Utilization Rate of Agents

Goal of the analysis:

The objective is to quantify how effectively customer service agents’ paid time is spent on productive work, and to pinpoint structural and operational levers to optimize capacity, cost per contact, and service levels without compromising customer experience. For executives, utilization is a core efficiency KPI tied to workforce costs (often the largest cost line in support), forecast accuracy, and the ability to absorb demand volatility. Managed well, it reduces overstaffing, protects agent wellbeing, and improves service-level reliability and unit economics.

Data required:

• Workforce management and scheduling data:
  • Scheduled hours by agent (paid vs unpaid), shift start/end, breaks.
  • Schedule adherence and exceptions (late, early out, overtime).
  • Shrinkage categories: PTO, sick, training, coaching, meetings.
• Telephony/CCaaS and omnichannel handling data:
  • Talk time, hold time, after-call work (ACW), ready/available time, not-ready codes.
  • Contacts handled by channel (voice, chat, email, messaging, social).
  • Concurrency data for digital channels (e.g., chats handled simultaneously).
• Case/ticket system data:
  • Case handling times, statuses, reassignment, backlog.
  • Work-type tagging (queue, skill, product, severity).
• Operational context and outcomes:
  • Forecasted demand and AHT, actual volume by interval.
  • Service levels (e.g., % answered in X seconds), ASA, abandonment rate.
  • Quality/CSAT/FCR metrics to balance efficiency with experience.
• People and cost data:
  • Agent profile (tenure, team, location, skills), attrition/absenteeism.
  • Payroll/timekeeping for paid hours and overtime.
• Historical and benchmarks:
  • Time-series of utilization/occupancy, seasonal patterns.
  • Internal targets and any external benchmark references.

Detailed step-by-step instruction on how to conduct the analysis:

1. Define the metric and scope.
   • Utilization (primary): Productive time / Paid time. Productive time = on-contact time (Talk + Hold + ACW) + approved productive off-phone work (cases, emails, escalations). Paid time = total paid hours excluding unpaid breaks.
   • Occupancy (reference): (Talk + Hold + ACW) / (Ready + Talk + Hold + ACW). Use to diagnose real-time loading for contact channels.
   • Set the analysis window (e.g., weekly/monthly) and units (agent-day, 30-min intervals).
2. Extract data from systems.
   • WFM (e.g., NICE, Verint): schedules, adherence, shrinkage codes, exceptions.
   • CCaaS/ACD (e.g., Genesys, Five9, Amazon Connect, Cisco, Zendesk, Service Cloud): interval-level handle times and states.
   • Ticketing (e.g., Zendesk, ServiceNow): case handling durations, backlog work.
   • Payroll/timekeeping: paid hours, overtime, unpaid breaks.
3. Normalize and classify time codes.
   • Map agent states into: On-contact, Available/Ready, ACW, Productive off-phone, Non-productive (idle), Shrinkage (PTO, training, meetings), Unpaid breaks, System downtime.
   • Align time zones and ensure intervals match across systems.
4. Calculate foundational metrics per agent per interval.
   • Paid time = scheduled paid time + overtime – unpaid breaks.
   • Productive time = On-contact + ACW + Productive off-phone.
   • Occupancy = (On-contact + ACW) / (Ready + On-contact + ACW).
   • Shrinkage rate = Shrinkage time / Paid time; Adherence = Time in planned states / Scheduled time.
5. Adjust for channel specifics.
   • Digital concurrency: attribute concurrent chat time carefully to avoid double-counting. Use effective handle time per agent = total active chat time across sessions capped by elapsed time.
   • Case/back-office: use work-item handle time and queue work to capture productive off-phone tasks.
6. Aggregate and segment.
   • Roll up by team, site, channel, queue, tenure bands, and time-of-day/weekday.
   • Create distributions (median, quartiles) to identify outliers and variability.
7. Benchmark and trend.
   • Compare to targets and prior periods; identify seasonality and post-change impacts.
   • Overlay with service levels, ASA, CSAT, QA to ensure efficiency does not degrade outcomes.
8. Diagnose drivers and gaps.
   • Decompose time with a waterfall: Paid → Shrinkage → Available → On-contact → ACW → Productive off-phone → Idle.
   • Quantify the contribution of forecast error, schedule mix, adherence, routing inefficiency, or excessive shrinkage.
9. Size the improvement opportunity.
   • Capacity gap (hours) = (Target utilization − Current utilization) × Total paid hours.
   • Convert to FTE equivalent and potential cost impact or volume absorption.
10. Validate findings with operations.
    • Review outliers and code mappings with supervisors and WFM; check for policy or system issues (e.g., misused not-ready codes).

