AI workflow automation and automated ticket handling are strongly linked to higher public-sector IT support efficiency, explaining over half the variation in service performance in a 214‑respondent survey; robust governance boosts cost savings, but accessibility practices lag.

Main Finding

An AI-powered automation framework for IT support in public sector organizations is associated with substantial gains in service delivery efficiency (model R² = 0.58). Workflow automation (β = 0.41, p < .001) and automated ticket handling (β = 0.36, p < .001) are the strongest predictors of performance. Cost-efficiency gains are positive but smaller (β = 0.22, p = .003; R² = 0.46). Strong governance and risk management both directly improve performance (β = 0.31, p < .001) and moderate the automation → cost-efficiency relationship (interaction β = 0.19, p = .005). Digital accessibility shows only partial integration (M = 3.56, SD = 0.65), indicating that accessibility benefits are not automatic and require deliberate inclusion in design and operations. Reliability of constructs is high (Cronbach’s α = 0.84–0.91).

Key Points

• Study design: quantitative, cross-sectional survey of 214 senior IT managers, service desk leaders, and digital transformation roles across ministries, departments, and public agencies.
• Adoption levels (means ± SD):
  • Workflow automation: M = 4.02, SD = 0.59 (higher)
  • Automated ticket handling: M = 3.95, SD = 0.62 (higher)
  • Predictive analytics: M = 3.41, SD = 0.71 (lower)
  • Digital accessibility indicators: M = 3.56, SD = 0.65 (moderate)
• Statistical results:
  • Service delivery efficiency model explains 58% of variance (R² = 0.58).
  • Cost-efficiency model explains 46% of variance (R² = 0.46).
  • Significant positive betas for workflow automation and automated ticket handling on efficiency; smaller but significant beta for automation on cost efficiency.
  • Governance/risk management: significant direct effect on performance and significant moderator of automation → cost-efficiency.
• Mechanisms discussed:
  • NLP for automated intake, classification, similarity matching and conversational interfaces reduces misclassification, reassignments, and handling time.
  • Workflow automation and RPA execute routine tasks (provisioning, resets, approvals) reducing touch time and cycle time.
  • Operational intelligence (log/event correlation, anomaly detection) improves detection/diagnosis and links tickets to monitoring data.
  • Accessibility treated as an operational outcome (accessibility-related incident detection, prioritization, resolution time) rather than solely a design/compliance metric.
• Measurement & reliability: constructs operationalized with ticketing and performance metrics and survey items; internal consistency strong (α = 0.84–0.91).
• Caveats: cross-sectional survey—associations not definitive causal proof; self-reported measures and organizational heterogeneity (legacy systems, governance contexts) may affect generalizability.

Data & Methods

• Sample: 214 respondents in senior IT/service desk/digital transformation roles from public sector organizations (ministries, departments, agencies).
• Design: quantitative, cross-sectional survey supplemented by descriptive operational indicators drawn from ITSM/ticketing contexts.
• Analyses:
  • Descriptive statistics (means, standard deviations) to assess adoption levels and accessibility indicators.
  • Reliability testing (Cronbach’s alpha) to assess internal consistency across constructs (α = 0.84–0.91).
  • Multiple regression models to estimate relationships between automation capabilities and outcomes:
    • Service delivery efficiency model: predictors included workflow automation, automated ticket handling, predictive analytics, governance/ITSM maturity, etc.; R² = 0.58.
    • Cost-efficiency model: R² = 0.46; governance tested as moderator (interaction β = 0.19, p = .005).
  • Significance reported for key predictors (p-values, standardized betas).
• Operationalization highlights:
  • Automation intensity: % tickets auto-classified, % requests fulfilled without human intervention, automated event correlation rates.
  • Service performance: response time, resolution time, first-contact resolution rate, reassignment and escalation frequencies, backlog size, SLAs.
  • Accessibility: accessibility-related incident counts, resolution time, recurrence, user feedback.
• Limitations acknowledged by study: cross-sectional design, variability across public organizations, partial reliance on survey responses rather than uniform operational logs.

Implications for AI Economics

• Productivity and value capture:
  • Large measured effect on service delivery (R² = 0.58) implies significant productivity returns from investing in workflow automation and automated ticket handling—these are likely high-priority, high-ROI AI investments within IT operations.
  • Cost-efficiency gains are present but more modest (β = 0.22), suggesting that process speed and quality improvements may precede or outweigh direct bottom-line cost reductions in public-sector settings.
• Investment prioritization:
  • Prioritize automation that substitutes repetitive operational labor (provisioning, password resets) and automates high-volume classification/routing tasks (NLP-driven ticket intake). These show the largest association with performance.
  • Predictive analytics adoption lags (lower mean); it may have longer payback periods or higher upfront data/integration costs—economic models should account for delayed benefits and integration friction.
• Governance and procurement policy:
  • Strong governance increases returns and moderates cost outcomes; public-sector procurement and governance frameworks materially affect the economic viability of AI automation. Investments without governance/enforceable auditability may yield lower cost benefits and higher risk.
  • Expenditure decisions should internalize governance costs (audit logs, explainability, exception handling) as necessary overhead to realize net benefits.
• Labor and skill effects:
  • Automation will reallocate labor from routine ticket handling to higher-value tasks (escalation, governance, exception handling); training and role redesign are necessary economic considerations.
  • Short-term transitional costs (reskilling, change management) should be included in cost-benefit analyses.
• Accessibility and distributional impacts:
  • Accessibility benefits are not automatic—if AI automation is not designed to flag and prioritize accessibility-related incidents, the technology may improve aggregate metrics while leaving vulnerable users behind. Economically, failure to embed accessibility leads to negative externalities (exclusion costs, reputational and compliance penalties).
  • Policymakers should treat accessibility-sensitive automation as a public good with potential positive externalities (inclusion, reduced social costs).
• Evaluation and measurement:
  • The study's operationalization ties AI interventions to concrete metrics (response/resolution times, first-contact resolution, accessibility incident rates), providing a template for cost-effectiveness and ROI measurement in procurement/business cases.
  • Future economic evaluations should use longitudinal operational logs and quasi-experimental designs to better estimate causal returns, amortization schedules, and total cost of ownership.
• Scaling and heterogeneity:
  • Heterogeneous legacy systems in the public sector imply variable marginal returns; economic models must include integration complexity and vendor/legacy lock-in risks.
  • Centralized vs. decentralized IT structures will change bargaining power and unit costs—scale economies likely where common platforms and data standards exist.