Demand calculation
Demand calculation is the foundational supply-and-demand math of capacity planning — the calculation that converts a forecasted volume of work into a base FTE requirement. It is the demand half of the capacity equation; paired with workforce supply modeling, it produces the cost-bearing capacity plan. See Capacity Planning Methods for where this calculation sits in the full planning workflow.
The standard formula
A simple demand calculation leveraged across the industry to generate an approximate base FTE required is as follows:
Base FTE Required = Volume Offered × AHT / 3600 / Total Annual Work Hours / Occupancy / (1 − Shrinkage)
This calculation establishes a base FTE needed to service an annual workload — a budget anchor and a starting point for the cost stack. Reading the formula left to right: total seconds of work demanded, divided by seconds per agent-hour, divided by hours per producing-FTE, grossed up by occupancy and shrinkage to convert producing time into paid time. The result is the count of paid headcount required to deliver the demanded work over the year.
The variables
Each variable in the formula carries assumptions practitioners get wrong in predictable ways.
- Volume Offered — the count of contacts offered to the queue, not forecasted, not handled, not answered. Forecast volume is what you predicted; offered volume is what would have arrived if the queue were infinite. Most demand-calculation errors trace to substituting forecasted volume for offered. Multi-channel operations sum offered across channels — voice, chat, email, async — each with its own series.
- AHT — average handle time per contact, in seconds. The math wants handle time (talk + meaningful after-call work), not gross paid time. Putting non-handle time into AHT and again into shrinkage double-counts and over-staffs the plan.
- Total Annual Work Hours — paid hours per producing-FTE per year. Common values: 2,080 (52 × 40, US baseline); 1,840 after PTO and statutory holidays. Pitfall: using 2,080 and including PTO in shrinkage double-counts the same time. Pick a definition and align shrinkage to it.
- Occupancy — fraction of paid-and-on-the-phone time that is actually handle time. Common values: 0.80–0.85 voice; higher back-office; lower premium service. Erlang queueing dynamics make occupancies above ~0.90 produce service-level collapse — the bound is real, not aspirational.
- Shrinkage — fraction of paid time that is not available to the queue. PTO, training, meetings, coaching, breaks, system downtime, adherence variance. Common values: 0.30–0.40. Define components mutually exclusively or the math is wrong.
The supply-and-demand framing
Demand calculation is one half of capacity planning. The math above produces required producing-FTE. The cost-bearing question is required paid headcount, which depends on attrition, ramp, training duration, and proficiency drag. Workforce Cost Modeling is the unified treatment: required producing-FTE flows from demand calculation; required paid headcount flows from supply modeling on top. A plan with only the demand half consistently underdelivers.
Relationship to interval-level Erlang staffing
The annual base-FTE formula is sufficient for budgeting and hiring plans. It is not sufficient for execution-level staffing, which requires interval-by-interval Erlang-derived math. Reasons: volume is not uniform across intervals; service-level commitments are made at the interval level, not annual; Erlang models the queue dynamics directly (arrival rate, AHT distribution, abandonment) and the annual math has no notion of any of this.
The standard practice: annual base-FTE math sets the budget envelope; interval-level Erlang sets the schedule. The two should agree to within a few percent. The Power of One discipline is the bridge — at the interval level, single-agent presence dominates service-level outcomes, and the annual math cannot represent that sensitivity.
Probabilistic extension
The formula takes Volume Offered as a point estimate. In production, volume is a distribution. The probabilistic extension propagates that distribution through to the FTE answer; the output is no longer "we need 412 FTE" but "we need 380–445 FTE depending on volume realization, with 80% confidence within 395–430." The plan can then commit at a stated confidence level. Probabilistic Forecasting is the input methodology; the WFM Labs Risk Score™ formalizes plan-risk rating. The Level 3 capacity-planning shift is moving from point-estimate demand calculations to distributional ones.
Common failure modes
- Forecast volume instead of offered. Forecast is what you predicted; offered is what arrived.
- Gross AHT instead of handle-time-only. Including after-call work, hold, and idle in AHT, then again in shrinkage. The plan over-staffs.
- Inconsistent component definitions. PTO inside work-hours and inside shrinkage; meeting time inside AHT and inside shrinkage. Define mutually exclusively.
- Aspirational occupancy. Setting occupancy at 0.90+ because "we want to be efficient." Erlang queueing dynamics make this infeasible in execution.
- Single-channel math for a multi-channel operation. Voice + chat + email + async each have their own profile.
- Point estimate as truth. One number from a stochastic process committed as the plan, with no confidence band.
- Ignoring the supply side. Producing-FTE without supply-side gross-up underdelivers in execution.
Interactive tool
For a single-formula scenario, this page is enough. For full cost-stack capacity planning — adding turnover, training, and ramp overhead on top of the base FTE — use The Spot Capacity Calculator. The calculator extends this base demand calculation with annual attrition, training weeks, training attrition, day-1 AHT factor, and months to proficiency, so the FTE figure reflects what the operation must actually hire — not just the steady-state minimum.
Maturity Model Position
Demand calculation is foundational math; what changes across maturity levels is what an organization does with the answer.
- Level 1 — Initial (Emerging Operations) — demand math may not be done formally; headcount is set by precedent or budget pressure.
- Level 2 — Foundational (Traditional WFM Excellence) — the standard formula is used; a single-number answer drives the budget. Inputs are point estimates; component definitions are often inconsistent. The plan is committed as truth.
- Level 3 — Progressive (Breaking the Monolith) — inputs become distributions (Probabilistic Forecasting); output reports a range. The WFM Labs Risk Score™ rates plan risk. Component definitions are mutually exclusive and audited. Plan commits at a stated confidence level.
- Level 4 — Advanced (The Ecosystem Emerges) — demand calculation is part of a bottom-up value-based plan across the Three-Pool Architecture. Demand is segmented by routing destination (human, hybrid, AI-with-supervision); rebound and escalation tax modify the demand math.
- Level 5 — Pioneering (Enterprise-Wide Intelligence) — demand calculation runs continuously and at multiple resolutions across the enterprise; practitioner role shifts to oversight.
The lower-level form is a single number committed as truth. The higher-level form is a distribution, segmented by routing destination, audited for component consistency, and committed at a stated confidence level.
See Also
- Capacity Planning Methods — the master reference for capacity-planning techniques
- Workforce Cost Modeling — the supply-side companion; the unified cost-per-producing-FTE frame
- Forecasting Methods — the methods that produce the demand forecasts this calculation consumes
- Probabilistic Forecasting — distributional forecasts for the probabilistic demand calculation
- Power of One — interval-level sensitivity the annual math cannot represent
- WFM Labs Risk Score™ — risk-rating methodology for capacity plans built on this math
- Naive and Seasonal Naive Forecasting — the baseline benchmark for any volume forecast
- Exponential Smoothing — most common method used by WFM software for volume forecasting
- ARIMA Models — when ETS isn't enough for the forecast feeding demand calculation
- Annual Attrition, Training Attrition, Onboarding Costs, Length of Training, Speed to proficiency curve — supply-side components that gross up producing-FTE to required paid headcount
