Impact of Digital Assistants on Office Workflow Efficiency: A Field Study

Thesis Abstract

The rapid integration of digital assistants (DAs) into contemporary office environments has raised questions about their actual impact on workflow efficiency, information management, and employee performance, particularly in medium-sized organizations that adopt AI-powered scheduling, email triage, and task automation tools. This study addresses the problem of mixed evidence regarding whether DAs consistently enhance productivity or introduce new coordination frictions. The aim is to evaluate the effect of DAs on office workflow efficiency and identify contextual and individual factors that modulate this relationship. Specifically, the study tests three objectives (1) quantify changes in task completion time, error rates, and perceived workload after DA deployment; (2) examine variations in workflow efficiency across job roles, tenure, and prior DA exposure; and (3) explore organizational processes and user experiences that mediate or moderate observed effects. The theoretical framework integrates the Technology Acceptance Model (TAM) to explain user adoption and perceived usefulness, the Job Demands-Resources (JD-R) model to interpret workload fluctuations, and Activity Theory to frame tool-mediated work practices within organizational contexts. The research employs a mixed-methods design, combining a quasi-experimental field study with a cross-sectional survey and in-depth interviews. The population comprises employees at six mid-sized firms across finance, operations, and administrative support sectors that implemented DAs within the prior six to twelve months. A sample of 240 employees is selected through stratified random sampling by department and role, supplemented by 40 managers for validation and qualitative triangulation. Data collection instruments include a DA usage log and performance metrics extracted from enterprise systems, a standardized Productivity and Workload Questionnaire (PWQ) for subjective assessment, and semi-structured interview guides to capture contextual factors and user experiences. Validity and reliability are ensured through pilot testing of instruments, Cronbach’s alpha assessments for multi-item scales (? ? 0.80), and triangulation across quantitative and qualitative data. The data analysis employs hierarchical linear modeling (HLM) to assess the impact of DA use on objective efficiency indicators while accounting for nested data structures (employees within teams within firms). Regression analyses determine relationships between DA engagement, perceived usefulness, and workflow outcomes. Thematic analysis of interview transcripts identifies mediators and moderators—such as task type, standard operating procedures, and management support. Descriptive statistics, t-tests, and ANOVA are used to explore group differences, with robustness checks via sensitivity analyses and multicollinearity diagnostics. Key expected findings include (a) modest but statistically significant reductions in average task completion time (mean decrease of 12–18%), accompanied by a decline in error rates (5–9%), and a measurable reduction in perceived cognitive load for routine tasks; (b) heterogeneous effects across job roles, with clerical and coordination tasks benefiting more than highly specialized expert tasks, and with greater gains in teams exhibiting higher DA training and SOP alignment; (c) identification of critical mediators such as perceived ease of use, supervisor encouragement, and predefined workflow rules, and moderators including prior DA experience, task complexity, and organizational process maturity. The study anticipates discovering potential trade-offs, including increased time spent in DA configuration and monitoring, which may offset some efficiency gains in certain contexts. The contribution to knowledge lies in providing empirical, context-rich evidence on how DAs influence office workflow efficiency in real-world settings, clarifying the boundary conditions under which benefits prevail and offering a validated analytical framework that integrates TAM, JD-R, and Activity Theory to explain the observed phenomena. The findings will inform management practice by outlining actionable guidelines for DA selection, user training, and process redesign to maximize productivity while minimizing disruption. Policy implications pertain to organizational governance around data privacy, human–computer interaction standards, and change management strategies. The final conclusions will emphasize that DA-driven efficiency gains are contingent on task design, organizational readiness, and sustained user engagement, with recommendations emphasizing iterative optimization, targeted training, and continual performance monitoring to sustain improvements.