Poor employee well-being exacts a heavy toll on organizations, manifesting in reduced productivity, increased absenteeism and higher turnover (22–24). Meta-analytic evidence reveals a robust positive correlation between well-being and job performance: employees reporting higher life satisfaction and positive affect consistently deliver better quality work and customer service (25–28). Conversely, chronic stress and disengagement undermine cognitive function and motivation, eroding both individual and team outcomes (29, 30). Gallup’s (31) global survey identifies low engagement and high stress as two of the greatest organizational challenges worldwide, with highly engaged teams outperforming their counterparts by up to fivefold on key business metrics (29, 32). Moreover, the reciprocal relationship between engagement and well-being means that interventions targeting one domain often benefit the other (33, 34).

Workplaces have become a critical battleground in the global struggle against rising mental-health challenges: the World Health Organization estimates that one in eight people suffers from a mental disorder, and depression alone now accounts for over 12 billion lost workdays and US $1 trillion in forgone productivity each year (7, 8, 35). Yet despite clear evidence that psychological well-being is even more predictive of job performance than physical health (36), HRM has historically focused on operational metrics and “hard” outcomes, leaving employee flourishing to chance or ad hoc wellness perks (19, 20). The result is striking: global employee engagement stagnated, and overall employee wellbeing declined, collectively costing economies some US$8.9 trillion annually, or 9 percent of global GDP (31). In South Africa, where this study’s intervention was tested, engagement and thriving rates are among the world’s lowest, and workplace stress contributes heavily to both absenteeism and turnover (31). Without a structured, well-being program embedded into core talent and leadership practices, organizations will continue to shoulder unsustainable human and economic costs. A targeted training intervention—co-developed with employees to reflect local cultural and organizational realities—is therefore essential not only to safeguard mental health but to unlock higher engagement, resilience, and long-term performance.

Well-being is a multifaceted concept with no universally accepted definition, varying across disciplines and theoretical perspectives (37, 38). There is currently no unified theoretical framework explaining how employee well-being influences work performance, and there is a lack of consensus on the fundamental definitions of both well-being and performance (39). Reviews have categorized well-being into multiple dimensions: hedonic well-being, life satisfaction, eudaimonic well-being, mental well-being, social well-being, and physical health (40–42). WHO (7) recognizes mental health as an integral part of well-being and a fundamental human right. However, the lack of a universal definition contributes to inconsistencies in conceptualization and measurement, making it essential to adopt a structured approach to studying workplace well-being (43).

The measurement of well-being is equally diverse, with at least 69 well-being scales identified across different disciplines (42). Another scoping review identified 109 workplace mental health instruments (44). Given this variability, this study focuses on three widely used dimensions of workplace well-being: life satisfaction, flourishing, and mental well-being. These are measured using the Satisfaction With Life Scale (SWLS), the Flourishing Scale, and the Warwick-Edinburgh Mental Well-being Scale (WEMWBS), respectively (40–42, 45). Together, these constructs evaluate well-being, integrating cognitive assessments of life, psychological resilience and broader personal development.

Beyond well-being, this study also considers work engagement and stress, which significantly impact workplace productivity and employee functioning. Engagement, measured using the Utrecht Work Engagement Scale (UWES), is consistently linked to higher job performance and reduced turnover (31). Similarly, stress, assessed through the Perceived Stress Scale (PSS), plays a crucial role in burnout, workplace efficiency, and mental health outcomes (46, 47). By integrating these measures, this study ensures a holistic assessment of workplace well-being, allowing for a nuanced understanding of how well-being interventions impact employees’ engagement, stress levels, and overall workplace experience.

The Good Life training program was developed as a structured, multi-component HR intervention aimed at enhancing workplace well-being, employee engagement, and stress management (see Figure 1). The design process was participatory and grounded in both theory and employee input. This participatory process unfolded over three phases: evidence-based construct selection via literature review, pilot testing with a diverse set of employees, and a focus group discussion to co-refine the content. For instance, ‘relationships’ was added and ‘forgiveness’ was retained based on strong participant endorsement during the focus group. This ensured the intervention aligned with both scientific evidence and workplace realities. The authors (who are also HR professionals) developed detailed operational guidance, learning activities, and evidence-based tools to ensure real-world applicability. The session structure and key theoretical anchors are summarized in Table 1. Unlike many theoretical models, The Good Life offers a practitioner-ready, scalable intervention that addresses the gap between psychological theory and Human Resource Development (HRD) practice, especially in low- and middle-income settings.

