WFM guideAdvanced planning

Contact centre multi-site planning

Q: Should multi-site contact centres measure service level per site or across all sites?

Multi-site contact centres should measure service level at the overall queue level (across all sites handling that contact type) as the primary metric, with per-site breakdowns used for diagnostic purposes only. The reason is that in a pooled multi-site operation, contacts are routed to whichever site has available capacity — so an individual site's SL at any moment is a function of routing priority and queue state, not of that site's staffing quality. Reporting per-site SL as if sites were independent creates false variance: a site handling overflow during another site's staffing shortage appears to have good SL while the sourcing site appears to have poor SL — but both are actually functioning correctly within the routing design. If sites are genuinely operating independently (not sharing contacts via routing), per-site SL is valid, but this is rarely the most efficient operating model.

Multi-site planning adds a layer of complexity that single-site models cannot handle: different time zones, site-specific holiday calendars, cross-site routing rules, and the question of whether service level is measured at the total queue level or per site.

Centralised vs. distributed WFM planning

Centralised planning model

A single WFM team manages the forecast, capacity plan, and schedule for all sites. Sites receive their staffing requirements and schedules from the central team.

+Consistent methodology and assumptions across all sites

+Cross-site optimisation is possible — surplus at one site can offset shortage at another

+Single source of truth for service level reporting

−Central team may lack site-specific knowledge (local holidays, site patterns, agent preferences)

−Communication lag between central team and site operations managers

−Site operations managers have less WFM ownership — may not engage with planning inputs

Distributed planning model

Each site has its own WFM resource (analyst or team). Sites plan independently; a central function provides methodology governance and consolidates reporting.

+Site-specific knowledge embedded in the planning process

+Faster response to site-specific operational changes

+Operations managers have direct access to the planning team

−Inconsistent methodologies across sites — assumptions differ, benchmarking is difficult

−Cross-site optimisation requires coordination between separate teams

−Higher total WFM headcount cost than a centralised model

Hybrid recommendation: For 3–5 sites with shared contact types, a centralised planning team with embedded site coordinators (one per site) typically provides the best balance: consistent methodology from the centre, site-specific knowledge from the coordinator. The coordinator feeds local knowledge into the central model and translates planning outputs back to the site operations team.

Cross-site routing: how contacts are allocated across sites

In a pooled multi-site operation, contacts are routed to agents based on availability and skill, regardless of site. The routing engine does not know that an agent is at Site B rather than Site A — it only knows whether the agent is available and proficient for the contact type.

Fully pooled routing

How it works

All agents at all sites are in the same routing group. A contact is offered to the first available proficient agent regardless of site.

Best for

Sites with identical skills and no geographic specialisation required. The most efficient model from a staffing perspective — full pooling captures the pooling efficiency benefit of the combined headcount.

Consideration

Requires consistent quality standards and processes across all sites. QA calibration across sites is essential.

Preferred site with overflow

How it works

Contacts are offered to Site A agents first (the preferred site). If no agent is available after a defined wait threshold, the contact overflows to Site B.

Best for

Sites with different language capabilities or specialist knowledge. Primary site handles most of the volume; secondary site handles overflow.

Consideration

The overflow threshold must be set carefully — too short and Site B receives too much overflow; too long and customers wait unnecessarily before overflow activates.

Site-allocated contact types

How it works

Different contact types are allocated to specific sites. Site A handles product queries; Site B handles complaints; Site C handles retention.

Best for

Sites with genuine specialisation by contact type. Each site is staffed for its allocated contact types independently.

Consideration

Less flexible — surplus at Site A cannot be used for overflow to Site B without cross-training and routing reconfiguration. Staffing risk is site-specific.

Multi-site planning: considerations unique to distributed operations

Time zone differences

If sites operate in different time zones, the staffing requirement calculation must be done in a shared reference time zone (UTC or the primary market time zone) and then converted to local site time. A contact arriving at 8am Eastern US is 1pm in the UK — if both sites handle the same queue, the scheduling must reflect when agents are available in local time against when demand arrives in universal time.

Site-specific bank holidays

Different sites will have different public holiday calendars — UK sites have different bank holidays to sites in Scotland (which has additional holidays), and both differ from US or European sites. The WFM model must apply site-specific holiday calendars rather than a single universal calendar.

Service level measurement at total queue vs. per-site

In a pooled routing model, total queue SL is the correct metric. Per-site SL creates false variance: a site handling overflow appears to have good SL while the site losing agents to overflow appears poor — but both are functioning correctly. Measure total queue SL as primary; use per-site data only for diagnostic analysis.

Cross-site surge management

In a volume spike, the ability to activate cross-site support quickly is a key multi-site advantage. The WFM team must have pre-agreed protocols: which skills can cross-site agents handle, what routing threshold triggers cross-site activation, and who authorises it. Having the protocol designed in advance means activation takes minutes rather than hours.

Multi-site planning questions

Should multi-site contact centres measure service level per site or across all sites?

In a pooled routing model, measure service level at the overall queue level as the primary metric. Per-site SL in a pooled operation is misleading: contacts are routed to whichever site has capacity, so a site absorbing overflow appears to have high SL while the source site appears to have poor SL — but both are working correctly within the routing design. Report total queue SL to senior management; use per-site data only for internal diagnostic purposes (investigating a site-specific incident or staffing shortfall). If sites are genuinely independent (handling entirely separate contact types with no cross-routing), per-site SL is valid as a primary metric.