Contacts per Agent

Goal of the analysis:

The objective is to measure frontline productivity by quantifying how many customer contacts each agent handles over a defined period, by channel and queue. This metric helps leaders balance service levels, cost-to-serve, and experience. It illuminates where throughput is constrained (e.g., long handle times, low occupancy, poor adherence) and where performance may be unsustainably high (e.g., quality or recontact risks). Executives use it to set staffing levels, optimize channel strategy (voice/digital), calibrate outsourcing, and drive process and system improvements that reduce cost while preserving CSAT and First Contact Resolution (FCR).

Data required:

• Contact volume and handling data:
  • Contacts handled by agent, by channel (voice, chat, messaging, email, social), queue/skill, and time stamp.
  • Average Handle Time (AHT) including talk/chat time and After-Contact Work (ACW).
  • Concurrency for digital channels (configured limits and observed concurrent sessions).
  • Disposition/resolution codes, contact reason taxonomy, FCR flags, repeat contact identifiers (if available).
• Workforce and time data:
  • Agent roster, role/skill/tenure, vendor vs captive, location.
  • Paid hours, staffed/available hours, on-queue time, shrinkage codes (PTO, training, meetings), schedule adherence.
  • Occupancy (handling time ÷ staffed time) by agent and interval.
• Service and quality data:
  • Service level, ASA, abandonment rate by interval/queue.
  • Quality assurance scores, CSAT/NPS post-contact, complaint flags, escalations.
• Context and benchmarks:
  • Historical performance by channel/queue.
  • External benchmark ranges by industry/channel (if available) and internal targets.
  • Seasonality markers, promotion or incident windows, product launches.
• Systems:
  • CCaaS/ACD: Genesys, NICE/inContact, Five9, Amazon Connect, Talkdesk.
  • CRM/case: Salesforce Service Cloud, Zendesk, Freshdesk.
  • WFM: NICE IEX, Verint, Calabrio; HRIS/timekeeping for paid hours.

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

1. Define the metric precisely. Use a clear definition and period (day, week, month):
   • Contacts per Agent (CPA) = Total contacts handled ÷ Average number of active agents in period.
   • Contacts per Paid Hour (CPPH) = Total contacts ÷ Total paid hours.
   • Contacts per Staffed Hour (CPSH) = Total contacts ÷ Total staffed/on-queue hours. Prefer CPSH for operational comparisons; use CPPH for cost.
2. Extract data. From CCaaS/CRM pull interval-level agent detail: contacts handled, AHT, ACW, channel, queue, disposition, FCR. From WFM/HRIS pull paid hours, staffed hours, adherence, shrinkage. Ensure consistent agent IDs across systems.
3. Clean and join. Standardize time zones; map queues to products/regions; exclude trainees if appropriate. Join on agent ID and interval (e.g., 15/30-minute buckets) to align contacts with staffed time.
4. Calculate core measures.
   • CPSH by agent/interval = Contacts / Staffed hours.
   • CPA per day = Σ contacts per agent per day; alternatively contacts ÷ average active agents per day.
   • Occupancy = (Talk/Chat time + ACW) ÷ Staffed hours.
   • Concurrency efficiency (digital) = Observed contacts per hour ÷ (60 ÷ AHT). Values >1 indicate concurrent handling.
5. Decompose throughput drivers. Build a productivity tree to attribute CPA differences:
   • CPSH ≈ Occupancy × (60 ÷ (AHT + ACW)) × Concurrency efficiency.
   • CPA (per day) ≈ CPSH × Staffed hours per agent × Adherence.
6. Segment the results. Break down by channel, queue/skill, contact reason, agent tenure cohorts (e.g., <90 days, 90–365, 1+ years), site/vendor, shift/time-of-day, weekday vs weekend, and complexity tier (Tier 1 vs Tier 2/back-office).
7. Quality and recontact overlay. Link CPA with FCR and CSAT. Flag repeat contacts within X days for same reason. Create a “quality-adjusted CPA” removing repeat contacts to avoid rewarding throughput that drives rework.
8. Service-level guardrails. Align CPA with SL/ASA. High CPA achieved by understaffing can inflate abandonments—include SL and abandons alongside throughput.
9. Benchmarking. Compare channels/queues to internal top quartile agents and to historical baselines. Where available, add external benchmarks for similar complexity and industry.
10. Trend and seasonality. Plot CPA, CPSH, AHT, occupancy weekly for 6–12 months. Tag events (product releases, outages). Use control charts to distinguish normal variation from shifts.
11. Identify outliers and root causes. Use box plots to find low and high outlier agents. Drill into their AHT, ACW, occupancy, adherence, and contact mix. Review QA notes/tool friction for low performers; assess quality for very high performers.
12. Synthesize insights. Quantify lift opportunities: e.g., “If bottom quartile agents reached median CPSH, we’d process +X contacts/day, avoiding Y FTE or reducing backlog by Z days.”

