WFM Labs Erlang-O™

From WFM Labs
WFM Labs Erlang-O™

WFM Labs Erlang-O™

Status — think-tank piece. WFM Labs Erlang-O™ is an exploratory framework, not an established practice. It is proposed here as an alternative methodology to traditional Erlang-C / Erlang-A staffing math, examining what it would look like to bake interval-level overhead, forecast volatility, and intraday shrinkage directly into the staff line rather than treating them as reactive adjustments. The framework has not been empirically validated at scale; the calculator on this page is an idea companion to the math, not a vendor-grade staffing engine. This page lives in the WFM Labs think-tank space alongside the Value-Based Planning Model — both interrogate where traditional models break and propose alternatives, with Erlang-O focused on the staffing-math layer and VPM on the planning architecture above it.

WFM Labs Erlang-O™ is proposed by Ted Lango at WFM Labs to address the limitations of traditional staffing models — Erlang-C and Erlang-A — which often break down under the operational realities of today’s contact centers. As proposed, Erlang-O is not a replacement for a base staffing algorithm; it is an overlay layer that adjusts any interval-level staffing calculation for variance, volatility, and intraday shrinkage. The intent is to make staffing plans resilient to real-world conditions while opening up cost optimization and employee-development capacity.

The "O" in Erlang-O stands for "Overhead," representing the calculated capacity buffer integrated into staffing requirements to absorb natural variance and allow for real-time adjustments.

Why Traditional Models Fall Short

Contact centers have historically relied on base algorithms like Erlang-C and Erlang-A to determine the number of agents required to meet service levels. However, these models are based on theoretical assumptions that often misalign with operational reality:

  • Erlang-C assumes infinite queue capacity and no caller abandonment.
  • Erlang-A accounts for abandonment but still presumes uniform agent capabilities and steady-state arrival patterns.
  • Both models assume that demand and capacity can be tightly planned in advance.

These assumptions fail in modern contact centers characterized by:

  • Intraday fluctuations in call arrivals and handle times.
  • Real-time variability in agent availability.
  • Multi-channel operations and skill-based routing.
  • The need to balance queue performance with agent development and alternative work.

As a result, staffing plans based purely on Erlang-C or Erlang-A are fragile, requiring reactive interventions when demand deviates from forecasts.

WFM Labs Erlang-O™: An Overhead Adjustment Framework

WFM Labs Erlang-O™ does not replace base interval staffing calculations; it enhances them. Organizations may continue using Erlang-C, Erlang-A, or custom algorithms tailored to their operations. Erlang-O works alongside these models, modifying their outputs to account for operational realities.

Base staffing algorithms estimate the number of agents required for a given volume, average handle time (AHT), and service level objective. Erlang-O applies adjustments to this estimate to account for:

  • Minimal Interval Variance (MV): The statistical variability inherent in call arrivals within small time intervals. Even under accurate forecasts, arrivals are random, causing fluctuations that must be absorbed in real-time.
  • Forecast Volatility (VX): The potential for unpredictable spikes in volume or fluctuations in AHT that exceed expected minimal variance.
  • Intraday Shrinkage (IS): Time allocated to off-phone activities such as training, coaching, and breaks, along with alternative work (e.g., email or back-office tasks). Erlang-O treats these as dynamic levers to optimize both service level and agent development during the day.

Core Components Explained

1. Base Staffing Algorithm

WFM Labs Erlang-O™ begins with a base staffing calculation, often using Erlang-C or Erlang-A. However, contact centers with more complex environments may employ custom algorithms based on:

  • Skill-based routing efficiencies.
  • Concurrency in digital channels.
  • Customer patience factors.
  • Discrete event simulation (DES) models or machine learning estimators.

WFM Labs Erlang-O™ is agnostic to the base staffing algorithm. It enhances the result, regardless of the complexity of the underlying calculation.

2. Minimal Interval Variance (MV)

Minimal Interval Variance captures the inherent randomness in call arrivals at the interval level. Even with accurate forecasts, call volumes within a 15- or 30-minute period will vary. MV is calculated using:

MV = √(2 / π * Forecasted Call Volume)

This adjustment increases staffing slightly to buffer against these natural fluctuations, preventing queue performance from collapsing when arrivals spike.

3. Forecast Volatility (VX)

Volatility accounts for larger, unpredictable deviations beyond minimal variance. It reflects past patterns of demand surges, AHT swings, or unanticipated events. VX is specific to each contact center, requiring historical data analysis to determine:

  • The frequency of volatility spikes.
  • The average magnitude of those spikes.

