Evaluating the Effectiveness of Digital Bookkeeping Applications for the UK Small and Medium Enterprises: A User-Centric Approach with the Mediating Role of User Satisfaction

2.4. Literature Gap

Even though there is a growing amount of research on digital accounting and bookkeeping systems, there are still a number of key gaps that are apparent in the present literature. First, many of the previous studies have been predominantly centered on the topic of technology adoption, intention to use, or continuance intention (Kholid & Asri, 2021), as opposed to considering perceived effectiveness as a result, according to the perspective of the user. The studies that do address performance or decision-making outcomes tend to assume that the outcomes are the direct results of using technology without unraveling the psychological mechanisms by which these outcomes are produced (Almgrashi, 2025; Kurniawan et al., 2025). SME literature that focuses on digital accounting, confirms user satisfaction to be essential in post-adoption (Al-Hattami & Almaqtari, 2023; Imtihan et al., 2025; Vărzaru, 2022). But user satisfacation is not studied in bookkeeping setterig for its mediating effect on perceived ease of use and perceived usefulness and effectiveness outcomes. Lastly, the available empirical data is heavily systematised in non-UK contexts (e.g., Indonesia, Jordan, Bangladesh, and South Africa), which restricts contextual transferability in the UK, with a unique regulatory framework and high usage of HMRC-developed digital compliance resources. This study filled these gaps by having a post-adoption and user-centric focus, where directly and indirectly associated constructs (perceived effectiveness, user satisfaction, perceived ease of use, and perceived usefulness) were analyse from UK SMEs sample. The methodology contributes to the theoretical rigor of TAM and gives practically applicable information about the way digital bookkeeping success can be realised in practice, as opposed to being simply adopted.

3. METHODOLOGY

3.1. Research Design

(Thomas & Zubkov, 2023) advocated quantitative design for empirical study especially hypotheses testing which this study employed. The method is suitable for the current study since it attempts to measure the associations between user perceptions of a bookkeeping application and outcome variables like perceived effectiveness and user satisfaction, thus allowing the testing of hypotheses and statistical analysis.

3.2. Data Collection Tool

For data collection a closed-end questionnaire was developed through Google Form with likert scale. The questionnaire included items that measured perceived effectiveness, user satisfaction, perceived ease of use, and perceived usefulness (Appendix A). All of these items were conceptualised as a reflection of the respondents’ evaluative perceptions towards their firm’s use of bookkeeping application.

Perceived usefulness was evaluated using questions that evaluated how much the bookkeeping application enhances financial accuracy, regulatory compliance, decision-making, and the overall financial performance of managing the financial management. Perceived ease of use was used as the measure of the simplicity and the usability of the application interface and functions, and user satisfaction was used as the measure of the user experience and trust in the system. The perceptions of effectiveness were quantified based on operational results which were related to bookkeeping efficiency, financial process control, information organisation, and effectiveness of activities of day-to-day financial management. The distribution of the survey was through Google form, which, as per (Kumar, 2021), is a fast and cost-effective method.

3.3. Population, Sampling Technique, and Sample Size

The population of this study were UK’s small and medium-sized enterprises (SMEs), where employees are not more than 250 employees (European Commission, 2024). SMEs represent more than 99% of the UK business population and play a significant role in the overall employment and economic performance of the country, and therefore, their context is highly pertinent when studying the perception of technology in the actual business setting (Hart et al., 2020). The sampling strategy used was purposive, in which the researcher selected the respondents based on their respective organisational position, like owners, managers, accountants, and finance officers, and their direct exposure to the bookkeeping practices. This strategy is used to make sure that respondents are qualified and have the responsibility to make decisions that are informed and valid (Tajik et al., 2025), and it is specifically suitable when the focus of the research is on particular knowledge holders instead of the general population. Nonetheless, LinkedIn-based purposive sampling procedure can bias digitally mature SMEs and companies that are already active participants in online professional networks.

The data were gathered through LinkedIn invitations, which provided 545 responses, with 500 being complete and usable responses after the screening. This is larger than the recommended minimum sample size, with G*Power indicating at least 100 respondents to identify a medium effect size (0.5) with power of 0.80 and significance of 0.05 probability (Kang, 2021). In such a way, the final sample of 500 is sufficient in terms of statistical power and improves the strength and external validity of the results. Moreover, statistical power analysis, i.e., minimum R2-based techniques, underlies the rationale of using larger samples in PLS-SEM, specifically in models that imply mediation and latent constructs (Hayes, 2021).

