WFM Labs Risk Score

Three-component risk score: Monte Carlo, staffing gaps, and shrinkage variance.

The WFM Labs Risk Score is a composite metric developed by Ted Lango for quantifying the probability-weighted risk that a contact center will fail to meet its Service Level target in a given planning period. The score aggregates inputs from three primary uncertainty domains—forecast variance, staffing gap, and shrinkage variance—and produces a single index that reflects the overall service delivery risk profile of the operation. It is used in Capacity Planning Methods, intraday management, and executive reporting to translate complex, multi-variable uncertainty into a single actionable figure.

The Risk Score addresses a structural limitation of traditional service level reporting: historical service level performance describes what happened, while the Risk Score projects what is likely to happen given current conditions. It moves WFM governance from lagging indicator management to leading indicator management.

Conceptual Foundation

Traditional contact center planning assumes that forecast demand, staffed headcount, and shrinkage rates are known quantities. In practice, all three are estimates subject to material uncertainty. A staffing plan that appears adequate under point-estimate assumptions can produce systematic service-level failure when forecast error, shrinkage variance, and headcount gaps compound simultaneously.

The Risk Score formalizes this compounding uncertainty. Rather than asking "are we staffed to plan?"—a binary question that ignores distributional uncertainty—the Risk Score asks "what is the probability-weighted expected service level shortfall given our current uncertainty profile?" This reframing produces a risk measure that is actionable before a service-level miss occurs rather than explanatory after one.

The conceptual approach draws on multi-objective optimization principles and is positioned as a Level 3–4 capability within the WFM Labs Maturity Model. It is a component of the broader Value-Based Planning Model framework in operations that have adopted that model.

Inputs

The Risk Score aggregates three input domains, each characterized by both a point estimate and an uncertainty distribution.

Input 1: Forecast Uncertainty

Forecast uncertainty captures the historical distribution of the gap between forecast demand and actual demand at the interval or day level. It is expressed as a forecast error distribution—typically the Mean Absolute Percentage Error (MAPE) and the standard deviation of percentage error across a trailing observation window.

Operations with stable, well-modeled demand patterns exhibit narrow forecast error distributions. Operations handling high-volatility contact drivers—weather events, product outages, promotional campaigns—exhibit wide distributions. The width of the forecast uncertainty distribution is the primary determinant of the Risk Score's sensitivity to demand-side risk.

Inputs required: interval-level or day-level forecast error distribution from the preceding 4–12 weeks, segmented by day of week and time of day where pattern differences are material.

Input 2: Staffing Gap

Staffing gap captures the difference between the staffing level required to meet service level targets (the Erlang-C or simulation-based requirement) and the staffing level actually expected to be in place. The gap arises from three sources:

• Headcount shortfall — current rostered headcount is below the capacity plan requirement, typically from attrition outpacing recruiting or from Training Attrition reducing the graduating cohort below plan
• Scheduling efficiency loss — the conversion of available headcount into scheduled, on-queue staffing is imperfect; scheduling constraints, preference accommodations, or planning errors reduce effective coverage below the theoretical maximum
• Shrinkage rate variance — actual shrinkage exceeds the budgeted shrinkage allowance embedded in the staffing calculation, reducing on-queue agents below the scheduled number

Staffing gap is expressed as a percentage of required staffing: a 5% gap means the operation expects 5% fewer on-queue agents than the service-level model requires.

Input 3: Shrinkage Variance

Shrinkage variance captures the historical distribution of the gap between planned shrinkage rates and actual shrinkage rates. While the staffing gap input captures the current expected gap, shrinkage variance captures the additional uncertainty around that expectation—the probability that actual shrinkage will be worse than expected in the planning period.

Shrinkage variance is expressed as the standard deviation of the percentage-point gap between planned and actual shrinkage rates across trailing periods, segmented by day type.