Format of the output of analysis:

• Executive summary slide with CPA/CPSH by channel and variance vs target, plus key driver tree.
• Summary table by queue/segment with CPA, CPSH, AHT, ACW, occupancy, adherence, SL, CSAT, FCR.
• Line charts showing trends over time; heatmaps by hour-of-day/day-of-week.
• Box-and-whisker plots of agent-level CPSH and CPA to show distribution and outliers.
• Waterfall/driver decomposition from paid hours to quality-adjusted contacts.
• Benchmark comparison views (internal top quartile vs median; external where applicable).
• Scatter plots of CPA vs CSAT/FCR to visualize trade-offs.
• Dashboard (BI tool) with filters for channel, queue, tenure, site/vendor.

How to interpret results:

• High CPA/CPSH: Indicates strong throughput. Positive if SL and quality are stable. Risk if accompanied by declining CSAT/FCR, rising repeats, or QA issues—may reflect rushing or inadequate diagnosis.
• Low CPA/CPSH: Often driven by long AHT/ACW, low occupancy, poor adherence, tool friction, or handling complex contacts. Acceptable in Tier 2/back-office; concerning in Tier 1 unless justified by complexity or failure demand.
• Channel effects: Voice typically lower throughput than chat/messaging due to lack of concurrency. Do not compare raw CPA across channels without adjusting for complexity and concurrency.
• Segment differences: New hires should trend up as proficiency grows. Vendors/sites with materially lower CPSH warrant process/tool parity checks. Certain contact reasons (billing disputes, technical escalations) appropriately show lower CPA.
• Benchmark context: Use internal top quartile as pragmatic targets. External benchmarks are directional; align by complexity and channel.
• Trends: Sustained improvement with stable quality suggests real efficiency gains. Volatility may indicate staffing misalignment or unstable demand.

Steps a company can take to improve on this measure:

• Process and policy simplification:
  • Eliminate failure demand by fixing top contact drivers; simplify policies that require lengthy explanations or approvals.
  • Standardize best-practice call flows and chat scripts; expand guided workflows.
  • Introduce tiering/triage to route complex issues to specialists and keep Tier 1 streamlined.
• Data, systems, and tooling:
  • Integrate CRM, billing, and knowledge to reduce toggling; enable screen-pop with customer/context.
  • Automate repetitive steps (RPA), reduce ACW via macros and templated responses.
  • Tune digital concurrency safely (e.g., 2–3 chats) and provide workload balancing; improve IVR/IVA containment to deflect simple contacts.
  • Enhance WFM: schedule to demand curve, reduce shrinkage, and raise occupancy without breaching SL.
• Capability, coaching, and governance:
  • Targeted coaching on diagnostics and resolution efficiency; micro-coaching from QA insights.
  • Knowledge base hygiene; quick-reference guides; hot cards for spikes/incidents.
  • Set balanced scorecards (CPA/CPSH with CSAT/FCR/QA) to avoid throughput-at-all-costs behavior.
• Product, targeting, and channel strategy:
  • Proactive communications (status pages, notifications) to prevent avoidable contacts.
  • Move simple intents to self-service and asynchronous channels; protect voice for complex/emotional cases.
  • If CPA high but FCR/CSAT low: slow concurrency, reinforce quality steps, refine knowledge.
  • If CPA low and AHT high on simple intents: redesign flows, fix tooling, and consider specialization.

Benchmark comparisons:

General benchmarks:

• Voice Tier 1: 7–12 contacts/hour typical depending on AHT (4–7 minutes) and ACW; 45–70 contacts/day for a fully staffed day.
• Chat (concurrency 2–3): 12–20 contacts/hour; higher with short intents and good macros.
• Messaging/asynchronous: 15–25 contacts/hour for simple tickets; 5–10 for complex cases.
• Tier 2/back-office: 2–6 cases/hour reflecting higher complexity and research time.
• Occupancy targets commonly 80–90%; sustainable levels depend on break structure and complexity.

Segment- or industry-specific benchmarks:

• Retail/e-commerce simple order inquiries: upper end of ranges; financial services and healthcare trend lower due to compliance and complexity.
• Technical support often 3–8 voice contacts/hour; enterprise B2B even lower.
• For outsourcing, compare vendors on like-for-like queues and ensure tool/process parity; top quartile internal performers provide robust internal benchmarks.

If robust external benchmarks are unavailable, construct internal benchmarks by queue and channel: compare current month vs prior 12-month averages, and set targets at the top quartile agent performance adjusted for tenure and complexity.