3.4. Data Analysis

PLS-SEM or Partial Least Squares Structural Equation Modelling method was used to analyse the data collected. As per (Memon et al., 2021), it is suitable to predict and explain models with the use of latent variables. PLS-SEM is favoured where a model structure is complicated, non-normal data distribution, and mediation effects are involved, and it is, therefore, suitable for the goals of this research (Ahmed et al., 2024).

It was selected because it can test both direct and indirect associations. The constructs were all modelled as reflective because all indicators were supposed to reflect the underlying latent variable and not constitute it. This specification is in line with the conceptualisation that the responses to the survey are an expression of respondents’ perceptions of the bookkeeping application in terms of being useful, easy to use, satisfying, and efficient, which can reliably be estimated and interpreted in the PLS-SEM framework.

3.5. Common Method Bias

Common method bias (CMB) was checked to reduce research bias and for that Harman’s single-factor test was conducted. The test uses exploratory factor analysis to confirm if the measurement items’ variance explained by a single factor is less than 50%, showing not significant bias (Howard et al., 2024). In the given research, the initial factor contributed to the total variance by 38.7%, which proved that the issue of CMB did not pose a serious problem. The assessment of non-response bias was conducted through the comparing of early and late respondents using independent sample tests, adhering to the accepted practices, including those suggested by (Armstrong & Overton, cited in France et al., 2023). According to survey research literature, the average response rate of online surveys is 15-20%, and it is important to measure non-response bias when there is a possibility that respondents are not representative of non-respondents (Bhattacherjee, 2019). Early and late responses of 30 respondents were analysed and the constructs showed no significant differences from the rest of the respondents. This supports that the results were not impacted by the non-response bias significantly. Moreover, another approach to ensure that CMB is not a significant issue is to confirm that the full collinearity VIF values of all constructs are above 3.3. Table 2 shows that all the values fulfil this criterion, thus CMB is not a serious issue here.

Table 2. Collinearity statistics (VIF).

3.6. Ethical Consideration

Lastly, the study was conducted with ethical consideration. These involved informed consent, voluntary participation, anonymity and confidentiality, and secure data handling procedures to protect the study participants and the integrity of the research. Participants were informed about the purpose of the study, its procedures, and their rights in a clear way beforehand. Transparency and autonomy are key best practices in online survey research (Kang & Hwang, 2023; Stokes, 2025). No personal identifiable data was gathered, and all answers were stored safely to protect privacy, as required by the existing ethical standards in conducting internet-based research. All these actions served to adhere to the ethical guidelines of research involving human subjects.

4. RESULTS

Table 3 presents the demographic characteristic of the 500 respondents, which reveals a balanced age distribution with the highest percentage of 35-44 (31.2%) and 25-34 (28.4%), indicating that the users are of the mid-career SMEs segment. The proportion of males (55.2%) is a bit more than that of females (39.6%), and a minor percentage are non-binary or do not state it. The majority of the respondents are owners (37.6%) and managers (20.4%). The leading sectors are Services (33.6%) and Retail (24.8%). Most have been using digital bookkeeping solutions for 1-3 years (37.2%), followed by 4-6 years (30.8%) and more than 6 years (17.2%), which shows consistent use of digital solutions.

Table 3. Demographic analysis.

The results of the measurement model are introduced in Table 4, and they are calculated with the help of measurement model. The measurement model is used to assess the reliability of indicators, internal consistency, as well as convergent validity, which are essential components of construct measurement in SEM (Hair et al., 2021). All constructs in this research satisfy these thresholds, and thus it is clear that the items are good measures of perceived effectiveness, ease of use, usefulness, and user satisfaction, which attests to the quality of the measurement.

Table 4. Measurement model assessment.

Note: All constructs meet thresholds; Factor Loadings ≥ 0.70; CR ≥ 0.70; AVE ≥ 0.50 (Hair et al., 2021).

4.1. Discriminant Validity

Table 5 shows the results of discriminant validity based on the HTMT criterion, which as per (Henseler, 2026), examines whether constructs are empirically different in the model of measurement. Discriminant validity is defined as the degree to which theoretically distinct constructs are practically dissimilar (Afthanorhan et al., 2021). The value of HTMT that is below a conservative level of 0.85 would suggest sufficient discriminant validity among constructs (Cheung et al., 2023). Table 5 presents that all of the HTMT values are not more than 0.85, which support that user satisfactiom, perceived usefulness, and perceived ease of use are not similar constructs and, thus, the interpretation of relationships in the model can be considered structural.