Calculation Methodology

The WFM Labs Risk Score combines the three inputs through a simulation-based approach rather than a closed-form formula. The simulation proceeds as follows:

1. For each planning interval, draw a demand realization from the forecast uncertainty distribution
2. Apply the staffing gap to reduce expected on-queue agents
3. Draw a shrinkage realization from the shrinkage variance distribution to adjust on-queue agents further
4. Run the resulting (demand, agents) pair through an Erlang-C or Erlang-A model to calculate the service level that would result
5. Record whether the resulting service level meets or fails the target
6. Repeat for a large number of simulation iterations (typically 1,000–10,000)

The Risk Score is then computed as the percentage of simulation iterations that produce a service-level failure—that is, the probability of missing the service-level target given the current uncertainty profile.

The output is indexed to a 0–100 scale for reporting purposes, where 0 indicates no simulated failures (near-zero risk) and 100 indicates failure in all simulated iterations. Threshold values for risk tiers are calibrated to organizational tolerance:

• Green (0–25): Low risk; current conditions are unlikely to produce a service-level miss
• Yellow (26–60): Moderate risk; conditions warrant monitoring and contingency planning
• Red (61–100): High risk; proactive intervention is warranted before the period begins

Integration with Planning Processes

Weekly Capacity Planning

The Risk Score is calculated weekly as part of the capacity planning review cycle. A high Risk Score for the upcoming week triggers specific review actions: examination of the staffing gap root cause, escalation of recruiting urgency if headcount shortfall is the driver, and contingency planning for intraday flex lever activation.

Operations that have adopted the Risk Score as a planning tool replace the binary "are we staffed?" question in weekly planning reviews with a "what is our risk profile this week?" review that surfaces distributional uncertainty rather than point-estimate adequacy.

Intraday Application

An interval-level Risk Score variant can be calculated at the start of each interval using real-time inputs for current on-queue agents, current AHT, and actual versus forecast volume. This intraday application gives ROC analysts a forward-looking risk figure that precedes service-level failure rather than following it, enabling earlier intervention decisions.

Executive Reporting

The Risk Score translates multi-variable WFM uncertainty into a single figure accessible to operational leadership. Rather than presenting three separate metrics (forecast MAPE, headcount gap percentage, shrinkage variance), the Risk Score communicates the integrated service-level risk in probability terms that align with business decision-making. A Risk Score of 70 communicates "we have a 70% probability of missing service level this week" in a way that a +3% shrinkage variance figure alone does not.

Relationship to Other Frameworks

The Risk Score is a component of the broader Value-Based Planning Model and complements the Interior Optimum concept, which identifies the staffing level that optimally balances cost and service-level risk rather than minimizing one at the expense of the other. A Risk Score that is chronically elevated indicates the operation is operating to the right of its interior optimum—understaffed relative to the optimal cost-service balance point.

In operations using the Three-Pool Architecture, Risk Scores are calculated separately for Pool Spec and Pool Collab, as the uncertainty profiles and service-level consequences differ across pools. Pool Spec Risk Scores are typically higher due to the derived demand nature of specialist contacts, which inherits uncertainty from both Pool AA containment variance and Pool Collab escalation variance.

The Risk Score feeds directly into multi-objective optimization frameworks by providing a probability-weighted cost of service-level failure that can be balanced against staffing cost in the objective function.

Maturity Model Considerations

The WFM Labs Risk Score is a Level 3–4 capability on the WFM Labs Maturity Model. Prerequisites include systematic forecast error measurement (Level 3), simulation-based staffing capability (Level 3–4), and structured shrinkage variance tracking (Level 3).

At Level 1 and 2, service-level risk is not formally quantified. Operations rely on historical service-level performance as a proxy for current risk, which provides no forward-looking signal.

At Level 3, operations may calculate simplified risk indicators—such as the probability of missing service level based on headcount gap alone—without the full three-input simulation model.

At Level 4, the full Risk Score methodology is implemented, and the score is embedded in weekly planning reviews and executive reporting.

At Level 5, Risk Score calculation is automated and updated at interval frequency using real-time data feeds, with automated alert routing when scores cross threshold values.

Related Concepts

• Service Level
• Capacity Planning Methods
• Forecasting Methods
• Shrinkage
• Erlang-C
• Erlang-A
• Value-Based Planning Model
• Three-Pool Architecture
• Interior Optimum
• Multi-Objective Optimization in Contact Center
• WFM Labs Maturity Model