WFM guideBooked demand

Appointment-based demand WFM

Q: How does workforce planning differ for appointment-based operations?

In a random-arrival operation (a typical inbound contact queue), customers arrive when they choose, demand is a given, and the planning problem is 'how many agents do we need to meet this arrival pattern at our service target?' — solved with a queuing model like Erlang C. In an appointment-based operation (clinics, field-service visits, scheduled callbacks, booked consultations), the arrival curve is not a given — it is shaped by the booking system. Customers are offered slots, and the operation controls how many slots exist at each time. This inverts the planning problem: instead of staffing to meet demand, you shape demand to fit capacity, by deciding how many appointment slots to open per interval based on how many staff are available. The model is closer to capacity-and-slot planning than to queuing theory. It is in some ways an easier problem — you are not at the mercy of random spikes, and utilisation can be very high because demand is matched to capacity by design. But it introduces concerns random-arrival WFM doesn't face: no-shows (booked demand that doesn't materialise, wasting the reserved capacity), overruns (appointments running longer than booked, cascading delays), and the hybrid case where booked appointments coexist with random walk-in or urgent demand that must be buffered for.

Most WFM assumes demand arrives randomly and you staff to meet it. Appointment-based operations invert that: the arrival curve is something you control through the booking system. The problem flips from "how many agents for this demand?" to "how much demand should we accept given this capacity?" — a different, and in ways easier, problem.

The inversion: shape demand to capacity

In a random-arrival queue, demand is a force of nature — you forecast it and staff to meet it, and the only lever you hold is staffing. In an appointment-based operation, demand is partly yours to control: you decide how many slots to open at each time, so you can match the booked demand to the staff you have available. The planning problem inverts — rather than asking "how many agents do I need for this demand?", you ask "how many slots should I open given the staff I have?". Because supply and demand are matched by design, utilisation can run higher than a random queue (no forced idle gaps between unpredictable arrivals) — provided you manage the appointment-specific risks of no-shows and overruns.

Random-arrival vs. appointment-based: six differences

Dimension	Random arrival	Appointment-based
Who controls arrivals	The customer — they call/contact when they choose.	The operation — through the slots it opens in the booking system.
The planning question	How many agents to meet this demand at the SL target?	How many slots to open given the staff we have available?
Core model	Queuing theory (Erlang C) — random arrivals, service level.	Capacity-and-slot planning — match slots to available staff-hours.
Achievable utilisation	Capped (~80-85%) — idle gaps between random arrivals are unavoidable.	Higher — demand is matched to capacity by design, so less forced idle time (subject to no-shows).
Main risk	Unforecast spike → queue → SL breach / abandonment.	No-shows (reserved capacity wasted) and overruns (appointments cascading late).
Lever to balance supply & demand	Adjust staffing (the only lever — demand is fixed).	Adjust slots AND staffing — demand is a controllable lever too.

Appointment-specific planning concerns

No-shows

Booked demand that doesn't materialise wastes the reserved capacity — the slot was held, the staff were there, no one came. Track the no-show rate by slot type and time, and consider controlled over-booking (offering slightly more slots than capacity, matched to the no-show rate) to recover the lost utilisation — carefully, because if everyone shows, you overrun.

Overruns

Appointments that run longer than their booked duration cascade: each overrun pushes the next appointment late, compounding through the session. Build realistic slot durations from actual handling-time data (not optimistic targets), and consider buffer slots to absorb overruns before they cascade.

Booked + walk-in hybrid

Many operations mix booked appointments with random walk-in or urgent demand (a clinic with appointments plus emergencies; a service desk with scheduled and ad-hoc tickets). This needs a hybrid model: capacity-and-slot planning for the booked portion PLUS a queuing buffer for the random portion. Don't plan the whole thing as either pure-booked or pure-random.

Slot granularity vs. flexibility

Fine-grained slots (e.g. 10-minute) match demand to capacity precisely but leave no slack for overruns; coarse slots (e.g. 30-minute) absorb variation but waste capacity on short appointments. Choose slot granularity to balance utilisation against overrun resilience, the same trade-off as forecast interval length in a queue.

Appointment-based WFM questions

How does workforce planning differ for appointment-based operations?

In a random-arrival operation, customers arrive when they choose, demand is a given, and you staff to meet it with a queuing model like Erlang C. In an appointment-based operation (clinics, field-service, scheduled callbacks), the arrival curve is shaped by the booking system — you control how many slots exist at each time. This inverts the problem: instead of staffing to meet demand, you shape demand to fit capacity by deciding how many slots to open given available staff. The model is capacity-and-slot planning, not queuing theory. It's in some ways easier — no random spikes, and utilisation can run higher because supply and demand are matched by design — but it adds concerns random-arrival WFM doesn't face: no-shows (reserved capacity wasted), overruns (appointments cascading late), and the hybrid case where booked appointments coexist with random walk-in or urgent demand that still needs a queuing buffer.