Table 5. Discriminant validity.

Note: HTMT values < 0.85

4.2. Path Coefficient

Table 6 displays the results of the structural model and gives the path coefficients, t-statistics, p-values as well as effect sizes (f²) to test the hypotheses proposed. The results indicate differences in the extent of the explanatory power of the constructs; were user satisfaction and perceived usefulness exhibit stronger effects than perceived ease of use.

Table 6. Path coefficient.

Note: Significant paths are indicated using asterisks: *** = p < 0.01 (1%).

The findings show that the perceived usefulness significantly and positively correlates with the perceived effectiveness (β = 0.176, t = 2.726, p = 0.006, f² = 0.032). This indicates that SMEs who consider digital bookkeeping applications useful are more apt to consider them helpful in facilitating the bookkeeping and financial management processes. The relationship is statistically significant though the effect size is small. Thus, H1 is accepted.

The results demonstrate that, there is no significant direct association between perceived ease of use and perceived effectiveness (β = 0.080, t = 1.350, p = 0.177, f² = 0.008). This implies that the ease of use is not a conclusive factor in determining the effectiveness of digital bookkeeping applications in the UK SMEs. There is also a negligible effect size which means that the explanatory contribution is weak. Thus, H2 is rejected.

User satisfaction showed significant mediation effect on the association between the perceived usefulness and the perceived effectiveness (β = 0.267, p = 0.001). In addition, perceived usefulness significantly influences user satisfaction (β = 0.522, t = 10.235, p = 0.001, f² = 0.347), while user satisfaction significantly predicts perceived effectiveness (β = 0.512, t = 8.690, p = 0.001, f² = 0.285). The relationships are both significant, and thus, the mediation is partial. Because the perceived usefulness and effectiveness are directly and indirectly associated through a mediator. Thus, H3 is accepted.

The findings also suggest that user satisfaction is a strong mediator between perceived ease of use and perceived effectiveness (β = 0.101, p = 0.001, t = 3.290). The relationship between perceived ease of use and perceived effectiveness is not significant (β = 0.198, t = 3.539, p = 0.001, f² = 0.052) but the relationship between perceived ease of use and user satisfaction is significant (β = 0.198, t = 3.539, p = 0.001, f² = 0.052). Since the direct association between perceived ease of use and perceived effectiveness was not significant, this mediation through user satisfaction is complete. H4 is thus accepted.

Combined, the findings suggests that in the UK SMEs, conversion of user perception for digital bookkeeping applications into perceptions of effectiveness is mediated by user satisfaction.

4.3. Model Explanatory Power and Predictive Relevance

Table 7 presents the explanatory power of the model through the values of R-squared, which (Hair et al., 2021) defined as the degree to which the exogenous constructs account for endogenous constructs variance. The greater the R-squares, the better the explanatory power. In this research, the model moderately explains perceived effectiveness (0.463) and user satisfaction (0.416) whereby perceived usefulness and ease of use, mediated by a satisfaction, significantly explains the differences in effectiveness and satisfaction. Moreover, the Q-square value greater than zero shows that for endogenous constructs predictive capability exists in the model. The Q-square values in Table 7 show satisfactory predictive relevance.

Table 7. Model R-square and Q-square values.

5. DISCUSSION

H1: Perceived Usefulness and Perceived Effectiveness.

H1 is supported as the study found a positive and significant association between perceived usefulness and perceived effectiveness (β = 0.176, p = 0.006). The effect size, however, is low (f² = 0.032), which means that, as much as usefulness is an important factor in the perceptions of effectiveness, its contribution as an independent factor is low. This implies that SMEs might not be assessing the effectiveness of bookkeeping in terms of functional usefulness alone, but also using other experiential and contextual criteria. In the context of UK SMEs, digital bookkeeping applications are more probable to be viewed as effective in the case when they enhance the accuracy of financial data, adherence to regulations, and decision-making. The results are in line with (Ghani et al., 2022; and Kurniawan et al., 2025) who linked usefulness perceptions with better organisational performance in digital accounting settings.