The theoretical framework presented in Figure 2 integrates key concepts from well-being science, HRM, and HRD to guide the design and delivery of the training intervention. It positions well-being as a multidimensional construct, developed through structured learning and embedded in organizational practice to enhance both individual thriving and firm-level performance.

At the center of the framework is the HR training program designed to address three key challenges facing modern workplaces: low engagement, high stress, and poor well-being. Grounded in both experiential learning and transformational learning theories, the intervention aims to create opportunities for employees to reflect on their lived experiences, question limiting beliefs, and adopt new, evidence-based strategies for improving well-being. This learning is not abstract but applied through real-world exercises and peer interactions that encourage behavioral change and deep personal insight.

The training intervention draws from multiple well-established psychological frameworks to support its design and outcomes. These include PERMA (48), the broaden-and-build theory (49), self-determination theory (SDT) (50), and the job demands–resources (JD-R) model (51). Together, these theories underscore the importance of positive emotions, intrinsic motivation, autonomy, and personal resources in driving engagement and resilience in the workplace.

Seligman’s (52) well-being formula—H = S + C + V— informs the HR program’s focus on voluntary activities (V), which represent an area where individuals can intentionally foster greater well-being. While set-point (S) and life circumstances (C) are relatively fixed, voluntary actions such as gratitude, mindfulness, social connection, and strength-based reflection can be shaped through targeted learning experiences. These activities are integrated into the HR training curriculum through experiential modules that allow employees to practice new behaviors, reflect on outcomes, and iterate on their strategies.

These voluntary activities are not isolated practices but are woven into the program’s learning architecture, fostering deeper change through transformational learning (53). Employees are encouraged to critically examine their assumptions about stress, work, and personal capabilities, shifting from low awareness to a high understanding of well-being and its applications at work. In parallel, experiential learning (54) is applied through real-time learning cycles that involve concrete experience, reflective observation, abstract conceptualization, and active experimentation.

Further, the program introduces employees to multiple constructs like building relationships, signature strengths and job crafting—that directly support the JD-R and SDT framework by enhancing personal resources. Teaching employees how to identify and apply their core strengths increases competence and fulfillment, while job crafting enables them to proactively reshape their roles to better align with their values and skills. These interventions not only mitigate the negative effects of job demands but also foster motivation, engagement, and psychological resilience.

The study adopts a pragmatic research philosophy (61, 62), recognizing that workplace well-being is shaped by multiple, context-dependent realities (ontology), and that knowledge emerges through an iterative blend of quantitative surveys and qualitative interviews (epistemology). It also embraces researcher values and ethical considerations as integral to the inquiry (axiology). By first measuring well-being outcomes statistically and then exploring participants’ experiences qualitatively, the explanatory sequential mixed-methods design ensures the training program is both evidence-based and responsive to employees’ lived contexts.

The study was conducted in a South African pharmaceutical firm employing 108 individuals, predominantly female and largely based in the sales department (Table 2). From this workforce, 95 employees (excluding HR and senior management) were invited to participate via non-probability convenience sampling. Of these, 87 gave informed consent and completed the pre-intervention survey. Participants were allocated into three distinct intervention cohorts: two classroom-based groups (one delivered over two full days and the other over four half-days across 4 weeks), and an online group delivered through four half-day sessions spread over 4 weeks, each comprising 16 contact hours in total. Initially, a control group of 15 participants was planned and kept on a waitlist for 6 months; however, this approach was subsequently discarded as control participants frequently interacted with the experimental groups, potentially biasing results. Moreover, significant attrition occurred within this control group, as approximately eight out of the original 15 participants left the company during the intervention year. Of the original 69 participants who attended the training, only 50 completed the follow-up survey 3 months post-intervention. Those participants who did not complete the follow-up survey were excluded from the analysis.

