WFM guide

Peak staffing for contact centres

Q: How do you staff a contact centre for peak periods?

Contact centre peak staffing requires five steps: (1) Forecast peak volume using prior year peak actuals adjusted for growth; identify contact-to-order ratio for demand-linked peaks (e-commerce) or event-type for campaign-linked peaks. (2) Model peak staffing by running Erlang C on peak interval volumes; the staffing requirement is non-linear (doubling volume more than doubles agents needed due to Erlang's queuing physics). (3) Assess the gap vs. normal staffing to determine how many additional agents are needed above the normal rostered team. (4) Design a flex strategy. The gap sources, in priority order: voluntary overtime from existing team, agency/temporary staff trained ahead of peak, outsourced overflow partner, reduced non-phone activities. (5) Test the plan by simulating peak scenarios in your WFM tool and confirming minimum staffing floors are met across the peak period.

Q: How do you forecast contact volume for a peak event?

Peak event volume forecasting methods: (1) Historical same-event comparison: use prior year peak actuals for the same event. Apply a year-on-year growth factor. (2) Uplift modelling: for promotional or campaign-linked peaks, model expected volume as baseline × peak uplift factor (e.g. Black Friday = 8–15× normal Monday for e-commerce). (3) Demand-linked forecasting: for WISMO (where-is-my-order) and delivery contacts, forecast based on dispatch volume × historical contact-to-order rate. (4) Scenario planning: model best case, expected case, and worst case. Staff to expected; prepare contingency for worst case. Do not staff to best case, because underestimating peak demand produces SL failures that drive further contacts.

Q: What flex staffing options are available for contact centre peaks?

Flex staffing options for peaks, in typical deployment order: (1) Voluntary overtime: ask existing trained staff to work additional shifts; cost-effective and quality-safe since agents know the product. (2) Temporary/agency staff: hired months in advance of peak, trained and nesting alongside permanent staff 4–8 weeks before the event. (3) Outsourced overflow partner: contacts routed to an outsourced partner during peak; lower quality risk if partner is briefed and trained in advance; higher quality risk if partner receives warm transfer without preparation. (4) Callback / virtual queue: reduce the demand for simultaneous agent availability; does not reduce contact volume but smooths the intraday distribution. (5) Extended operating hours: open earlier and/or close later to spread the same volume across more hours; reduces peak interval intensity.

Q: How much staff do you need for a 2× volume peak?

Due to Erlang C's queuing mathematics, a 2× volume peak requires significantly more than 2× the normal agent count. This is because the relationship between volume and agents is non-linear at higher utilisation levels. Example: if 10 agents handle 100 calls/hour at 80% occupancy with a 20-second ASA, then 200 calls/hour might require 22–24 agents (not 20) to maintain the same service level, because the marginal agent at higher volume has a disproportionate queue-clearing effect. The exact multiplier depends on AHT, target SL, and current occupancy. Always re-run Erlang C at peak volume rather than simply doubling normal staffing.

Peak periods are the moments when WFM planning is most visible, and most punishing when it fails. A well-planned peak runs within SL at sustainable occupancy; a poorly planned one produces queue failures, burnout, and permanent reputational damage. The maths is unforgiving: Erlang C is non-linear, and a 2× volume spike needs more than 2× the agents.

Types of contact centre peaks

Seasonal peaks

Examples: Christmas (retail), summer (travel/hospitality), tax season (financial services), winter (utilities/energy)

Forecast method

Prior year actual for same period, adjusted for growth factor. Build month-by-month seasonal index from 2–3 years of history.

Required lead time

3–6 months for temp recruitment and training

Key risk

Underestimating growth year-on-year: if order volumes grew 20%, contact volumes likely grew proportionally

Promotional/campaign peaks

Examples: Black Friday, flash sales, product launches, marketing email campaigns, TV advertising spots

Forecast method

Baseline × uplift factor from prior similar campaigns. For new campaigns: commercial team's response rate × expected contact-per-response ratio.

Required lead time

2–4 weeks for internal flex; 6–8 weeks for temp staff trained in advance

Key risk

Marketing not informing operations of campaign timing: contacts spike before staffing plan is activated

Disruption events

Examples: System outages, product recalls, billing errors, delivery failures, adverse weather

Forecast method

Cannot be forecast in advance. Must use tiered response escalation (Tier 1: 1–3×, Tier 2: 3–8×, Tier 3: 8×+) triggered by real-time volume monitoring.

Required lead time

Zero: must be managed with pre-approved contingency playbook

Key risk

No documented response playbook; operations improvise during crisis and staff recovery is slower

Growth-driven peaks

Examples: Gradual increase in customer base outpacing planned headcount growth

Forecast method

Trend analysis from weekly volume data. Trigger headcount approval when forecast shows planned staffing will breach SL target within 8 weeks.