Once assessed, VX adjustments are added to both forecasted calls and AHT.

4. Intraday Shrinkage (IS)

Traditional staffing models treat shrinkage as a pre-scheduled deduction from available capacity. WFM Labs Erlang-O™ redefines shrinkage as a real-time capacity lever:

  • Discounted Productive Shrinkage (DPS): Time for coaching, training, and breaks delivered dynamically in response to queue conditions.
  • Other Shrinkage (OS): Unplanned shrinkage (e.g., absenteeism, unproductive time).
  • Alternative Channel (AC): Time allocated to less time-sensitive work (e.g., email, back-office).

Intraday shrinkage is applied as an inflator to the final staffing requirement, reflecting the need to balance queue work and agent development.

Conceptual Staffing Adjustment Framework

WFM Labs Erlang-O™ adjusts the base staff requirement as follows:

Staff Required = Base Staffing Algorithm(Adjusted Calls, Adjusted AHT, SL) / (1 - IS)

Where:

  • Adjusted Calls = Forecasted Calls * (1 + MV + VX1)
  • Adjusted AHT = Forecasted AHT * (1 + VX2)
  • IS = DPS (discounted based on real-time delivery), OS, and AC

The Role of Real-Time Automation

WFM Labs Erlang-O™ is most effective when paired with real-time automation, allowing contact centers to:

  • Continuously monitor queue health and agent availability.
  • Dynamically deliver productive shrinkage and alternative work.
  • Optimize staffing buffers in real time.

Automation ensures that surplus capacity added through MV and VX adjustments is not wasted but used productively.

Key Benefits of WFM Labs Erlang-O™

  • Builds resilience into capacity plans by embedding variance and volatility buffers.
  • Reduces reliance on manual interventions during peak intervals.
  • Maximizes agent development time without sacrificing service levels.
  • Supports multi-channel and skill-based routing environments.
  • Enhances cost efficiency by reducing reliance on overtime and minimizing service level failures.

When to Use WFM Labs Erlang-O™

Erlang-O is ideal for contact centers operating in environments with:

  • Frequent intraday variability.
  • Skill-based routing and multi-channel complexity.
  • A desire to maximize agent training and development.
  • Real-time automation capabilities to deliver dynamic adjustments.

Future Considerations

WFM Labs Erlang-O™ is designed as a flexible framework that will continue to evolve alongside contact center technologies. Future advancements may include:

  • Machine learning for volatility detection.
  • AI-driven forecasting refinements.
  • Integration with workforce engagement tools for personalized scheduling.

As a think-tank proposal, WFM Labs Erlang-O™ asks: what would WFM look like if interval overhead, volatility, and intraday shrinkage were first-class inputs to the staff line rather than reactive adjustments? The math above is one answer. The empirical question — does it produce more resilient operations than Erlang-C with conventional shrinkage handling? — is open and is the next step for any operation willing to A/B test the approach against its current method.


Maturity Model Position

  • Level 1 — Initial (Emerging Operations) — Erlang-O is unreachable. There is no interval-level forecasting infrastructure to feed MV / VX adjustments.
  • Level 2 — Foundational (Traditional WFM Excellence) — Erlang-O is unreachable. Operations run static Erlang-C with conventional shrinkage deduction; the framework's premise (overhead as a planning input, not a deduction) is invisible from inside this paradigm.
  • Level 3 — Progressive (Breaking the Monolith) — Erlang-O is approachable. Variance Harvesting gives the operation visibility into MV; Probabilistic Forecasting gives the distributional inputs VX needs. The framework can be piloted at this level.
  • Level 4 — Advanced (The Ecosystem Emerges) — Erlang-O is operable as the staffing-math layer for traditional single-pool work. In the Three-Pool Architecture, it sits inside Pool Spec staffing as one of the candidate methods; Pool AA and Pool Collab use different math entirely.
  • Level 5 — Pioneering (Enterprise-Wide Intelligence) — Erlang-O is closed-loop: MV and VX parameters are calibrated continuously from in-house data; intraday shrinkage delivery is automated against real-time queue health.

The honest framing: at any level, Erlang-O is one option among several for the staffing-math layer. Its adoption depends on whether an operation has the Level 3 instrumentation to feed it and the appetite to A/B test against incumbent practice.

See Also

Retrieved from "https://wiki.wfmlabs.org/w/index.php?title=WFM_Labs_Erlang-O™&oldid=2581"