Call Center Shrinkage, Occupancy & FTE Staffing Planner

1. The Workforce Management Dilemma: Bridging Erlang C to the Real-World Floor

In telecommunications and contact center operations, queueing theory formulas such as Erlang C or Engset calculate the theoretical minimum number of “bodies in chairs” (N_net) required to answer incoming calls within a defined Service Level Agreement (SLA). Erlang C assumes that every assigned server is 100% physically available, logged into the Automatic Call Distributor (ACD), wearing a headset, and ready to accept calls every second of the half-hour scheduling interval.

In reality, human contact center employees cannot remain glued to telephone queues for eight consecutive hours. Employees must take statutory rest breaks, eat lunch, participate in daily team huddles, receive quality calibration coaching, complete compliance training, take paid vacations, and occasionally stay home due to illness.

Shrinkage (S) is the percentage of paid employee time during which staff are unavailable to handle customer contacts despite being rostered on payroll. Without a precise shrinkage cushion, an operation that dimensions staff purely to Erlang C output will collapse into perpetual queue backlogs and severe SLA failures the moment the first agent steps away for a scheduled rest break.

2. Internal vs. External Shrinkage: Slicing the Loss Categories

Workforce planners categorize shrinkage into two operational domains based on whether the loss occurs on-site or off-site:

• External (Off-Queue / Out-of-Office) Shrinkage: Represents time when an employee is absent from the contact center facility or disconnected from virtual telework systems. Key components include Paid Time Off (PTO / Vacation), statutory public holidays, sick leave, short-term disability, jury duty, bereavement, and unexcused absenteeism. External shrinkage typically accounts for 18% to 24% of annual hours and is primarily managed by Human Resources and long-term capacity planning teams.
• Internal (On-Queue / In-Office) Shrinkage: Represents time when an employee is clocked in and physically present, but unavailable to take customer calls. Key components include paid 15-minute rest breaks, 1-on-1 supervisory coaching, team meetings, software glitches or desktop reboots, customer callbacks, and special projects. Internal shrinkage typically accounts for 12% to 18% of the paid shift and is actively monitored by Real-Time Analysts (RTAs) to protect intra-day intervals.

3. The Mathematics of Shrinkage: Linear vs. Compounded Formulations

A frequent debate among workforce management architects centers on whether to combine internal and external shrinkage additively or multiplicatively:

The linear model simply sums the percentages (e.g., 15% internal + 20% external = 35% total shrinkage). While widely used due to simplicity, it contains a minor theoretical flaw: an agent who is absent on vacation cannot simultaneously take an internal coffee break or attend a team meeting.

The compounded model evaluates successive survival probabilities of available time. If external shrinkage is 20%, only 80% of annual time remains. If internal shrinkage is 15% of that remaining time, available time is 0.80 × 0.85 = 0.68 (68%), yielding a compounded shrinkage of 32.0%. In an enterprise operation of 500 agents, this 3.0% discrepancy represents 15 FTEs—translating to over $675,000 in annual payroll variance.

4. The Correct Gross Rostering Formula: Why Multiplying by (1 + S) Fails

One of the most destructive and widespread errors in contact center management is attempting to calculate gross headcount by multiplying net agents by (1 + S):

Suppose an Erlang C calculation requires 100 net agents in chairs during peak hour, and total shrinkage is 30%. Using the erroneous multiplication formula:
Gross Staff = 100 × (1 + 0.30) = 130 agents.
If 130 agents are rostered and 30% shrinkage occurs on the day of operation:
Absent Agents = 130 × 0.30 = 39 agents.
Actual Agents Remaining in Chairs = 130 - 39 = 91 agents.
The contact center experiences an immediate 9-agent deficit, causing queues to spike and SLAs to collapse.

Because shrinkage is defined as a proportion of the gross scheduled payroll, the equation must divide by the available factor:
Gross Staff = 100 / (1 - 0.30) = 100 / 0.70 = 142.86 → 143 scheduled FTEs.
Validating the math: 143 × 0.30 = 42.9 absent, leaving exactly 143 - 42.9 = 100.1 agents in chairs, perfectly fulfilling the Erlang requirement.

5. Agent Occupancy vs. Shrinkage: Preventing the Burnout Spiral

While shrinkage measures unavailable time, Agent Occupancy (ρ) measures how intensively an agent works while logged in. It is defined as the percentage of logged-in time spent handling customer interactions (Talk Time + Hold Time + After-Call Work) versus sitting idle waiting for a call:

When shrinkage is under-budgeted, fewer agents arrive on the floor than required. To maintain service levels, the ACD routes incoming calls back-to-back with zero idle buffer time. As occupancy rises beyond 85% to 90%, cognitive fatigue sets in. Research consistently confirms that sustained occupancy above 88% causes sharp spikes in unplanned sick calls, higher handle times (as exhausted agents stretch wrap-up codes to catch their breath), and rapid employee turnover—creating a vicious cycle of rising shrinkage and degrading service.

6. Omni-Channel Concurrency & Shift Multipliers

In modern digital contact centers, agents frequently handle non-voice channels where concurrent interactions are possible. Live webchat agents typically handle 2 concurrent sessions, while asynchronous messaging agents (e.g., WhatsApp, SMS, in-app chat) can manage 3 or 4 simultaneous threads. Concurrency effectively scales agent throughput, allowing a smaller net roster to support high customer volumes.

Additionally, for contact centers operating beyond a standard 5-day business week (such as 7-day extended support or 24/7 operations), rostered headcount must be multiplied by a weekend coverage factor. Because a full-time employee works 5 days per week, covering a 7-day continuous schedule requires a baseline multiplier of 7 / 5 = 1.40×, before factoring in shift rotation overlaps.