The use of non-probability sampling was appropriate given the study’s focus on a specific organizational setting where logistical and operational constraints necessitated targeted recruitment. This approach facilitated access to a representative cross-section of employees across various roles and work locations, aligning with the study’s objective to evaluate differential responses to training delivery formats (i.e., in-person versus online). While this approach facilitated targeted recruitment within a specific workplace context, it may limit the generalisability of findings to other organizational settings. Future studies should consider stratified or random sampling to enhance representativeness.

To gain a deeper understanding of participants’ experiences with the training intervention, semi-structured interviews were conducted with a purposive subsample of intervention participants 3–6 months after completion. A total of 15 employees (five from each training cohort) were selected to ensure representation across delivery modes (two-day intensive classroom, four half-day classroom, and four half-day online) and roles (sales versus non-sales). Interviewees were purposively chosen based on demographic variation (gender, tenure, and job grade) and quantitative outcomes (high versus low change in WEMWBS scores) to capture diverse perspectives regarding program impacts and barriers to behavioral change. This sample size aligns with recent findings by Squire et al. (63), who demonstrated that a sample of 15–23 participants typically achieves saturation in qualitative interviews, indicating minimal benefit in collecting additional data beyond this point.

The training program evaluation employed an explanatory sequential mixed-methods design grounded in pragmatism (61), whereby quantitative pre- and post-intervention surveys first established the statistical effects of the program on well-being, engagement and stress; this was subsequently followed 6 months later by semi-structured interviews to elucidate participants’ lived experiences and to understand why certain employees exhibited stronger or weaker responses; finally, these quantitative and qualitative findings were integrated to formulate practical, HR-driven recommendations for optimizing future workplace well-being interventions.

Quantitative data collection for the training program evaluation relied on five validated instruments—the Perceived Stress Scale (PSS-4), the Utrecht Work Engagement Scale (UWES-3), the Satisfaction With Life Scale (SWLS), the Flourishing Scale (FS) and the Warwick–Edinburgh Mental Well-Being Scale (WEMWBS-14)—administered online at two key time points: immediately before the intervention and 3 months afterwards. Each survey incorporated informed-consent, confidentiality and opt-out provisions to ensure ethical compliance, yielding comprehensive data on respondents’ perceived stress, work engagement and overall well-being.

Specifically, each instrument was selected on the basis of its robust psychometric properties and relevance to workplace well-being. The PSS is among the most widely used measures of perceived stress, capturing respondents’ appraisal of unpredictability, uncontrollability and overload in daily life (64–67). Work engagement was assessed with the ultra-short UWES-3, which operationalizes engagement as a positive, fulfilling work-related state reflected in vigor, dedication and absorption; this version demonstrates excellent reliability and valid associations with job resources and satisfaction, while showing minimal correlation with burnout (68–70). Life satisfaction was measured by the SWLS, a concise instrument favored for non-clinical populations that correlates with quality-of-life dimensions such as anxiety, optimism and sleep quality (71, 72). The FS provided a single eudaimonic well-being score based on eight items addressing relationships, optimism, self-esteem and purpose (73–75). Finally, the WEMWBS-14 comprises 14 positively phrased items scored on a five-point Likert scale (14–70), with higher scores indicating better mental health; which exhibits strong convergent validity, no ceiling effects and cultural applicability in the South African sample (76–79). The specific variables, their operationalization and classification are summarized in Table 3.

Qualitative data were also gathered via semi-structured interviews with a purposive subsample of 15 participants. Each 45–60 min interview followed a standardized guide, explored participants’ motivations for enrolling, perceived benefits and challenges, and suggestions for improving the program, and was audio-recorded with informed consent.

Data trustworthiness was established through multiple strategies to ensure credibility, dependability, confirmability, and transferability in the mixed-methods design. Credibility was enhanced by triangulating quantitative findings with qualitative interview data, allowing for cross-validation of well-being outcomes through participants’ reflections on the intervention. Verbatim quotations further substantiated key themes, providing direct evidence of participants’ lived experiences. Dependability was reinforced by maintaining a comprehensive audit trail that documented each phase of data collection and analysis, with independent coding checks conducted to verify theme consistency. Confirmability was addressed through reflexive memos recorded throughout the qualitative analysis to mitigate researcher bias and ensure neutrality. Additionally, the coding framework was independently reviewed by the co-authors, who provided feedback on theme coherence and consistency. Transferability was supported by providing rich, thick descriptions of the research context, participant characteristics, and intervention delivery, enabling readers to assess the applicability of findings to similar workplace settings. These measures collectively contributed to a systematic and transparent qualitative process, reinforcing the credibility and reliability of the findings.

