WFM guidePlanning

WFM scenario planning & what-if analysis

Q: What is what-if analysis in contact centre workforce management?

What-if analysis (scenario planning) recomputes the staffing requirement, predicted service level, and cost under a changed set of assumptions, so you can see the consequence of a decision before you make it. A baseline plan answers 'how many agents do I need for my forecast?'. A what-if answers questions like: 'what happens to SL and cost if volume rises 20%?'; 'how many fewer agents do I need if I cut AHT by 30 seconds?'; 'what does raising chat concurrency from 2 to 3 do to my staffing and my response-time risk?'; 'if I hold headcount flat, how far does SL fall?'. Each scenario flexes one or more input variables and runs the same capacity model (Erlang C for voice, the concurrency model for chat, the backlog model for deferred work) on the changed inputs. The output is a comparison: each scenario's required headcount, predicted SL, weekly cost, and cost per contact, with the deltas against the baseline. This turns a resourcing conversation from opinion ('we need more people') into a quantified trade-off ('+20% volume needs 12 more FTE at £18k/week, or SL drops to 68% if we hold headcount').

Scenario planning turns "I think we need more people" into "at +15% volume, holding our 80/20 target, we need 12 more FTE at £18k/week — or we hold headcount and accept SL falling to 68%." It makes the cost-service trade-off explicit before the decision, not after.

What a scenario adds to a baseline plan

A baseline plan answers one question: how many agents do I need for my current forecast and assumptions? A scenario changes one or more of those inputs and re-runs the same capacity model, so you can see the consequence of a possible future or a possible decision before committing to it. The discipline is to flex the inputs deliberately and compare the outputs side by side — required headcount, predicted service level, weekly cost, and cost per contact — with the delta against the baseline. The decision then rests on a quantified trade-off, not an assertion.

Five variables worth flexing

Volume (±%)

Question it answers

What if demand rises or falls — a campaign, a launch, a deflection programme, a downturn?

Effect on the model

Drives the requirement non-linearly through the queuing model. A 20% volume rise needs more than 20% more agents at high occupancy, because the queue dynamics worsen as utilisation climbs. The most common and most important scenario variable.

AHT (±seconds or ±%)

Question it answers

What if handle time changes — a new process, a system change, an AHT-reduction initiative, added wrap requirements?

Effect on the model

Scales the offered load directly. Cutting AHT reduces the requirement; a +10% AHT increase raises it by roughly the same proportion at the requirement level (and more at the SL level). Quantifies the staffing value of an AHT initiative.

Shrinkage (±%)

Question it answers

What if shrinkage changes — more training, an absence spike, a wellbeing programme reducing absence?

Effect on the model

Changes the gross-up from seated requirement to scheduled requirement. Higher shrinkage means more agents must be rostered to put the same number on the phones. A small shrinkage change moves the establishment requirement noticeably.

Service level target

Question it answers

What if we relax or tighten the SL target — a contractual change, a cost-saving review, a quality push?

Effect on the model

Moves the requirement sharply near the top of the curve. Going from 80/20 to 90/20 costs disproportionately more agents; relaxing to 70/20 frees up significant headcount. Exposes the true cost of each SL point.

Concurrency (chat)

Question it answers

What if chat agents handle more or fewer simultaneous conversations?

Effect on the model

Raising concurrency reduces the agent requirement but increases response-time risk and AHT inflation per conversation. A scenario makes the efficiency-vs-quality trade-off of concurrency explicit rather than assumed.

Comparing scenarios: an example

Scenario	Peak agents	Predicted SL	Weekly cost	Cost / contact
Baseline	80	80%	£60k	£3.00
+20% volume, hold SL	96	80%	£72k	£3.00
+20% volume, hold headcount	80	68%	£60k	£2.50
Cut AHT 30s, hold volume	72	82%	£54k	£2.70
Relax SL to 70/20	71	72%	£53k	£2.65

Illustrative figures. Note the highlighted row: holding headcount at +20% volume keeps cost flat and even lowers cost-per-contact — but only because SL collapses to 68%. Cost-per-contact on its own flatters a deteriorating service. Always read cost and service together.

Four mistakes that make scenario analysis misleading

Reading cost without service

A scenario that cuts headcount always looks cheaper on cost-per-contact — because the metric divides cost by volume, not by quality. A falling cost-per-contact alongside a collapsing SL is not an efficiency gain; it is service erosion. Always show SL next to cost.

Flexing one variable when two move together

Real changes rarely isolate one input. A volume spike from a campaign often comes with a different contact-reason mix and therefore a different AHT. A scenario that flexes volume but holds AHT constant may understate the true impact. Model correlated changes together.

Linear thinking on a non-linear model

Staffing does not scale linearly with volume near high occupancy — a 10% volume rise can need far more than 10% more agents, and a 10% staffing cut can drop SL far more than 10%. Do not eyeball scenario impacts; run them through the queuing model.

Treating a scenario as a forecast

A scenario is a conditional 'what if', not a prediction that it will happen. Use scenarios to bound the decision (best/worst/likely), not to replace the forecast. Staffing to a worst-case scenario as if it were certain is as wasteful as ignoring it entirely.

Scenario planning questions

What is what-if analysis in contact centre workforce management?

What-if analysis recomputes the staffing requirement, predicted service level, and cost under changed assumptions, so you see the consequence of a decision before making it. Where a baseline plan answers 'how many agents for my forecast?', a what-if answers 'what happens to SL and cost if volume rises 20%?', 'how many fewer agents if I cut AHT by 30 seconds?', or 'how far does SL fall if I hold headcount flat?'. Each scenario flexes one or more inputs and runs the same capacity model (Erlang C, concurrency, or backlog) on the changed inputs, then compares the scenarios on required headcount, predicted SL, weekly cost, and cost per contact. It converts a resourcing argument from opinion into a quantified cost-versus-service trade-off.