Peak season plans usually get one number right: how many extra agents the queue needs. The number that decides whether you get them in time is the start date of the seasonal class, and that one is often treated as an admin detail. It is not. A class that reaches the floor after the peak has started delivers about half the capacity of a class that reached it a few weeks before, so you need far more people to cover the same gap.
This guide is a companion to why September is already late for peak season, which covers lead times and the averaging trap. Here the focus is narrower: a simple ramp model that shows what the class start date does to your hiring number, the window to work back from, and what to do if you are already past it.
The setup
We modelled a seasonal class that must cover a peak of 6 weeks. The assumptions are the kind most contact centres can check against their own data:
- 12% of the class drops out during training and never reaches the floor.
- After reaching the floor, a new agent works at 40%, 55%, 70%, 80%, 88% and 94% of a fully ramped agent in weeks 1 to 6, and 100% from week 7.
- 1.5% of agents leave each week once on the floor.
For each floor start date we counted how much fully ramped capacity one hire adds across the 6 peak weeks. Then we asked a practical question: how many people do you need to hire to cover a gap of 20 fully ramped agents?
What the start date does to the hiring number
| Class reaches the floor | Capacity one hire adds over the peak | Hires needed for a 20 agent gap |
|---|---|---|
| 4 to 6 weeks before peak | about 77% of a ramped agent | 26 |
| 2 weeks before peak | about 73% | 28 |
| At the start of peak | about 60% | 33 |
| 1 week after peak starts | about 47% | 42 |
| 2 weeks after peak starts | about 35% | 57 |
| 3 weeks after peak starts | about 24% | 84 |
Read the chart in two parts. Between 6 weeks before and 3 weeks before the peak, the number barely moves, because the class is mostly ramped either way. After that, the line bends. Every week of delay costs more than the one before, because you lose the strongest weeks of the ramp and the peak keeps running.
A class that arrives on day one of peak needs about a quarter more people than a class that arrived a month early. A class that arrives 3 weeks into peak needs more than three times as many, and by then it is an HR problem, not a capacity plan.
Why the cost bends
Two things stack against a late class.
- The ramp is slow at the start. In its first two weeks on the floor a new agent works at roughly half the speed of an experienced one, and that is exactly when your queue is at its worst.
- Training pulls experienced agents off the phones. Nesting, coaching and side by side calls take your best agents away from the queue during peak.
If your handle time also rises during peak, the effect is larger. A longer handle time in peak weeks cuts the effective capacity of the whole team, as the unplanned absence guide shows for a similar reason: small shortfalls hit service level harder than the same surplus helps it.
Work back from the peak date
Industry guidance agrees on the timing. Brightmetrics suggests starting peak planning at least 6 to 8 weeks ahead, and September for the major holiday peak. Call Centre Helper notes that recruitment has to outpace attrition from August to October if the peak is to be covered.
Here is a simple way to turn the model into dates:
- Pick the date the class must be on the floor. From the table, 2 to 4 weeks before the peak is the sweet spot.
- Subtract training and nesting. If training takes 4 weeks, training starts 6 to 8 weeks before peak.
- Subtract your hiring lead time. Offers, notice periods and background checks often take 2 to 4 weeks, so offers go out roughly 8 to 12 weeks before peak.
For a peak that starts in late November, that points to late August to late September. If that date has passed, read the next section.
If you are already late
Hiring more for this peak is the weakest lever once you are past the window. The model says every extra week costs you more heads for the same result. Better options, in the order we would try them:
- Cross skill existing agents from quieter queues. They are already ramped. See multi skill pooling for how much it saves.
- Protect the team you have. Set the leave quota early and share peak week availability, so absence does not eat the capacity you planned. The effect of this is covered in the shift flexibility guide.
- Agree overtime and flex rules now, with senior leaders, before the first bad day.
- Use new hires on simple contact types first, so they add real capacity in week one instead of waiting to be fully trained.
- Keep hiring for the next quarter. Stopping now to save cost repeats the same gap later. See why a hiring freeze is not a staffing plan.
What to take to your leader
If you have to defend the date, use three lines:
- The gap in fully ramped agents for the peak.
- The hires needed at each possible start date, from the table above or your own ramp data.
- The date by which offers must go out, and what each week of delay adds.
That makes the timing visible as a cost, not a preference. For the permanent version of the same problem, see the hidden capacity cost of attrition, and for ramp time in hiring plans see AI agent assist, ramp time and your hiring plan.
Try it with your own numbers
Replace the assumptions with your own: training drop out, ramp curve, attrition and peak length. The shape of the answer holds for most operations, but the dates move. The Hiring and Class Size Planner on WFM Helper helps you size a class once you know your drop out and ramp.
Key takeaways
- A seasonal class that reaches the floor 4 to 6 weeks before peak adds about 77% of a ramped agent over a 6 week peak. One that reaches it on day one adds about 60%. One that arrives 3 weeks in adds about 24%.
- To cover a gap of 20 ramped agents you need 26 hires for an early class, 33 for a day one class and 84 for a class that arrives 3 weeks late.
- Work back from the peak: floor start 2 to 4 weeks before, training and nesting before that, and offers 8 to 12 weeks before the peak.
- If the window has passed, use cross skilling, leave rules, overtime agreements and simple first contact types before you add more hires.
The numbers in this article are model outputs based on the stated assumptions, not measured results. Use your own training drop out, ramp and attrition data for planning.