Fictional worked example

AI works, and the proposed reduction is supported

A 180-FTE customer-service organisation has operated AI self-service and agent-assist workflows at scale for nine months. Management proposes eliminating 32 customer-service positions at the start of the next quarter. The Stress Test supports the proposed amount and timing.

A productivity result can support permanent restructuring when the full decision chain survives the test.

Decision at a glance

The proposed reduction remains below the evidenced capacity available for choice.

Proposal Remove 32 positions next quarter

Reduce the customer-service organisation from 180 to 148 FTE.

Capacity available for choice Approximately 33 to 35 FTE

The proposed 32-position reduction stays below the low end of the tested available-capacity range.

Supported structural outcome Remove 32 positions next quarter

Use internal placement and temporary assignments where they improve implementation without preserving structurally unnecessary roles.

Decision status 01: Support the proposed reduction now

The amount and timing survive all four questions.

Employee capacity Customer service 180 FTE starting team AI operating at scale for 9 months

Evidence behind the case

Human work has fallen, service has held, and the released capacity forms a structural block.

Observed evidence

  • Total customer contacts are broadly unchanged.
  • Net human workload is 19% lower after human review, escalations, after-call work and new control activity are included.
  • The reduction is concentrated in three high-volume service queues.
  • The revised operating model requires approximately 34 fewer FTE at planned service levels.
  • Tested demand and service scenarios require approximately 145 to 147 FTE to remain. The proposal leaves 148 FTE.
  • Peak-hour demand in the affected queues is 16% lower.
  • Abandonment, response time, complaint rate and quality scores have remained within target for nine months.
  • The retained organisation preserves the scarce product and escalation skills required for exception work.

Economics and remaining assumptions

32-position annual carrying cost avoided ≈ £2.7m
Severance, transition and temporary-risk cost ≈ £0.6m
Equivalent capacity rebuild time ≈ 8 weeks
Suitable approved vacancies elsewhere 3 roles

Assumptions still carrying the recommendation

  • Contact demand remains within the agreed range.
  • Current AI performance and containment are sustained.
  • No material regulatory change increases human-review requirements.
  • Service and quality remain within agreed tolerance.
  • Equivalent capacity remains restorable in approximately eight weeks.

The capacity bridge

The proposal stays inside the evidenced capacity range.

The six-stage capacity bridge establishes capacity available for choice before Q4 tests whether permanent removal is economically stronger than the alternatives.

01

AI productivity

AI self-service and agent assist have changed how customer demand is handled.

02

Net human work change

Human workload is 19% lower after remaining and new work are included.

03

Released capacity

The changed workload releases capacity equivalent to roughly 34 FTE.

04

Structurally addressable capacity

The released work is concentrated in three high-volume queues and forms an identifiable block of approximately 34 FTE.

05

Capacity that must remain

Tested demand and service scenarios require approximately 145 to 147 FTE to remain.

06

Capacity available for choice

Approximately 33 to 35 FTE is available across the tested range.

33 to 35 FTE available for choice 32 positions proposed

The four questions

The proposed 32-position reduction survives the full test.

How much human work is actually no longer required?

Supported

Net human workload is 19% lower after human review, exceptions, escalations, after-call work and new control activity are included. The evidence has persisted for nine months while total customer contacts remain broadly unchanged.

The operation genuinely requires materially fewer human workload hours to serve approximately the same demand.

Is released capacity structurally addressable?

Supported

The released time is concentrated in three high-volume service queues rather than appearing as small fragments across the 180-FTE organisation. The revised workload maps to an identifiable block of approximately 34 FTE.

Approximately 34 FTE is structurally addressable. Q2 establishes the block, not whether all of it should be removed.

What capacity must remain to protect performance, coverage, skills and resilience?

33 to 35 FTE available

The base operating model requires approximately 146 FTE to remain. Sensitivity testing across the observed demand and service range produces a retained requirement of approximately 145 to 147 FTE. The proposal leaves 148 FTE, above the upper end of that tested requirement.

