OR in Manufacturing and Production Planning

Operations research in manufacturing is where the discipline was born. Frederick Taylor's time studies, the Gilbreths' motion studies, and the wartime operations research groups that optimized convoy routes and bomber maintenance schedules — these created the mathematical toolkit that WFM later adopted wholesale. Manufacturing OR and WFM OR solve the same fundamental problem: match capacity to demand over time, subject to constraints on resources, sequencing, and quality.

This page maps the manufacturing-WFM parallel explicitly. The Toyota Production System is not a metaphor for WFM — it is a direct template, with kanban mapping to real-time management, heijunka to demand leveling, and takt time to interval staffing requirements.

Overview

Toyota Production System parallels to real-time WFM

Manufacturing and contact center operations share structural DNA:

• Both convert variable demand into a production schedule constrained by resource availability
• Both face the aggregate planning problem: how much capacity (workers, machines) to provision over a planning horizon
• Both manage quality alongside throughput — Toyota's jidoka (stop-and-fix) parallels real-time quality monitoring and coaching
• Both deal with variability as the central enemy of efficiency

The differences are real — physical goods vs. services, inventory buffers vs. no buffers, deterministic processing vs. stochastic handle times — but the OR methods transfer with minimal modification.

Mathematical Foundation

Aggregate Production Planning

The classic aggregate planning model minimizes total cost over T periods:

$\min \sum _{t=1}^T[c_p{P}_t+c_i{I}_t+c_h{H}_t+c_f{F}_t+c_o{O}_t+c_s{S}_t]$

subject to:

$I_t=I_{t-1}+P_t-D_t\mathrm{\forall }t$ (inventory balance)

$P_t\le (W_{t-1}+H_t-F_t)\cdot k+O_t$ (capacity)

$I_t,P_t,H_t,F_t,O_t,S_t\ge 0$

where $P_t$ is production, $I_t$ is inventory, $H_t$ is hiring, $F_t$ is firing, $O_t$ is overtime, $S_t$ is subcontracting, $D_t$ is demand, and $W_t$ is workforce level.

This is a linear program — identical in structure to WFM annual capacity planning.

Takt Time

Takt time is the production pace required to meet demand:

$\text{Takt Time}=\frac{\text{Available Production Time}}{\text{Customer Demand}}$

In manufacturing, if 480 minutes of production time are available per shift and 240 units must be produced, takt time is 2 minutes per unit. Every workstation must complete its task within 2 minutes.

This is mathematically identical to the WFM staffing requirement per interval: if 120 calls arrive in a 30-minute interval and each takes 6 minutes, the staffing requirement is $120\times 6/30=24$ agents — the demand rate dictates the resource pace.

Line Balancing

Assembly line balancing assigns tasks to workstations to minimize idle time while respecting precedence constraints. The objective:

$\min \sum _{j=1}^m(C-\sum _{i\in S_j}{t}_i)$

where $C$ is cycle time, $S_j$ is the set of tasks assigned to station $j$, and $t_i$ is the task time. This minimizes total idle time across all stations.

Constraints:

• $\sum _{i\in S_j}{t}_i\le C\mathrm{\forall }j$ (no station exceeds cycle time)
• Precedence: task i must be assigned to an earlier station than task k if i precedes k

This is NP-hard — and structurally identical to multi-skill agent assignment where "tasks" are skill requirements and "stations" are agent schedules.

Economic Order Quantity (EOQ)

The EOQ balances ordering costs against holding costs:

$Q^{*}=\sqrt{\frac{2DK}{h}}$

where $D$ is annual demand, $K$ is the fixed ordering cost, and $h$ is the holding cost per unit per year. The total cost function is convex — the unique minimum is the EOQ.

Parallel Structure

Manufacturing Concept	WFM Equivalent	Key Difference
Aggregate Production Planning	Annual/Quarterly Capacity Plan	Manufacturing has inventory buffers; WFM does not (services cannot be stored)
S&OP (Sales & Operations Planning)	WFM Annual Planning Cycle	Same cross-functional demand-supply alignment process
Master Production Schedule	Monthly staffing plan by skill group	Manufacturing schedules products; WFM schedules people
Takt Time	Interval staffing requirement	Same concept: demand pace dictates resource rate
Cycle Time	Average Handle Time	Both measure the time to complete one unit of work
Line Balancing	Multi-skill agent assignment	Same NP-hard assignment problem; same heuristics apply
Kanban (pull-based production)	Real-time WFM (demand-driven routing)	Both pull work to capacity rather than pushing capacity to work
Heijunka (demand leveling)	Schedule smoothing / demand shaping	Both reduce variability by redistributing demand across time
Jidoka (stop-and-fix)	Real-time quality monitoring	Both halt production for quality failures; both prioritize quality over throughput
Andon (visual management)	Real-time WFM dashboards	Same principle: make status visible immediately
Kaizen (continuous improvement)	WFM continuous optimization cycles	Same PDCA structure
EOQ (Economic Order Quantity)	Optimal hiring batch size	Same cost trade-off: fixed cost per batch vs. carrying cost of excess capacity
Machine uptime / OEE	Agent productive time / occupancy	Same utilization metrics with different names
Changeover time (SMED)	Agent wrap-up / after-call work	Both are non-productive time between productive cycles
Work-in-process inventory	Calls in queue	Both represent demand awaiting processing

WFM Applications

S&OP Parallels WFM Annual Planning

Manufacturing S&OP follows a monthly cycle:

1. Demand review: Sales forecast by product family
2. Supply review: Capacity assessment by production line
3. Pre-S&OP: Gap analysis between demand and supply
4. Executive S&OP: Decision on hiring, overtime, outsourcing, inventory targets

WFM annual planning follows the same structure:

1. Demand review: Volume forecast by channel and skill
2. Supply review: Headcount projection (current staff minus attrition plus pipeline)
3. Gap analysis: Identify periods where projected supply fails to meet projected demand
4. Executive review: Decision on hiring classes, overtime budgets, outsource partnerships

The mathematical models are identical. The only structural difference: manufacturing can build inventory during low-demand periods to serve high-demand periods. WFM cannot — services are perishable. This makes WFM planning harder (no buffer) and makes the aggregate planning LP more constrained.

