Automation: share of manual operations, how to measure it
Automation (share of manual operations) shows what portion of processes is performed without human involvement or with minimal interaction. A high level of automation reduces costs and processing time, decreases errors, and improves scalability. It is important to combine automation with quality standards so as not to "speed up" inefficient processes.
What the Metric Measures
The share of transactions/requests/operations that involved manual steps (copying data, manual approvals, manual deployments, manual routing), as well as the share of fully "touchless" ones (straight-through processing, STP). For customer support — the share of tickets closed without agent involvement; for DevOps — the share of automated deployments; for the back office — automatic payments/reconciliations.
Calculation
- Manual Share, % = Operations with manual steps / All operations × 100
- Automation Rate, % = 100 − Manual Share
- STP (Touchless), % = Operations without a single manual step / All operations × 100
- Assisted Automation, % = Operations with 1 click/macro / All operations × 100
- Manual Steps per Case = Σ manual steps / Number of cases
- RPA Coverage, % = Operations involving an RPA bot / All operations × 100
- CI/CD Automation, % = Automated deployments / All deployments × 100
- Auto-Resolution, % (Support) = Closed automatically/by bot / All tickets × 100
Define "what a manual step is": physical clicks/data entry, manual approvals, copying between systems, manual imports. Account separately for exceptions/quality checks (human-in-the-loop).
Calculation example
3,000 orders were processed in a month: 1,950 went through automatically (import, invoice, reservation) and 1,050 needed manual entry or correction. Manual share = 1,050 / 3,000 × 100 = 35%. At 6 minutes per manual order that is 105 hours of work a month.
Data Sources
ITSM/CRM/Helpdesk (action logs), BPM/Workflow (step statuses), RPA orchestrators, integration/API logs, CI/CD (deployments/pipelines), ERP/1C/banking gateways (reconciliations/payments), event analytics and audit systems.
Interpretation
- Manual Share↑ — lack of integrations/APIs, non-standardized processes, manual approvals, low-quality data
- STP↑ with declining quality — risk of "over-acceleration"; check errors/rework
- Combine with SLA/CSAT, AHT, Error Rate, Rework%, and cost per operation
- Prioritize automations using the formula: volume × repeatability × errors × time
Useful Breakdowns
- Process: support/back office/logistics/finance/DevOps
- Channel/system: email/chat/API/RPA/ERP/CRM
- Case type: standard vs. exception; risk/criticality
- Team/shift/region; data source/supplier
Dashboard Visualizations
- Trend of Automation Rate and STP% with target corridors
- Pareto of manual steps/processes by labor intensity
- Heatmap of manual operations by system/team
- Before/After for automation launches (time/errors/cost)
- Correlations with SLA, AHT, Error Rate, Backlog
How to Set Targets
Benchmarks: STP ≥ 50–80% for standard transactions; Automation Rate ≥ 70–90% in mature processes; a 30–50% reduction in Manual Steps per Case. Implement an initiative queue (pipeline), a quality/exceptions gate, monitoring with a "kill switch," and ROI metrics (savings in time/errors/cost).
Common Mistakes
- Automating a chaotic process without standardization
- Ignoring quality/exceptions and having no manual route (fallback)
- Local macros without support and monitoring
- Incorrect accounting of manual actions (incomplete logs, "gray" work)
- No user feedback and no error metrics
Questions for Discussion
- Which 5 operations account for 80% of manual labor, and what is their automation ROI?
- Are there integration/API bottlenecks, and what quick wins are possible?
- What exception and human-in-the-loop rules are needed to ensure quality?
- How do we track the effect: time, errors, costs, SLA/CSAT?