Approximately 33 to 35 FTE is available for choice. The proposed 32-position reduction preserves at least 1 FTE of modelled headroom across the tested range.

Is the proposed reduction, at this amount and timing, economically better than the credible alternatives?

32 now is stronger

The organisation compares removal, retention, redeployment, temporary work, staging and a smaller reduction on feasibility, value, timing and reversibility.

Option A

Eliminate 32 customer-service positions now

≈ £2.7m annual carrying cost avoided · ≈ £0.6m transition cost

The proposal remains below the available-capacity range, leaves 148 FTE in place and can be reversed by rebuilding equivalent service capacity in approximately eight weeks.

Strongest current structural option.
Option B

Retain the 32 positions

≈ £2.7m annual carrying cost

Retention preserves maximum optionality, but no current service, peak, skill or resilience requirement justifies carrying the full block.

Optionality does not justify the recurring cost.
Option C

Redeploy suitable employees into approved vacancies

3 suitable roles · ≈ £0.12m recruitment and onboarding cost avoided

Internal placement preserves knowledge and fills roles the organisation already intends to fund. It changes which people leave, not whether the 32 customer-service positions are still required.

Use where valuable, without preserving the old roles.
Option D

Use some capacity for temporary backlog work

≈ 6 FTE for 4 months · ≈ £0.16m contractor spend avoided

Carrying six employees for the same period costs approximately £0.17m in payroll. The temporary use is roughly break-even before wider transition effects and ends after four months.

Useful implementation option, not a permanent-capacity case.
Option E

Stage the whole 32-position reduction

More flexibility, more carrying cost

Nine months of stable evidence, protected scarce skills, modelled headroom and an approximately eight-week restoration period reduce the value of carrying the whole block for longer.

No material requirement offsets the extended carrying cost.
Option F

Remove fewer than 32 positions

More headroom, higher recurring cost

A smaller reduction creates additional operating headroom, but the 32-position proposal is already below the low end of the tested available-capacity range.

No evidenced requirement justifies preserving the extra positions.
Eliminating 32 customer-service positions now, while using economically attractive redeployment and temporary assignments during implementation, is stronger than retaining, broadly staging or permanently preserving capacity that the operating model no longer requires.

Stress Test conclusion

Support the proposed reduction now.

The amount and timing remain supported after workload, structural addressability, protected requirements, economics, alternative uses and sensitivity have been tested.

Structural decision ≠ people decision. Some affected employees may continue elsewhere in the organisation. That does not recreate a requirement for the 32 customer-service positions.

What could change the answer

The reduction is supported now, but the operating assumptions still need monitoring.

Open conditions

  • Monitor demand, peak performance, service and quality monthly for the first six months.
  • Maintain agreed scarce-skill and escalation coverage.
  • Track whether the assumptions underlying the 146-FTE retained requirement remain valid.
  • Complete the identified internal-placement review before individual exits are finalised.
  • Use temporary assignments only where their value exceeds the cost or disruption of delaying the relevant exit.

Retest the decision if

  • Demand moves materially outside the agreed range.
  • Mandatory human-review requirements rise.
  • Service or quality breaches the agreed tolerance.
  • Scarce-skill coverage becomes inadequate.
  • Restoration economics deteriorate materially before implementation.
  • A materially higher-value permanent use for the capacity emerges before the structural change is completed.

What this example shows

AI productivity evidence can support permanent restructuring without ignoring the alternatives.

AI works human work genuinely falls released capacity forms an addressable block required capacity is protected capacity becomes available for choice credible alternatives are compared some alternatives improve implementation the 32-position reduction remains supported

Some employees are better redeployed and some released capacity has useful temporary value. Neither fact requires the organisation to preserve permanent positions that the operating model no longer needs.

AI Restructuring Business Case Stress Test

Apply the same test to your own case.

Use the parent resource to test the bridge from AI productivity to a proposed permanent capacity decision.

Open the Stress Test

More worked examples

Browse all worked examples