Service Level

Service level measures the percentage of contacts answered within a defined time threshold.

Service level is a workforce management metric that measures the percentage of customer contacts answered within a defined time threshold. Expressed as "X% of contacts answered in Y seconds" (e.g., 80% in 20 seconds, commonly written as 80/20), service level is the primary accessibility target in most contact center operations and a foundational input to workforce management staffing calculations.

Service level quantifies the tradeoff between customer wait time and staffing cost: higher service levels require more agents, which increases expense but reduces customer effort. It is one of the most widely tracked metrics in contact center management and a central output of the Erlang C staffing model.

Definition and Formula

Service level is calculated as:

$\text{Service Level}=\frac{\text{Number of contacts answered within threshold}}{\text{Total contacts offered}}\times 100\mathrm{\%}$

The threshold (also called the target answer time or acceptable wait time) is the maximum number of seconds a customer should wait before being connected to an agent. Common thresholds include 20, 30, and 60 seconds for voice, and longer intervals for digital channels.

Variations in the formula produce materially different results depending on how abandoned calls are treated:

Method	Abandoned calls	Effect
Offered	Counted as failures (not answered within threshold)	Most conservative; produces lowest service level
Removed	Excluded from both numerator and denominator	Common in ACD reporting; assumes abandoned calls are irrelevant
Positive	Short abandons (within threshold) counted as successes	Assumes callers who hung up quickly were satisfied or called in error

The method used must be consistent to allow meaningful trending and comparison. ICMI recommends documenting the calculation method alongside any reported service level figure.

Common Targets

The 80/20 Standard

The most widely cited service level target is 80/20 — 80% of calls answered within 20 seconds. This standard originated in the telecommunications industry in the 1980s and became entrenched through early WFM software defaults and industry benchmarking. Despite its ubiquity, there is no mathematical derivation that proves 80/20 is optimal; it represents a historically accepted balance between cost and accessibility.

Alternative Targets

Organizations increasingly set service level targets based on their specific business context rather than defaulting to 80/20:

Channel	Common Targets	Notes
Inbound voice	80/20, 80/30, 70/30	Higher-value or emergency lines may target 90/10 or 95/15
Chat	80/30, 80/60	Concurrent chat sessions complicate measurement
Email	80% within 4 hours, 95% within 24 hours	Measured in hours rather than seconds
Social media	80% within 60 minutes	Varies widely by brand promise
Back office	90% within SLA (hours/days)	Knowledge work often uses completion-time SLAs

The appropriate target depends on customer expectations, competitive benchmarks, revenue impact of wait time, and the organization's position on the CX/Cost/EX triad.

Relationship to Other Metrics

Average Speed of Answer (ASA)

Average speed of answer is the mean wait time across all answered calls. While service level measures the proportion of calls meeting a threshold, ASA measures the average wait experience. Both can be derived from the same Erlang C model, but they tell different stories:

• A center can meet an 80/20 service level while some callers wait several minutes (the 20% tail)
• ASA smooths those extremes into a single average, obscuring the worst-case experience

Abandonment Rate

Abandonment rate measures the percentage of callers who hang up before being answered. Higher service levels generally correlate with lower abandonment, but the relationship is non-linear — small changes in staffing near the knee of the service level curve produce disproportionate changes in abandonment.

Occupancy

Occupancy and service level move inversely: higher service levels require more available agents, which lowers occupancy. In small teams, achieving high service levels requires accepting low occupancy — a mathematical reality often misunderstood by operations leaders accustomed to manufacturing utilization targets.

The Erlang C Connection

The Erlang C formula provides the mathematical foundation for service level planning. Given the arrival rate, average handle time, and number of agents, Erlang C calculates the probability that a caller will wait in queue — from which service level can be derived.

The standard staffing process in contact center WFM is:

1. Forecast contact volume and AHT by interval
2. Calculate base staff required to meet the service level target using Erlang C
3. Add Shrinkage to determine scheduled staff
4. Build schedules to deploy that staff

Erlang C assumes Poisson arrivals, exponential handle times, infinite patience (no abandonment), and a single queue with homogeneous agents — assumptions that are frequently violated in practice. See Erlang C for a detailed treatment of the formula, its limitations, and alternatives.

Service Level in Multi-Channel Environments

As contact centers expand to multi-channel and omnichannel operations, service level measurement becomes more complex:

• Chat: Agents handle multiple concurrent sessions; the when answered trigger may differ from voice (first response vs. session assignment)
• Email: Measured in hours or days; service levels like "95% responded within 24 hours" bear little mathematical resemblance to voice service level
• Social media: Response time expectations vary by platform and are often publicly visible
• Back office: Completion SLAs (e.g., "process 90% of claims within 3 business days") replace real-time service levels

A unified service level metric across channels requires weighting by volume, customer impact, and channel-specific expectations. Few organizations have achieved this effectively, though the Future WFM Operating Standard proposes frameworks for multi-channel service optimization.

Criticisms and Limitations

Despite its dominance, service level has significant limitations as a primary performance target:

• Arbitrary threshold: The choice of 20, 30, or 60 seconds is typically convention rather than evidence-based
• Distribution blindness: An 80/20 service level says nothing about the experience of the 20% who waited longer
• Gaming risk: Agents or supervisors may take actions that improve service level but harm other outcomes (e.g., shortened AHT at the cost of resolution quality)
• Interval volatility: Service level fluctuates dramatically at the interval level; daily or weekly averages mask significant intra-day variation
• Cost discontinuity: The service level curve is non-linear — improving from 80% to 85% costs far more than improving from 50% to 55%

Modern Alternatives and Complements

Several alternative or complementary approaches have emerged:

• Customer Effort Score (CES): Measures perceived effort rather than wait time
• Patience-weighted service level: Weights the threshold by actual customer patience distribution
• Multi-objective optimization: Balances service level against cost and employee experience simultaneously
• Percentile-based targets: "95th percentile wait time under 120 seconds" addresses the tail better than X/Y
• Outcome-based metrics: First contact resolution, CSAT, and revenue per contact may better reflect actual customer value

Maturity Model Position

Service level measurement and targeting evolves across maturity levels:

• Level 1 (Reactive): Service level tracked ad hoc, often daily or weekly averages only
• Level 2 (Foundational): 80/20 target established, interval-level tracking, Erlang C-based staffing
• Level 3 (Integrated): Channel-specific targets, service level tied to forecast accuracy feedback loops
• Level 4 (Optimized): Dynamic service level targets that flex based on cost-benefit analysis per interval
• Level 5 (Adaptive): Multi-objective service targets optimized across human and AI agent pools

See Also

• Workforce Management — Overview of the WFM discipline
• Erlang C — Mathematical foundation for service level staffing
• Occupancy — Related metric that moves inversely to service level
• Average Handle Time — Key input to service level calculations
• Shrinkage — Factor applied between base staff and scheduled staff
• Forecast Accuracy Metrics — Upstream driver of service level achievement
• WFM Goals — The CX/Cost/EX triad
• Customer Access Strategy — Strategic framework for accessibility decisions
• First Contact Resolution — Complementary quality metric