effect of delivery-related capabilities ($\beta$ = 0.090, p = 0.057, $f^2$ = 0.049) is a noteworthy finding that contrasts with Sony et al. [44], who emphasised operational competence in implementing between pre-adoption intention and post-adoption implementation: adoption intention is primarily a cognitive-strategic phenomenon, shaped by a leader’s vision, relational influence, and adaptive capacity rather than by execution-oriented skills. This interpretation is consistent with Arroyabe et al. [47], who found that digital maturity and innovation capability, not operational competence, were the primary enablers of adoption among EU SMEs. Within the DOI framework, Rogers’ emphasis on perceived enablers of adoption characteristics and individual innovation as determinants of adoption intention, rather than implementation capacity, further supports this finding [43].

5.2. Mediation Through Organisational Innovation Climate

The partial mediation of the organisational innovation climate between strategic capabilities and AI adoption intention (indirect $\beta$ = 0.076, p = 0.003) and between interpersonal capabilities and AI adoption intention (indirect $\beta$ = 0.048, p = 0.024) offers important mechanistic insights. Leaders with a strong strategic vision and collaborative orientation cultivate innovation-supportive environments that, in turn, enhance organisational receptivity to AI technologies. This finding extends to Yansen and Yujie [56], who demonstrated that transformative digital leadership fosters multi-dimensional innovation by identifying the organisational innovation climate as a specific mediating mechanism. It also aligns with Sarkis and Pallotta [57], who found that AI adoption challenges simultaneously promote innovation-supportive organisational competencies. From a practical standpoint, this suggests that developing digital leadership capabilities alone may be insufficient; organisations must concurrently cultivate innovation-supportive climates to maximise the translation of leadership competencies into AI adoption readiness.

5.3. Moderation by Firm Size

The significant moderating effect of firm size on the relationship between interpersonal capabilities and AI adoption intention (${\beta }_{\mathrm{interaction}}$ = 0.109, p = 0.029) indicates that the impact of relational leadership competencies on adoption intention is amplified in medium-sized enterprises. This finding is consistent with Badghish and Soomro [17], who found stronger AI adoption–performance relationships in medium-sized Saudi Arabian enterprises, and with the resource-based logic that larger SMEs possess greater absorptive capacity, more formalised communication structures, and broader talent pools that enable interpersonal leadership competencies to operate more effectively [59]. A plausible explanation for why size specifically moderates the interpersonal path, but not the strategic, personal, or delivery paths, lies in the social mechanics of relational leadership: communication, coaching, and coalition-building yield returns only when there are enough organisational members and structural layers to act upon. In a micro-firm of a few employees, the owner’s interpersonal competence has a thin social field in which to operate, so its marginal effect on collective adoption intention is small; as the firm grows into the medium-sized range, the same competence governs a larger, more differentiated team and thus exerts a stronger effect. Strategic vision and personal adaptability, by contrast, shape the owner-manager’s decision calculus regardless of headcount, consistent with the size-invariant effects observed here.

5.4. Theoretical Implications

These findings contribute to the theory in several ways, extending some positions while challenging others. First, the study extends the DOI theory and the Tigre–Henriques–Curado digital leadership framework to a developing-economy SME context, demonstrating their explanatory utility in a non-Western setting and answering calls to test Western-derived leadership constructs in African organisations. Second, by identifying the organisational innovation climate as a partial mediator, it advances the literature beyond direct leadership, adoption associations, and specifies how individual competence becomes a collective adoption decision, a transmission account consistent with the view that leader-level resources require organisational conversion to influence firm outcomes [14]. Third, the non-significant delivery-related effect challenges the common assumption that operational competence drives technology uptake; our results instead support a strategy-before-execution logic in which cognitive and relational capabilities govern the pre-adoption stage, while execution capacity becomes salient only at implementation [45]. Fourth, the firm size moderation findings refine contingency perspectives by showing that digital leadership’s effect is not uniformly context-dependent but selectively so—confined here to the relational dimension. Finally, situating these mechanisms within the sustainability agenda extends the contribution beyond efficiency: where digital leadership initiates the adoption of AI among resource-constrained, production-oriented firms, it also conditions their capacity for green innovation and sustainable transition, a linkage that is especially pronounced for traditional manufacturing and resource-limited organisations of the kind that dominate the present sample [18].

6. Conclusions

This study examined the influence of four dimensions of digital leadership capability on the intention to adopt AI among Nigerian SMEs, integrating the organisational innovation climate as a mediator and the size of the firm as a moderator within a PLS-SEM framework. Based on 306 valid responses, the findings lead to several conclusions. First, strategic capabilities are the strongest driver of the intention of adopting AI, followed by interpersonal capabilities and personal attributes. Delivery-related capabilities, while positively associated, do not significantly predict the intention to adopt. This hierarchy demonstrates that cognitive, strategic, and relational leadership dimensions outweigh operational competencies in shaping pre-adoption decisions. Second, organisational innovation climate partially mediates the influence of strategic and interpersonal capabilities on AI adoption intention, revealing that leadership’s effect on adoption is amplified when organisations cultivate innovation-supportive environments.

Third, firm size moderates the interpersonnel relationship between onal capabilities, AI adoption, and intention relationdium-sized enterprises showing stronger effects, indicating that relational leadership competencies operate more effectively in organisations with sufficient structural complexity to benefit from enhanced communication and collaboration.

These findings translate into segment-specific rather than generic guidance. For micro-enterprise owners (fewer than 10 employees), the priority is strategic-cognitive: because relational competence has a thin social field at this scale, owners should first build a clear understanding of AI’s concrete business value and a feasible adoption plan before investing in execution tooling, avoiding blind expenditure on operational digital capability that the firm cannot yet exploit. For small firms, the emphasis shifts towards pairing that strategic cognition with the early formalisation of an innovation-supportive climate. For medium-sized firms, where the interpersonal pathway is strongest, managers should emphasise communication, coaching, and team cooperation to forge organisational consensus around AI adoption. For policymakers and SME-support agencies, capacity-building should be hierarchical and differentiated: training for micro-firms should emphasise AI’s commercial value and transformation planning; courses for medium firms should centre on team collaboration and innovation climate construction; and programmes should be regionally calibrated to local digital infrastructure conditions, so that better-resourced states accelerate application while less-developed states first build owner-manager digital cognition.

This study acknowledges several limitations that guide future research. First, the cross-sectional design precludes causal inference and cannot rule out reciprocal effects among leadership, climate, and adoption intention; longitudinal panels that track how leadership capabilities shape AI adoption trajectories over time, or quasi-experimental designs, would establish temporal ordering. Second, although the sample captures Nigeria’s most commercially active SME ecosystem, it is drawn from a single geopolitical zone, so cross-zone and cross-country comparative studies are needed to test the generalisability of the leadership–adoption mechanism across differing institutional and infrastructural settings. Third, future work should explore additional mediating mechanisms (e.g., technology readiness, organisational learning culture) and moderating variables (e.g., industry sector, founder digital experience, leader educational background) that may further explain the leadership–adoption nexus in developing economies. Finally, as AI adoption matures, the agenda should extend from technology adoption to human-centric system design, examining how SMEs can integrate AI in ways that keep human well-being, sustainability, and resilience central: the orientation of the emerging Industry 5.0 paradigm [67] and the work linking Industry 5.0 with green supply chain management for sustainable development in resource-constrained economies [68].