Interval Level Staffing Requirements

Interval-level staffing requirements refers to the process of translating forecasted contact volume and Average Handle Time (AHT) into the minimum number of full-time equivalent (FTE) agents needed during each discrete time period — typically 15 or 30 minutes — within a scheduling day. This calculation forms the foundation of Capacity Planning Methods, Schedule Generation, and Rostering, connecting the demand forecast to actionable staffing targets. The output is commonly expressed as a required number of concurrent agents per interval, derived from mathematical models that account for call arrival randomness and agent availability. Accurate interval-level requirements are prerequisite to efficient scheduling: errors at this stage propagate through every downstream staffing decision.

Core Mechanics

From Volume to Workload

The first step converts forecasted contact volume into total workload, measured in Erlangs — a dimensionless unit representing one full hour of agent occupancy. For a given interval of length $t$ (in hours), the workload in Erlangs is:

Erlangs = (Volume × AHT) / t

For example, 200 calls expected in a 30-minute interval, each with an AHT of 4 minutes (0.0667 hours), yields:

Erlangs = (200 × 0.0667) / 0.5 = 26.7 Erlangs

This Erlang value represents the minimum number of agents required if agents were perfectly utilized and no call had to wait — the theoretical lower bound. Real contact centers require more agents to achieve target Service Level because arrival patterns are random and queuing dynamics must be absorbed.

Applying Erlang-C

The Erlang-C formula translates workload into the required agent count given a specified service level target (e.g., 80% of calls answered within 20 seconds). Erlang-C assumes an infinite queue: all callers wait indefinitely. Given the Erlang traffic intensity and target service level, the formula iterates over candidate agent counts until the modeled service level meets or exceeds the target.

The key inputs are:

• Traffic intensity (A): Erlangs computed above
• Target service level (proportion of calls answered within threshold T seconds)
• Number of agents (N): iterated until service level is met

Erlang-A extends Erlang-C by modeling caller abandonment, which is particularly relevant in high-volume, high-wait environments where infinite-patience assumptions are unrealistic.

Accounting for Shrinkage

The Erlang-derived agent count represents agents who must be simultaneously available — on-phone and ready. This is the net staffing requirement. To translate this into the number of scheduled agents (gross staffing requirement), the planner must account for Shrinkage: time during which scheduled agents are unavailable for calls due to breaks, training, meetings, coaching, and unplanned absence.

Gross Requirement = Net Requirement / (1 − Shrinkage Rate)

If net requirement is 30 agents and shrinkage is 33%, gross requirement becomes approximately 45 scheduled agents. This step is often misapplied: using an average daily shrinkage rate rather than an interval-specific one can introduce systematic under- or overstaffing by time of day, since shrinkage is not uniformly distributed across a shift.

Interval Length Selection

The choice of interval length — most commonly 15 or 30 minutes — affects both the precision of staffing requirements and the computational burden downstream. Shorter intervals (15 minutes) provide greater granularity and are better suited to environments with volatile intraday arrival patterns, such as high-volume voice channels. Longer intervals (30 minutes) introduce smoothing that can mask within-interval peaking, understating required agents during the peak portion of the interval. See Schedule Horizon and Planning Interval Selection for a full treatment of this tradeoff.

Most Erlang-C implementations operate on a per-interval basis, so the interval length directly determines how many staffing requirement data points are generated per day and how many scheduling cycles must be satisfied.

Handling Multichannel and Multi-Skill Environments

In blended contact centers handling multiple channels (voice, chat, email, back-office), interval-level requirements must be computed separately for each channel and then reconciled. Chat and concurrent-channel work violates Erlang-C assumptions, which model single-concurrent interactions; specialized queuing models or simulation are required for accurate chat staffing.

Multi-Skill Scheduling adds further complexity: agents who handle multiple queues must have their availability allocated across competing demand streams, requiring optimization rather than simple summation. The interval-level requirement for each skill group becomes a constraint in the scheduling problem rather than a standalone output.

Common Errors and Biases

Several systematic errors frequently occur in interval-level staffing calculations:

• Interval boundary smoothing: Averaging volume across intervals rather than preserving peak subinterval arrival rates suppresses true peak requirements.
• AHT pooling: Using a single AHT across all intervals when handle time varies by time of day (e.g., longer calls in early morning) produces systematic errors.
• Shrinkage applied as a uniform scalar: As noted above, non-uniform shrinkage creates interval-specific gaps.
• Ignoring occupancy caps: Erlang-C solutions sometimes produce high-occupancy answers that are operationally unsustainable. An Occupancy ceiling (e.g., 85%) should be imposed as a constraint, which may raise required agent counts above the pure service-level-driven result.
• Forecast error passthrough: Interval-level requirements inherit any bias in the volume forecast. See Forecasting Methods for forecast accuracy considerations.

Relationship to Schedule Generation

Interval-level staffing requirements serve as the primary input to Schedule Generation. Schedulers or WFM systems translate the per-interval FTE requirements into shift structures — determining which shifts must be offered, how many agents per shift, and how shift boundaries align with requirement peaks and valleys. Requirement profiles with pronounced intraday peaks (e.g., Monday morning volume spikes) often necessitate staggered shift starts or part-time coverage that pure rostering models cannot easily accommodate without multi-shift or split-shift designs.

Maturity Model Considerations

Maturity Level	Typical Practice
Level 1	Requirements computed from average daily volume and a single AHT; often at 30-minute or hourly granularity. Shrinkage applied as a flat scalar. Limited use of Erlang models; headcount estimated empirically.
Level 2	Erlang-C applied per interval. Interval-specific volume forecasts used. Basic shrinkage adjustment. Service level target defined.
Level 3	Erlang-A used where abandonment is material. Interval-specific shrinkage applied. Multichannel requirements separated by channel. Requirements feed directly into automated Schedule Optimization systems.

At higher maturity levels (L4–L5), simulation-based staffing replaces closed-form Erlang models, enabling accurate modeling of complex blended environments, non-stationary arrival processes, and skill-based routing dynamics. See WFM Labs Maturity Model for the full maturity framework.

Related Concepts

• Average Handle Time
• Erlang-C
• Erlang-A
• Shrinkage
• Occupancy
• Service Level
• Schedule Generation
• Schedule Horizon and Planning Interval Selection
• Multi-Skill Scheduling
• Capacity Planning Methods
• Forecasting Methods