Usefulness in this study is not studied as the motivation of the adoption for the technology but the evaluation of its performance post-adoption. SMEs operating within the scope of HMRC-related compliance demands might be more concerned with the reliability of the functioning and practical value following the incessant use of a system than with the technological novelty. This result answers the research question of whether digital bookkeeping systems are operationally effective once implemented, as opposed to adopted. One practical implication is that software vendors can gain by focusing on the output-based characteristics like accuracy of reporting, compliance support, and financial decision support to enhance long-term effectiveness perceptions among SMEs.

H2: Perceived Ease of Use and Perceived Effectiveness

H2 is rejected as the result showed non-significant association between perceived ease of use and perceived effectiveness (β = 0.080, p = 0.177). The effect size is also insignificant (f² = 0.008), meaning that ease of use has a minimal contribution to the explanation of the effect on the perception of the effectiveness independently. This means that usability might not necessarily have a direct effect on the effectiveness of digital bookkeeping applications as assessed by UK SMEs in post-adoption environments. One fact that might explain it is that a lot of SMEs that already work in digitally advanced surroundings already demand bookkeeping systems to meet the minimum standards of usability. Ease of use can thus cease to be a significant point of distinction when the technology is already implemented and used regularly.

The results are contrasted with others like (Begum & Begum, 2025; and Putri et al., 2025) which established direct usability-performance links in less developed adoption settings. The present results indicate that the perceptions of effectiveness in UK SMEs are more tightly linked to operational reliability, workflow integration, and practical outcomes than interface simplicity. It is important to note that this finding is not evaluation of initial acceptance behaviour but an evaluation of post-adoption technology. One of the practical implications is that organisations adopting bookkeeping software might have to integrate usability-enhancing features with workflow support, training, and integration to enhance the perceptions of operational effectiveness.

H3: Mediating Role of User Satisfaction between Perceived Usefulness and Perceived Effectiveness

H3 is supported as current results supported the mediator role of use satisfaction for perceived usefulness and perceived effectiveness relationship (β = 0.267, p = 0.001). Since the indirect and direct association are significant, the mediation is partial. Moreover, the perceived usefulness shows a significant impact on the user satisfaction (f² = 0.347), which proves usefulness as a critical predictor of the positive user experiences in the digital bookkeeping settings. This implies that SMEs consider bookkeeping applications as effective not just due to the functional value, but also due to the fact that useful systems promote confidence, trust and satisfaction in terms of continued use.

These results are in line with (Al-Hattami & Almaqtari, 2023; and Imtihan et al., 2025) who discovered that satisfaction reinforces the connection between system characteristics and post-adoption results. The partial mediation also indicates that usefulness has both direct evaluative and indirect experiential role via satisfaction. In the UK SMEs, beneficial bookkeeping functions can solidify on perceptions of reliability and regulatory support, which in turn intensifies effectiveness judgments. Considering that this study only investigated only UK SMEs through purposive sampling and LinkedIn recruitment, it is important to cautiously take the mediation effect. The practical implication is that companies that adopt bookkeeping systems must not only consider some of the functional utility of the software, but they must consider the satisfaction of users in the process of having long-term interaction with the system.

H4: User Satisfaction as a Mediator of the Relationship between Perceived Ease of Use and Perceived Effectiveness

The results confirm the H4 hypothesis since the user satisfaction lies between the perceived ease of use and perceived effectiveness (β = 0.101, p = 0.001). The perceived ease of use has a small effect size on user satisfaction (f² = 0.052) and so usability is a contributing factor to user satisfaction, but to a smaller degree than perceived usefulness. The mediation is complete since the direct correlation between perceived ease of use and perceived effectiveness is not significant. It means that ease of use can only add to the perceptions of effectiveness when it enhances the user experience and satisfaction in general.

These results are consistent with those of (Koerniawan et al., 2024), who found that satisfaction was a mediator between system features and SME performance outcomes and with (Vărzaru, 2022), who asserted that satisfaction is a significant post-adoption variable in the use of accounting technologies. Reducing frustration, effort in learning and reliance on external support may help to improve positive experiences in the UK SMEs, which in turn enhances perceptions of effectiveness. To a certain degree, these findings can be used to help solve the research problem by demonstrating that usability does not necessarily result in positive operational ratings unless it is supported by satisfactory user experiences. A practical suggestion is that organisations can leverage onboarding assistance, user documentation, and continuous support instead of just depending on the simplicity of the interface to enhance long-term system reviews.