Hospitality contact centre staffing
Hospitality contact centres face some of the steepest seasonal volume swings of any industry — and some of the most operationally diverse queues. A hotel group may handle sub-3-minute room availability checks alongside 40-minute group event quotations in the same team. Getting the staffing model right means segmenting queues, not blending AHT.
Queue types and staffing models
Different hospitality queues require different capacity models. Using Erlang C for group booking enquiries (which should be a backlog) is as wrong as using a backlog model for an inbound reservations queue with a seconds-based SLA.
Hotel reservations (inbound)
High variation around group bookings; bimodal AHT common
Guest services / concierge
Lowest AHT for simple requests; complaints extend to 20+ min
Complaints and feedback
Voice complaints need dedicated queue; email SLA 4–24h
Group / event bookings
Rarely real-time; managed as case queue with assigned agents
Tour operator (voice)
High AHT for itinerary complexity; very seasonal (Q1 booking peak)
Restaurant booking line
Short, transactional; evening peaks sharper than hospitality hotel lines
Disruption / IRROPS
5–20× spike; pre-agreed tier escalation, not Erlang C modelling
Bimodal AHT: the hospitality staffing trap
Hospitality reservations lines frequently have two distinct contact types with very different handle times. Using the blended average produces wrong headcount for both pools.
Bimodal AHT example — hotel reservations line
Simple contacts (60% of volume)
5 min
Availability check, price enquiry, loyalty points
Complex contacts (40% of volume)
22 min
Group booking, itinerary, complaint, special requirements
Blended average (wrong)
11.8 min
0.6×5 + 0.4×22 = 11.8 — feeds wrong Erlang C for both pools
Wrong: blended AHT → Erlang C
Using 11.8 min AHT at 120 calls/hour, 80/20 SL, 35% shrinkage:
47 agents
Fails both simple (overstaffed) and complex (underpopulated) pools
Correct: per-pool Erlang C
72 calls at 5 min → 15 agents. 48 calls at 22 min → 26 agents. +shrinkage:
55 agents
8 fewer agents — the blended model was overstaffing by 17%
When bimodal is operationally separable
If complex bookings can be handled by a specialist team (groups and events desk), route them separately and run independent Erlang C models. This produces the most accurate headcount and allows you to hire specifically for complex-booking skills — which typically command a small salary premium.
Seasonality and volume events
The booking curve and the travel curve are not the same. WFM teams must plan for both — and maintain a separate contingency plan for unplanned disruption events.
Summer school holidays (UK: July–Aug)
1.6–2.4×Peak booking volume falls 6–10 weeks before travel — May–June is the enquiry spike
Easter holiday period
1.3–1.7×Enquiries peak in January–February; contact spike occurs during travel days
Christmas and New Year
1.4–2.0×Dual spike: booking in October–November, travel-day contacts in late December
Early May bank holiday
1.2–1.5×Short notice bookings; 2–3 week booking curve vs. 10 weeks for summer
Transport strike or airline grounding
5–20×Unplanned; requires Tier 2–3 contingency activation within 2 hours
Hotel/property closure or overbooking error
3–10×Affects specific property guests; targeted callback campaign reduces spike duration
Disruption response: three-tier plan
Hospitality contact centres should maintain a pre-agreed tiered disruption response. Activation thresholds must be agreed in advance — day-of governance approval during a 5× surge is operationally impossible.
Recall agents on RDO (voluntary first). Cancel planned training. Supervisor covers admin tasks only. Activate all callbacks and IVR deflection.
Activate outsourced/agency flex contract. Cross-trained internal volunteers from other departments. Reduce in-scope contacts to urgent-only. Proactive outbound to affected customers to reduce inbound spike.
Email-only channels during peak hours — suspend non-essential voice. Senior leadership communications owning customer-facing messaging. Dedicated response queue separate from BAU. Post-event debrief and shrinkage reconciliation.
Shrinkage profile: hospitality specifics
Hospitality contact centres have higher-than-average shrinkage during peak periods because short-term sickness tends to correlate with seasonal illness peaks — which often coincide with operational peaks.
Short-term sickness
6–10%Typically peaks in January (flu season) which is also a busy booking period for summer travel. Plan for up to 12% in January.
Annual leave clustering
8–15% (seasonal)School holiday leave requests cluster in July–August — exactly when volume peaks. Leave policy must cap approvals at sustainable shrinkage thresholds.
Training and onboarding
4–8%Pre-peak hiring (spring recruitment for summer peak) creates a training shrinkage spike in April–May. Schedule system and product training before the booking curve rises.
Meeting and briefing
3–5%Team briefs about property updates, promotions, and seasonal offers run frequently in hospitality. Keep to 15 minutes maximum during peak periods.
Hospitality staffing questions
What staffing model works for hotel reservations contact centres?
Erlang C is the right model for inbound reservations queues targeting a seconds-based service level (e.g. 80% in 20 seconds). It assumes steady-state queuing — valid for most reservation periods. Email and group booking enquiries use a backlog model (hours-based response target), not Erlang C. Abandoned calls must be tracked separately — Erlang C does not model caller patience.
How does seasonality affect staffing in hospitality contact centres?
Hospitality faces some of the most extreme seasonal patterns of any industry. Summer (July–August) typically produces 1.6–2.4× baseline January volume for UK leisure operators. Critically, the booking curve creates a second peak: volumes peak 6–12 weeks before the travel peak — in May–June for summer. WFM teams must plan for both the booking peak and the travel-day contact peak.
What is bimodal AHT and why does it matter for hospitality staffing?
Bimodal AHT describes a queue where handle times cluster at two distinct values. In hospitality, simple contacts (3–5 min) and complex contacts (15–35 min) coexist. Using the blended average produces wrong Erlang C for both pools. The correct approach: segment by contact type, run per-pool Erlang C, then sum agents. This typically adds 5–12% to headcount vs. blended modelling — but removes systematic over/understaffing.
How should hospitality contact centres plan for disruption spikes?
Disruption events (transport strikes, severe weather, overbooking) can produce 5–20× volume in under 2 hours with little warning. No capacity model predicts this — it requires a pre-agreed tiered contingency plan. Define 3 tiers with specific volume multiplier thresholds and pre-approved response actions (agency flex, channel restriction, proactive outbound). Activation must be a day-of operational decision, not a governance approval.
Model your hospitality reservations queue in Turnella
Run Erlang C for your inbound reservations line. Set separate seasonal multipliers for peak and shoulder periods. No account required.
Related guides
Contact centre staffing hub
All industries and staffing models
Retail staffing guide
Comparable seasonal surge patterns
Shift design guide
Building rosters around seasonal peaks
Call abandonment rate
Managing abandonment during surges
Multi-skill routing
Routing simple vs. complex contacts
Forecast accuracy (WAPE)
Measure seasonal forecast quality