Quantitative analyses were conducted in SPSS using a three-step non-parametric approach. First, Shapiro–Wilk tests (α = 0.05) assessed normality for each outcome and time point, confirming violations of normality assumptions for key well-being indicators [(80, 81); see Supplementary Table S1]. Second, within-group changes from pre- to three-month follow-up were evaluated with Wilcoxon signed-rank tests, and between-group differences across delivery formats at each time point were examined with Mann–Whitney U tests. Third, for the subset of participants with three measurements (pre-training, immediate post-training, three-month follow-up), Friedman tests assessed overall time effects, with post-hoc Wilcoxon signed-rank comparisons pinpointing significant pairwise contrasts (see Supplementary Tables S3–S5). All tests used a two-tailed α < 0.05 and reported exact p-values without Bonferroni correction to preserve statistical power in these small samples (82). Rank-biserial correlations (r) quantified effect sizes for all Wilcoxon and Mann–Whitney contrasts. Descriptive statistics—including medians, interquartile ranges, means, standard deviations, skewness, kurtosis and frequencies—summarized outcomes and demographics (see Supplementary Table S2), and Cronbach’s α coefficients established internal consistency for each scale. Cases with missing data were omitted via a complete-case approach, and data integrity was ensured by cross-checking SPSS imports against the original Excel files.

Qualitative interview transcripts were analyzed manually using thematic analysis (83). After familiarization through repeated readings, the researcher applied open coding (e.g., “increased motivation,” “practical application,” “emotional control”) to salient passages. Codes were then clustered into higher-order themes (e.g., “enhanced self-awareness and emotional resilience”), iteratively reviewed for coherence, and supported with verbatim quotations. Themes were reviewed against the raw data in multiple iterations, with detailed reflexive memos maintained throughout to ensure coherence and credibility.

Quantitative and qualitative data were synthesized during the final phase of analysis, enabling the integration of statistical outcomes with participant narratives to contextualize the observed intervention effects. This approach facilitated the identification of mechanisms underlying observed changes in well-being and engagement, contributing to a more nuanced interpretation of program effectiveness.

Quantitative and qualitative data were synthesized through narrative synthesis, whereby thematic findings from the qualitative phase were juxtaposed against statistical outcomes to identify areas of convergence and divergence. This integrative approach enabled a comprehensive interpretation of how the TGL training program influenced well-being and engagement, while also highlighting potential mechanisms of change underlying observed outcomes.

Key ethical safeguards were embedded throughout the quantitative phase to protect participants’ rights and well-being. Participation was entirely voluntary, with a clear right to withdraw at any time without repercussion, as specified in the online consent form. Before completing each survey, employees received comprehensive information on the study’s aims, duration, potential risks and anticipated benefits. To ensure anonymity and data security, responses were collected anonymously with support from the University of Johannesburg’s Statistics department and stored on a password-protected server accessible only to the main researcher and authorized co-authors. Measures were also taken to minimize survey fatigue and avoid psychological harm by spacing follow-up assessments and allowing participants to skip questions without penalty. Finally, although the pharmaceutical company’s HR department facilitated recruitment, care was taken to avoid any perception of coercion, ensuring that no employee felt obliged to participate.

In the qualitative phase, additional ethical safeguards were implemented to protect participant confidentiality and manage sensitive disclosures. Participants were informed of their right to withdraw from the interview at any stage without penalty, and all notes were securely stored on a password-protected server accessible only to the researcher.

Two full-day classroom (n = 19): Median WEMWBS rose by 7.0 points, Z = −2.474, p = 0.013, r = 0.57 (strong effect).

Two full-day classroom: Median UWES gain of 0.8 points, Z = −2.378, p = 0.017, r = 0.55 (strong effect).