Format of the output of analysis:

• Executive summary slide with current vs target utilization, capacity gap, and top drivers.
• Time-allocation waterfall chart from Paid time to Productive time.
• Heatmaps by 30-minute intervals showing occupancy/utilization and service levels.
• Distribution plots (box-and-whisker) of utilization by team/tenure to highlight variance.
• Segmented tables by channel/queue with AHT, volume, shrinkage, adherence, and utilization.
• Benchmark comparison view (internal quartiles; external ranges where available).
• Action plan matrix linking identified drivers to specific initiatives and estimated impact.

How to interpret results:

• High utilization (e.g., >75–80% sustained for voice): Indicates strong loading but risks burnout, increased AHT, quality drift, and attrition. If CSAT/QA are stable and service levels are met, it may be healthy; if complaints or rework rise, workload is too intense.
• Low utilization (e.g., <60%): Suggests overstaffing, forecast/scheduling mismatches, excessive shrinkage, routing issues, or idle time. Check occupancy and adherence to isolate the driver.
• Channel differences: Voice typically lower utilization than digital with concurrency; back-office case work shows lower occupancy but can have solid utilization if off-phone tasks are captured.
• Variability matters: High volatility by interval points to forecast or staffing mix issues; persistent low performers may reflect coaching or adherence problems.
• Benchmark influence: Use external ranges as directional; prioritize internal top-quartile comparisons and trend improvements to set realistic targets.
• Trend lines: Improving utilization alongside stable service and quality is positive. Improving utilization while service or quality deteriorates signals over-tightening.

Steps a company can take to improve on this measure:

• Process and policy:
  • Optimize shrinkage: schedule training/meetings in low-demand windows; cap discretionary meetings; formalize shrinkage approval.
  • Reduce ACW with templates, knowledge base improvements, and automated dispositioning.
  • Clarify and enforce state code usage to reduce “hidden idle” in not-ready codes.
• Data, systems, and tooling:
  • Improve forecasting (Erlang-C or simulation) and scheduling to align staffing with interval demand.
  • Enable channel blending and intelligent routing to balance load across queues.
  • Increase digital concurrency where appropriate and safe; leverage bots/IVR for simple intents and proactive communications to deflect avoidable contacts.
  • Implement real-time adherence dashboards and alerts; monitor system downtime.
• Capability, training, and governance:
  • Cross-skill agents to flex capacity across channels/queues; create a rapid response pool for spikes.
  • Coaching for handle-time efficiency and first-contact resolution without sacrificing quality.
  • Establish governance for utilization targets with guardrails on occupancy to protect wellbeing.
• Targeting, product, and workload reduction:
  • Fix top contact drivers (policy/product defects) and improve self-service content to reduce volume.
  • Use asynchronous messaging for non-urgent issues to smooth peaks.
• Example plays:
  • If utilization is low and occupancy is low: reduce staffing or improve forecast/schedule mix; consider cross-skilling.
  • If utilization is low but occupancy is high: too much shrinkage or ACW; streamline wrap, reduce meetings, improve tools.
  • If utilization is high and quality is falling: relax occupancy, add micro-breaks, bolster knowledge/tools, adjust staffing.

Benchmark comparisons:

General benchmarks:

• Overall agent utilization (productive time / paid time): commonly 60–75% in inbound support; high performers may sustain 70–80% with strong quality controls.
• Agent occupancy (voice): typically 75–85% during steady state; sustained >85–90% often degrades experience and increases burnout.

Segment- or industry-specific benchmarks:

• Voice-heavy inbound service: utilization 60–70%; occupancy 75–85%.
• Digital chat/messaging with 2–3 concurrent sessions: utilization 70–85%; effective occupancy can range 60–90% depending on concurrency and automation.
• Email/case back-office: utilization 55–75% given variability and batching; occupancy not applicable.
• Seasonal/retail peaks may temporarily push higher occupancy; use short-term guardrails and recovery plans.

If external benchmarks are sparse or definitions vary, construct internal benchmarks by comparing top-quartile teams, channels, and shifts over the last 3–6 months, and set targets by queue/channel, not a single enterprise-wide number.