Call centre capacity planning — the complete WFM guide

Collect and clean historical contact data

Capacity planning starts with data. You need at minimum 8–12 weeks of interval-level data: contacts received per 15- or 30-minute interval, AHT, and channel. The longer your history and the more granular the interval, the better your forecast will be. Clean out outliers (major incidents, bank holidays, system outages) before using data to build a forecast.

Forecast demand by interval and channel

A forecast breaks your expected volume into intervals, typically 15 or 30 minutes. Good forecasts apply day-of-week and intraday patterns multiplicatively: Monday midday is not simply the average of all intervals. Add planned volume drivers (campaigns, product changes, seasonal events) on top of the baseline. Use MAPE or weighted MAPE to measure forecast accuracy and iterate.

Calculate agent requirements: the right model for each channel

This is where most operations make the critical error: applying Erlang C to every channel. Voice queues need Erlang C. Live chat needs a concurrency model. Email, back-office, and case processing need a backlog flow model. Using the wrong model will overstaff or understaff your operation. The channel mix in your operation determines which calculators you need.

Apply shrinkage to convert seated to scheduled headcount

The Erlang C result tells you how many agents need to be actively handling calls: your seated requirement. Your scheduled headcount must be higher to account for all the time agents are not available: breaks, meals, meetings, training, coaching, administrative tasks, sickness, and lateness. Shrinkage is the total of these as a percentage. Scheduled = seated ÷ (1 − shrinkage).

Design shift patterns and schedule agents

With a scheduled headcount requirement by interval, you can design shifts. Shifts must cover the intervals that need coverage, respect working-time rules, and be acceptable to agents. Shift design is a constrained optimisation: you want to match supply to demand as closely as possible without creating unsustainable patterns. Most operations use a combination of fixed core shifts and flexible cover shifts.

Model cost and build the budget

Once you have a schedule, you can calculate labour cost: agents × shift length × hourly rate × employer on-costs. Compare this to your headcount budget. If cost exceeds budget, test whether AHT improvements or FCR gains can reduce the requirement before cutting headcount. Cost per contact is the most useful efficiency metric: it combines staffing cost with throughput in a single number.

Monitor intraday and manage deviation

A capacity plan is a model of the future. Reality will deviate. Intraday management is the discipline of detecting deviation early and responding: volume arriving higher than forecast, AHT longer than modelled, agent absence not captured in the plan. The faster you detect and respond, the smaller the impact on service level. Real-time adherence (RTA) monitoring feeds into this loop.