3. METHODOLOGY

In this study, a cross-sectional explanatory survey design was employed to test the hypothesised relationships among the adoption, automation, and ERP integration of CAIS and measurable performance outcomes of UAE SMEs, informed by the resource-based view, which posits that the adoption of CAIS, its automation, and integration with ERP systems are resources that lead to performance outcomes (Cho & Linderman, 2020). The information is not experimental and self-reported; therefore, it should be understood as correlational, not causal. The population consisted of UAE Small and Medium Enterprises (SMEs) that are using CAIS for accounting and financial management nowadays. To guarantee that respondents had first-hand experience with CAIS, the sampling method adopted was purposive and non-probability type, which is typical in information systems adoption research (Hojnik & Huđek, 2023). The eligible firms were those that met the official UAE SME classification and had a CAIS for at least three years, and respondents’ positions required them to examine the capabilities linked to the CAIS, which included accounting, financial reporting, or managerial decision-making roles (Putro, 2023). Firms that used only on-premises systems and respondents who did not have financial decision-making responsibilities were excluded, and screening questions at the beginning of the questionnaire verified eligibility.

Data were collected using a pre-designed online questionnaire that was circulated through SME associations, professional networks, and digital business platforms in the UAE. A total of 400 questionnaires were sent, of which 330 were completed and found to be valid after checking for completeness, consistency, and eligibility. The respondents ranged from micro to small and medium-sized firms across different sectors, such as trading, manufacturing, services, logistics, and technology, from several emirates in the UAE. The results obtained cannot be considered statistically representative of the UAE SME population because of the non-probability sampling technique employed. However, the number of respondents selected (n = 330) was larger than the minimum sample size required based on a priori power analysis using G*Power software (α = 0.05, power = 0.80, and medium effect size).

The questionnaire (Appendix A) contained closed-ended questions on a Likert 5-point scale (1 = strongly disagree to 5 = strongly agree). The measures of CAIS adoption, extent of automation, and ERP integration were adapted from previously validated measures with minor context changes for UAE SMEs, and performance outcomes were operationalised as financial decision speed, reporting accuracy, and operational efficiency; items were structured to avoid conceptual overlap between predictors and outcomes. The questionnaire was reviewed by three academic experts with experience in accounting information systems and SME digitalisation to establish face and content validity before data collection.

Potential common method bias (CMB) was minimised and addressed as part of the data collection procedures as follows: i) the responses were anonymous, ii) the wording of the items was kept simple and unambiguous, and iii) the design of the instrument minimised patterns of socially desirable answers. According to the single factor test suggested by Harman, the first factor contributed less than 50% of the variance, and the VIF values were less than 3.3, indicating that there was no significant problem with CMB (Table 1). In all procedures, ethical conduct of research was observed with anonymous, informed, and voluntary participation. The data were analysed using the partial least squares structural equation modelling (PLS-SEM) method, which is suitable for predicting data using complex research models and small to medium samples (Pereira et al., 2024). The analysis was separated into two parts: first, to test the measurement model and test for reliability and validity (Cronbach’s alpha, composite reliability, average variance extracted, and discriminant validity); second, to perform the structural path analysis, along with bootstrapping to estimate path coefficients, t-statistics, and p-values for the hypothesised relationships between CAIS adoption, automation, ERP integration, and the performance outcomes.

4. RESULTS

The demographic traits of the respondents are presented in Table 2. The statistics indicate diversity in age, gender, education, and professional experience. The demographic profile revealed that most of the respondents were in the ages of 31-35 years (40.91%), which implies that the sample was mature enough in its professional activities to provide valid information about the use of CAIS. The majority of the respondents were male (69.70%), and many of them had graduate or postgraduate degrees (86.66%), which is a highly educated sample. Regarding experience, half of them had between 6-10 years of experience which gave them good exposure to financial and accounting practices. These factors ensured the reliability, relevance, and extent to which the data collected were broadly representative of the SMEs that had embraced CAIS; however, a purposive sampling approach prevents claims of representativeness.

Measurement Model was evaluated to determine the reliability and validity of the constructs (Table 3). Factor loadings, Cronbach’s alpha, composite reliability, and average variance extracted (AVE) were analysed to ascertain that the construct measures were accurate. Factor loadings were all above 0.700, which indicates that individual items represented constructs very strongly. The Cronbach’s alpha values ranged from 0.900 to 0.829, indicating good internal consistency. The composite reliability was above 0.800, and the AVE was above 0.700 which proved convergent validity. These findings indicate that there was reliability and validity in the measurement model which suggests a strong background for testing the structural association among variables.

Table 4 presents the evaluation of discriminant validity based on the Heterotrait-Monotrait (HTMT) ratio, which assesses the extent to which the constructs are empirically different. HTMT values below 0.850 indicate discriminant validity. All the HTMT values of the Automation Extent, CAIS Adoption, Financial Decision Speed, ERP Integration, Operational Efficiency, and Reporting Accuracy constructs were less than 0.850, which indicates satisfactory discriminant validity. This proves that every construct includes a distinct attribute of CAIS capabilities and performance outcomes, and that multicollinearity among the latent variables is unlikely to be an issue in the measurement model.