Required lead time

8–12 weeks from approval to trained agent on phones

Key risk

WFM team identifies the growth trend but headcount approval cycle is too slow, so SL declines while waiting for approvals

Why 2× volume needs more than 2× agents

Erlang C non-linearity at peak: worked example

Calls/hourAHTTarget SLAgents requiredvs. 100/h baseline

1005 min80/2010—

1505 min80/20161.6×

2005 min80/20222.2×

3005 min80/20343.4×

5005 min80/20575.7×

At 2× volume (200 vs. 100 calls/hour), you need 2.2× agents, not 2×. At 5× volume (500 calls/hour), you need 5.7× agents. The multiplier grows with volume because each additional agent at high occupancy has a disproportionate queue-clearing effect. Always run the actual Erlang calculation for peak volume rather than scaling headcount proportionally.

Flex staffing strategy for peaks

A peak flex strategy should work in layers: deploy each option in priority order, from highest quality to lowest quality at the margin.

Voluntary overtime (existing trained staff)

Quality: HighestCost: 1.5–2× hourly rateLead time: 2–4 weeks notice; depends on volunteer take-up

Existing agents know the product. Quality and AHT are identical to normal operation. Take-up is predictable only with a good relationship with the team; do not rely on volunteers if engagement is low.

Temporary staff (trained 4–8 weeks ahead)

Quality: Good (after training)Cost: Agency margin (20–35% uplift on hourly rate)Lead time: Start recruitment 12–16 weeks before peak; training 4–8 weeks before

Temp agents trained to 80–90% productivity before peak. AHT may be 10–20% higher than permanent staff during first month. Must be included in ramp FTE calculations.

Outsourced overflow partner (pre-briefed)

Quality: ModerateCost: Per-minute or per-contact rate (typically higher than in-house cost)Lead time: Overflow routing activated in real time; partner preparation 4–6 weeks ahead

Quality depends on briefing depth and monitoring. Works well for routine contacts; avoid routing complaints or vulnerable customers to an overflow partner without specific training.

Callback / virtual queue

Quality: No quality riskCost: Platform cost only; no additional headcountLead time: Can be enabled within hours if IVR supports it

Reduces the peak demand for simultaneous agent availability; does not reduce total contact volume. Customers call once, leave a callback request, and agents work through the queue. Effective for non-urgent contacts.

Reduced non-phone activities during peak intervals

Quality: Operational trade-offCost: None (resource reallocation)Lead time: Hours; internal decision

Pause back-office work, training, team meetings, and coaching during peak intervals; agents available on phones instead. Short-term only, because deferred back-office work creates a secondary post-peak problem.

Disruption response tiers

Unforecast volume spikes require a pre-approved playbook. Document this before the peak season so that decisions are made on policy, not in crisis.

Tier 1

1–3× normal volume

→ Activate voluntary overtime shift via text/app notification to trained pool
→ Move back-office agents to inbound phones
→ Pause non-urgent training and coaching
→ Real-time WFM dashboard alert to operations manager

Tier 2

3–8× normal volume

→ Activate overflow routing to outsourced partner
→ IVR message: estimated wait time and callback option
→ WFM analyst takes over intraday management full-time
→ Operations manager notified; service recovery comms plan activated

Tier 3

8×+ normal volume (major incident)

→ IVR: suspend non-critical contact types; route only emergency contacts
→ Email-only or callback-only for all other contact types
→ Incident management team convened; senior leadership notified
→ External communications drafted; FAQ on website to deflect repeat callers

Peak staffing questions

How do you staff a contact centre for peak periods?

Five steps: (1) Forecast peak volume using prior year actuals adjusted for growth. (2) Run Erlang C on peak interval volumes to determine staffing requirement. (3) Assess the gap vs. normal staffing. (4) Build a flex strategy: voluntary overtime → temp staff → outsourced overflow → callback → reallocation from non-phone activities. (5) Test the plan in your WFM tool against multiple volume scenarios.

How do you forecast contact volume for a peak event?

For seasonal peaks: prior year actual for the same event, adjusted for growth. For promotional peaks: baseline × uplift factor from prior campaigns. For demand-linked peaks (WISMO, delivery contacts): dispatch volume × contact-to-order rate. Always model best/expected/worst case, staff to expected, and prepare contingency for worst.

What flex staffing options are available for contact centre peaks?

In priority order by quality: (1) Voluntary overtime from existing trained staff. (2) Temporary/agency staff hired and trained 4–12 weeks before peak. (3) Outsourced overflow partner (pre-briefed and tested). (4) Callback/virtual queue (smooths intraday demand without adding headcount). (5) Reallocation from non-phone activities (short-term only).

How much staff do you need for a 2× volume peak?

More than 2×, due to Erlang C's non-linear queuing mathematics. At 2× volume, expect to need approximately 2.2× agents to maintain the same service level. At 5× volume, expect 5.7× agents. Always run the actual Erlang calculation for peak volume rather than scaling headcount proportionally.

Calculate your peak staffing requirement

Run Erlang C on your peak interval volume to get the precise agent count for any peak scenario. Don’t estimate. The non-linearity makes rough calculation meaningfully wrong at high volumes.

Erlang C calculator →Headcount projection →