Toyota Production System as WFM Template

Kanban → Real-Time WFM: Kanban is a pull system — production starts only when downstream demand signals. In WFM, real-time routing is pull-based: agents are assigned interactions only when they become available, and work is "pulled" from the queue rather than "pushed" based on a pre-set schedule.

Heijunka → Demand Leveling: Toyota levels production by mixing product types across the day rather than producing in large batches. WFM equivalents:

• Scheduling callbacks to spread demand away from peaks
• Offering self-service options that divert simple contacts from peak hours
• Staggering outbound campaigns to avoid demand spikes

Jidoka → Quality Monitoring: Toyota empowers any worker to stop the production line when a defect is detected. In WFM:

• Real-time speech analytics flag quality issues during live interactions
• Supervisors can intervene (barge/whisper) when quality scores drop below threshold
• Automated systems can route interactions away from struggling agents

EOQ for Hiring Batch Size

Hiring has a fixed cost (recruiting, onboarding, training class setup) and a variable cost (each new hire's salary). The EOQ analog:

$H^{*}=\sqrt{\frac{2\cdot A_{\text{annual}}\cdot K_{\text{class}}}{c_{\text{carry}}}}$

where $A_{\text{annual}}$ is annual attrition (units of replacement demand), $K_{\text{class}}$ is the fixed cost per training class, and $c_{\text{carry}}$ is the cost of carrying one excess FTE (salary + overhead during ramp-down).

If annual attrition is 120 agents, each training class costs $50,000 to run, and each excess FTE costs $45,000/year in carrying cost:

$H^{*}=\sqrt{\frac{2\times 120\times 50,000}{45,000}}\approx 16.3\to 16\text{ agents per class}$

This yields approximately 7-8 hiring classes per year — a quarterly-plus cadence. Running smaller classes more frequently would increase fixed costs. Running larger classes less frequently would increase carrying costs from overstaffing between classes.

Worked Example

Problem: Apply aggregate production planning to a WFM quarterly capacity plan.

Given:

• 4 months: January through April
• Forecasted monthly volumes: 180K, 160K, 200K, 220K
• Current headcount: 400 agents
• Productivity: 500 contacts/agent/month (fully productive)
• Hiring cost: $5,000 per agent (recruiting + training)
• Separation cost: $2,000 per agent (severance, knowledge loss)
• Overtime cost premium: $800/agent/month (limited to 10% of workforce)
• Understaffing penalty: $3,000/agent/month (estimated from service degradation + customer churn)

LP Formulation:

$\min \sum _{t=1}^4[5000H_t+2000F_t+800O_t+3000U_t]$

subject to:

$W_t=W_{t-1}+H_t-F_t$

$500W_t+500O_t+500U_t\ge D_t$

$O_t\le 0.10\cdot W_t$

Optimal solution:

Month	Demand (K)	Workforce	Hire	Separate	Overtime	Understaff
Jan	180	400	0	40	0	0
Feb	160	360	0	0	0	0
Mar	200	360	40	0	36	0
Apr	220	400	40	0	40	0

Total cost: $5,000 × 80 + $2,000 × 40 + $800 × 76 = $400,000 + $80,000 + $60,800 = $540,800

The optimizer separates agents in January (when demand drops) and rehires in March-April (when demand surges), using overtime to bridge the March gap while the new hires ramp up. This is precisely the "chase strategy" in aggregate planning — matching workforce to demand rather than maintaining a level workforce.

Maturity Model Position

• Level 2 (Developing): Headcount planning uses spreadsheets with manual capacity calculations; no formal production planning methods
• Level 3 (Advanced): Aggregate capacity planning with LP; takt time concept applied to interval requirements; basic lean principles adopted
• Level 4 (Leading): Full S&OP cycle with demand-supply integration; TPS principles systematically applied; EOQ-style analysis for hiring cadence
• Level 5 (Innovating): Advanced production planning (stochastic aggregate planning, robust capacity optimization); digital twin of the operation modeled as a production system; continuous flow optimization with real-time kanban-style routing

See Also

• Capacity Planning in WFM
• Linear Programming in WFM
• Demand Forecasting in WFM
• Lean Principles in WFM
• Operations Research in Workforce Management

References

• Hopp, W.J. & Spearman, M.L. (2011). Factory Physics. 3rd ed. Waveland Press.
• Ohno, T. (1988). Toyota Production System: Beyond Large-Scale Production. Productivity Press.
• Nahmias, S. & Olsen, T.L. (2015). Production and Operations Analysis. 7th ed. Waveland Press.
• Womack, J.P. & Jones, D.T. (2003). Lean Thinking. 2nd ed. Free Press.
• Silver, E.A., Pyke, D.F. & Thomas, D.J. (2017). Inventory and Production Management in Supply Chains. 4th ed. CRC Press.