When a floor is chronically short, the diagnosis is almost always "we cannot hire fast enough." Sometimes that is true. More often the headcount model itself was wrong from the start, and recruiting is being asked to close a gap that arithmetic created.
Two numbers cause most of it: occupancy and shrinkage. They are widely reported and widely misunderstood.
What each one actually measures
The distinction matters because they fail in opposite directions. High occupancy looks like efficiency and is usually a warning. Underestimated shrinkage looks like nothing at all until the floor is short every single week.
- Occupancy — of the time an agent is logged in and available, what share is spent handling contacts. It excludes breaks, training and meetings. It answers: how hard are the people who are on the phone working?
- Shrinkage — of the total paid hours you buy, what share is not available for handling contacts. Holidays, sickness, breaks, training, coaching, meetings, system downtime. It answers: how much of what I pay for actually reaches the queue?
The occupancy band nobody wants to hear
Sustained occupancy above roughly 85 to 90 percent is not efficiency. It is a queue with no recovery time in it, and it converts directly into attrition — usually with a lag of two to three months, which is exactly long enough for nobody to connect the two.
What happens operationally is that agents lose the seconds between contacts. Those seconds are where notes get written, where a difficult call gets shaken off, where an agent asks a colleague a question. Remove them and quality degrades first, then people leave. Then occupancy rises further on the remaining agents, which is the loop that turns a staffing gap into a staffing crisis.
Below about 70 percent sustained, you are carrying more heads than the volume needs — which is a cost problem rather than an operational one, and a much easier problem to have.
Shrinkage is where headcount models break
Ask an operator what shrinkage they plan for and the answer is often a single number carried over from a previous year. Real shrinkage moves seasonally, differs by site and by tenure, and is systematically underestimated because the small components are easy to forget.
A realistic build looks less like one number and more like a stack:
- Planned leave and public holidays — varies significantly by country, which is why an offshore or nearshore site cannot inherit your domestic assumption.
- Sickness and unplanned absence — higher in the first 90 days than at tenure, which matters enormously during a ramp.
- Breaks and legally required rest periods.
- Training, coaching, calibration and team meetings — the components most often left out entirely.
- System downtime and shift handover friction.
The practical consequence: if your model assumes shrinkage in the low twenties and reality is in the mid thirties, you are structurally short by more than a tenth of your floor before a single person resigns. No recruiting function can out-hire a modelling error.
Compare outsourcing against staffing before you commit.
We can map the seat count, hiring calendar, and replacement plan that fits your call center.
The number that ties it together
Required headcount is not contacts divided by capacity. It is the staffing an Erlang calculation gives you for your service-level target, divided by (1 minus shrinkage), plus a buffer for attrition during the period.
Two things follow from that, and both are routinely missed. First, service level is a choice with a price — moving from answering most contacts quickly to answering nearly all of them quickly costs disproportionately more headcount, because the last few percent are the expensive ones. Second, attrition has to be in the model as a rate, not as an event. A floor with meaningful annual attrition is hiring continuously just to stand still, and a hiring plan that only funds growth will always be behind.
Our /tools/staffing-calculator runs this arithmetic against your own volume and shrinkage assumptions rather than generic ones. If the diagnosis is that your cohorts keep landing under-sized rather than that the model is wrong, /insights/call-center-understaffed covers the calendar side of the problem.
Providers in our group
Alongside the providers above, the following companies are part of our own group. We are listing them because they are relevant options, and marking them because you should know the relationship before weighing them against the independent providers on this page.
Thirteen of the fifteen are group companies; the remaining two are independent and are marked where they appear. Both of those are larger than anything in our group, so if your requirement is global multilingual delivery under one contract they remain the realistic shortlist.
- Global Empire Corporation: Healthcare, finance, customer support, back office
- Intelemark: B2B appointment setting & lead generation
- Call Motivated Sellers: Real estate outbound calling
- Customer Communications Corp: Scalable omnichannel customer support
- Call Center Staffing: Rapid agent deployment & seasonal scaling
- B2B Appointment Setting: SMB outbound sales & pipeline growth
- Contact Center USA: US-based call center services
- Call Center Communications: Large-scale enterprise BPO
- Business Process Outsourcing: Global CX & digital customer engagement
- Canada Contact Centre: Enterprise process transformation
- B2B Telemarketing: IT + BPO hybrid outsourcing
- Telemarketing Services: AI-driven process automation
- Appointment Setting: Digital-first outsourcing
- Teleperformance (independent): Telecom & IT-enabled services
- Concentrix (independent): BPO & digital CX
