How many people does your operation really need? Demand-driven vs. fixed-post staffing

Two regimes coexist in many companies: demand-driven (headcount follows incoming volume) and fixed-post (headcount follows operational structure). Knowing which applies, and when both appear together, is the first step in any serious sizing. In both, typical shrinkage of 25%–35% must be absorbed without SLA disruption.

Demand-driven: when the queue dictates the headcount

In receptions, contact centers, retail and customer service, volume varies along the day and the week. Sizing by historical headcount is what generates peak-time queues and idle time in valleys. We size from the real demand curve, using Erlang and shrinkage, and build the shift mesh on top of that.

Queue models: Erlang C and Erlang A

Erlang C calculates the minimum number of agents so that a fraction of demand is served within a target time (e.g., 80% in 20s). Erlang A incorporates the abandonment rate, more adherent to modern contact centers. For field and retail operations, a backlog model with average handling time (AHT) per task type is used. Each model delivers a different FTE count per interval.

Fixed-post: when demand does not dictate the headcount

Not every operation forms queues. Hospital beds, production lines, gatehouses and central sterilization units have fixed posts that require continuous presence. The headcount comes from the structure (posts, shifts and schedule regime), not from the incoming volume. Here an empty post does not create waiting: it creates clinical risk, line stoppage or norm violation.

The fixed-post formula

HC = posts × shifts/day × schedule factor × (1 + shrinkage). The schedule factor depends on the regime: 1.17 on 6x1 and 1.40 on 5x2 for 7-day operations. For 24/7 posts on 12x36, the base is 4 crews per post plus relief. Typical shrinkage sits between 25% and 35% and absorbs vacations, absenteeism, training, breaks, meetings and tardiness.

Shrinkage: the cost that does not show in the headcount

Shrinkage is paid time that does not turn into service: vacations (8.33%), absenteeism (2–6%), training (1–3%), regulated breaks (1–5%), meetings and back-office (3–8%), and tardiness or early leaves (1–3%). The sum usually lands between 25% and 35%. Ignoring shrinkage is the most frequent sizing mistake, and it is what most explains 'out of nowhere' overtime at month close.

When both regimes coexist

Many operations have both regimes at the same time, and treating them with the same method is what generates chronic overtime or structural idleness. In a hospital, the ER is demand-driven, but the nursing post is fixed. In an e-commerce, customer service is demand-driven, but the DC gatehouse is fixed. The method must be hybrid.

KPIs that matter after publishing

Four minimum KPIs: (1) coverage rate per interval (planned vs. actual), (2) cost per productive hour, (3) overtime as % of base payroll, (4) schedule adherence (presence vs. plan). Monthly review for the first 6 months, quarterly thereafter, always comparing against the baseline defined before the change.