The proposed relationships among the adoption of CAIS, automation, and ERP integration (independent variables) and financial decision speed, accuracy of reporting, and operational efficiency (dependent variables) were empirically examined using PLS-SEM (Table 5). The extent of automation had a significant positive impact on operational efficiency (β = 0.209, p < 0.01) and accuracy of reporting (β = 0.157, p < 0.05), but not on financial decision speed (β = 0.060, p = 0.317). The adoption of CAIS positively impacted financial decision speed (β = 0.211, p < 0.01) but had no significant influence on operational efficiency (β = -0.084, p = 0.278) or reporting accuracy (β = -0.124, p = 0.140). ERP integration significantly and positively influenced all three performance indicators: financial decision speed (β = 0.485, p < 0.01), operational efficiency (β = 0.395, p < 0.01), and reporting accuracy (β = 0.440, p < 0.01).

Note: *** indicates significance at 1%, ** indicates significance at 5%

The explanatory power of the model is presented in Table 6. The model accounted for 44.8%, 21.9 %, and 20.6% of the financial decision speed variance, operational efficiency variance, and 20.6% of the variance in reporting accuracy variance, respectively.

The f-square results show the relative sizes of the predictors on the dependent variables (Table 7). ERP integration had the most significant effect. The effect on financial decision speed (f² = 0.253) is large, while the effects on reporting accuracy (f² = 0.145) and operational efficiency (f² = 0.119) are small-to-medium, collectively pointing to the critical importance of ERP integration as a predictor. CAIS adoption showed small effect sizes despite its statistical significance. The extent of automation has insignificant to weak impacts, especially on the speed of financial decisions (0.005). In general, ERP integration has the strongest impact, and automation and CAIS have a minor effect on enhancing organisational and performance outcomes.

The Q² prediction values showed a high predictive relevance of the model (Table 8). The speed of financial decision-making (0.428) has high predictive relevance, whereas operational efficiency (0.193) and reporting accuracy (0.176) have moderate predictive relevance. Overall, the model has a satisfactory to moderate predictive value for the three outcomes, especially the speed of financial decision, which could facilitate its use in practice.

The results of the hypotheses are shown in Table 9. 9 hypotheses were tested, of which 6 were supported and 3 were not.

5. DISCUSSION

The results suggest that the capabilities of CAIS do not necessarily lead to uniform outcomes; rather, effectiveness varies depending on the extent and depth of integration of CAIS capabilities into organisational procedures. This is in line with dynamic capability theory (Bleady et al., 2018), which holds that performance benefits from CAIS resources occur when the resources are adapted, integrated, and reconfigured within the firm, and not just acquired.

As in H1, the financial decision speed significantly increased with CAIS adoption, confirming previous findings that cloud systems accelerate managerial response time by allowing access to financial data in real time and eliminating the need for manual financial consolidation (Rahman & Hossain, 2024; Hojnik & Huđek, 2023). This advantage of adoption may be particularly significant in the UAE, as the regulatory intensity and compliance requirements are substantial. No significant differences were found in H2 (improved reporting accuracy) or H3 (operational efficiency), indicating that the main value of adoption lies in access to information, not the change in underlying processes. Integrated systems, employee training, and automated controls do not seem to be mutually exclusive and are likely required to achieve true accuracy and efficiency (Ayoub & Sopuru, 2026). When adopted superficially, with manual controls or parallel record-keeping, they may not result in accuracy or efficiency improvements (Putro, 2023). This aligns with the perspective that technology adoption alone is insufficient to achieve better reporting or operational outcomes (Sihotang et al., 2024).

The situation was different for automation: it had a significant positive impact on the accuracy of reporting and efficiency of operations (H5 and H6 supported) but not on the speed of financial decisions (H4 not supported). This aligns with the findings that automation enhances reporting accuracy by standardising processes and minimising human error (Kimani, 2024) and increases operational efficiency by reducing repetitive tasks and allowing resources to be allocated for more valuable activities (Ojika et al., 2022). It also shows that despite its limited effect on decision speed, automated outputs are still dependent on interpreting and analysing information, which automation does not supply, but instead may increase the reliability of the information (Al-Dmour, 2018; John, 2024).

The most widespread impact was in the area of ERP integration, which had a strong impact on all three performance outcomes (H7–H9 supported), reflecting evidence of ERP’s ability to synchronise financial and operational data across departments while also eliminating data fragmentation (Barna et al., 2021; Faccia & Petratos, 2021). However, ERP integration combines the features of adoption and automation, and it seems to yield greater performance benefits overall than the adoption or automation features alone. This impact can be further amplified in the UAE through regulatory compliance pressure and digital transformation initiatives by the government, which are pushing for integration over independent systems (Hadouga, 2023) that can enhance cross-functional visibility and minimise process duplication.

Overall, these findings confirm that performance gains are more likely to be realised through the process of transforming and combining technological resources into a new configuration rather than simply owning them (Ginting et al., 2025), and reinforce the findings that depth and integration are more important than adoption or non-adoption of technology to realise performance gains (Alao et al., 2024).

The results have several implications for the practice of SME managers and policymakers but need to be viewed with caution because the data are cross-sectional and self-reported. To maximise the value of these findings for SME managers, investments in managerial digital skills may be warranted to enable them to understand ERP outputs that can then be used to guide timely decision-making (Nofel et al., 2024; Rakibuzzaman et al., 2025). Policy measures can facilitate this transition by implementing targeted training initiatives and incentives for digital infrastructure, such as supporting ERP integration and promoting more extensive use of CAIS in line with the UAE’s overall digital transformation and economic diversification goals.