Two full-day classroom: Positive trends in life satisfaction (SWLS, p = 0.055, r = 0.44), flourishing (FS, p = 0.075, r = 0.41) and stress reduction (PSS-4, p = 0.053, r = 0.44)—all moderate effects.

Overall, the two full-day classroom format emerged as the most effective delivery method, demonstrating moderate to strong effects across all primary and secondary outcomes. Specifically, significant improvements were observed in well-being (r = 0.57), engagement (r = 0.55), and stress reduction (r = 0.44), with effect sizes indicating practical significance. Conversely, the online format produced a modest gain in well-being (r = 0.53), while the fragmented half-day format yielded no significant changes across any outcomes, suggesting that continuity and immersive learning are critical to intervention success.

Strong Significant Changes (p ≤ 0.05):

Overall, intensive in-person training (two full days) yielded the most consistent and largest improvements across well-being, engagement and related outcomes. The online format improved well-being only, while the half-day classroom format produced a moderate boost in engagement but no reliable changes in well-being or secondary measures.

In interpreting effect sizes, small effects (r ≈ 0.10–0.30) suggest minimal practical impact, moderate effects (r ≈ 0.31–0.50) indicate noticeable but limited changes, and strong effects (r > 0.50) reflect substantial and potentially impactful outcomes (143). In this study, the two full-day format produced strong effect sizes for well-being (r = 0.57) and engagement (r = 0.55), underscoring its effectiveness relative to other formats.

Qualitative themes of immersive peer interaction, structured reflection and managerial support help explain why the full-day format was most effective.

Our findings align closely with the theoretical framework (Figure 2), which integrates PERMA, broaden-and-build, SDT, and JD-R models underpinned by experiential and transformational learning. The two-day, in-person workshops—the format that produced the largest WEMWBS and UWES gains—enabled uninterrupted cycles of concrete experience, reflective observation, abstract conceptualization, and active experimentation (54). This format allowed participants to fully engage in well-being practices, fostering positive emotions (PERMA) through structured activities such as gratitude journaling, strengths identification, and meaning-focused reflections. These activities not only broadened cognitive and social resources (49) but also addressed core SDT needs for autonomy, competence, and relatedness (50).

Moreover, the integration of PERMA constructs (gratitude, character strengths, relationships, meaning) effectively functioned as job resources within the JD-R model (51), equipping employees with personal and social resources to buffer job demands and sustain well-being. The full-day format further facilitated job crafting and peer support, reinforcing the application of well-being strategies in the workplace.

In contrast, the half-day and online formats fragmented these learning cycles, limiting opportunities to internalize well-being practices, as similarly found by Stratton et al. (94). Participants in these formats reported struggling to maintain focus amid work-related distractions, reducing the impact of voluntary activities (e.g., mindfulness, gratitude) that Seligman (52) identifies as drivers of sustainable happiness. This disparity underscores the importance of immersive, uninterrupted training in embedding well-being practices, particularly in high-demand workplace settings.

Thus, the mixed-methods data suggest that dedicated, intensive training time—supported by organizational endorsement—enhances the translation of theoretical frameworks into tangible workplace well-being outcomes, highlighting the critical role of experiential and transformational learning in HRD.

Meta-analyses and systematic reviews consistently demonstrate that well-being interventions yield moderate to large improvements in mental health, life satisfaction and psychological functioning in both clinical and occupational settings (86–89). Meanwhile, the TGL training program extends this evidence by showing that delivery format, organizational context and participant engagement critically shape outcomes. In particular, our finding that full-day classroom delivery produced larger improvements in WEMWBS and UWES scores than the online and half-day formats underlines the importance of immersive, interactive approaches.

Previous reviews have highlighted a pronounced WEIRD (Western, Educated, Industrialized, Rich, and Democratic) bias in positive psychology interventions and a paucity of workplace studies in African settings (90–92, 144). By implementing and evaluating a structured, program in South Africa, this study responds directly to those calls. Moreover, whereas many workplace interventions rely solely on randomized trials, TGL adopts an explanatory sequential mixed-methods design to integrate quantitative pre–post measures with qualitative interviews, thereby offering richer causal insights (47, 89).