BUS 721 Week 6 Data and Decision Making Example

Which Order Ships First, Where to Add Capacity and When to Warn a Builder: Data-Driven Operational Decisions at Tri-State

[Student Name]

University of Phoenix

BUS/721: Issues in Optimizing Operations

Week 6 Assignment

[Instructor Name]

[Date]

Tri-State Cabinet Company and all data are composites written for a model paper.

Tri-State Cabinet Company, a composite kitchen cabinet manufacturer, is building order visibility through business intelligence and planning integration through a new ERP. Data alone will not improve delivery unless decisions change. Three recurring decisions matter most: which orders to prioritize each day, where and when to add capacity and when to warn builders about delays. Data change outcomes only through decisions, and decisions are made by people with habits, pressures and blind spots that data cannot simply override. This paper examines data and decision making at Tri-State.

Management Practices and Performance

Research using structured interviews in thousands of firms found that management practices such as monitoring operations, setting targets and linking rewards to performance varied widely and were associated with higher productivity and profitability (Bloom & Van Reenen, 2007). Data-driven decisions depend on these practices: data must be monitored, compared with targets and acted on.

Judgment and Biases

Tversky and Kahneman (1974) described heuristics that lead to systematic errors, including anchoring on initial values, availability of vivid examples and overconfidence. Operational decisions under time pressure are especially prone to these biases, such as anchoring capacity plans on last year's volume or overreacting to one memorable late order.

Operations Science

Hopp and Spearman (2011) explained that lead time depends not only on processing time but on queueing, which grows sharply as utilization rises and as variability increases. Running resources near full utilization in a variable environment produces long, unpredictable lead times, a principle often misunderstood by managers who equate high utilization with efficiency.

Decision One: Daily Order Priorities

Plant schedulers currently prioritize orders by style to minimize changeovers, a habit from the cost strategy that made sense when most orders came from home centers ordering single styles in bulk. Data from the new dashboard show that this delays mixed builder orders. Priority rules based on due dates and order completeness, finishing the components needed to complete the oldest builder orders first, would reduce late orders, at some cost in changeovers.

Testing the Priority Rule

A simulation using three months of order data compared the style-based rule with a due-date rule. The due-date rule reduced late orders by 38 percent while increasing changeovers by 15 percent. Supervisors reviewed the results and suggested a hybrid that groups orders by style within due-date windows, cutting the changeover increase to 8 percent with most of the delivery benefit.

Decision Two: Capacity

Finishing lines run at about 95 percent utilization during peak months. Plant managers view this as efficient. Queueing principles suggest it is a cause of long lead times: at such high utilization, small disruptions create long queues. Data show finishing wait times of three to five days in peak months versus one day in slower months, even though processing time per part is the same in both.

Adding Capacity Where It Matters

Adding a partial second shift on finishing lines during peak months, lowering utilization to about 80 percent, is predicted to cut finishing wait times by more than half. The cost, about $420,000 a year, is less than the expected value of retained builder business and fewer expedited shipments.

Reducing Variability

Capacity is not the only lever. Reducing variability through the rework improvements from Week 5 and more stable order release would lower queues even at the same utilization. Combining modest capacity with lower variability is more effective than either alone, and it is cheaper than adding a full shift.

Decision Three: Warning Builders

Customer service now warns builders only when an order is already late. With dashboard data, Tri-State can predict at-risk orders five to seven days ahead. Earlier warnings let builders adjust installation schedules, preserving trust even when delays occur. Builders interviewed in Week 1 said a warning a week ahead is far less damaging than a surprise on delivery day.

Why Managers Favor High Utilization

Plant managers' preference for high utilization is rational given their incentives: they are measured on cost per unit, which falls as equipment runs more. The cost of long queues appears elsewhere, in late orders handled by sales and customer service. This mismatch between where costs are measured and where they occur is common in operations and explains why the queueing insight is resisted. Changing plant scorecards, as recommended in earlier weeks, removes the incentive to keep utilization at its maximum.

Decisions About Suppliers

Data also inform supplier decisions. Hardware and glass suppliers caused 15 percent of late orders. Dashboard data on supplier delivery performance will support quarterly reviews, consignment stock for the most common hinges and handles and, where performance does not improve, a second source. Decisions that once rested on relationships and price alone will include delivery reliability.

Data Literacy

Supervisors and schedulers vary in comfort with data. Short training sessions on reading dashboards, understanding variability and interpreting prediction scores will help them use the tools. Analysts will sit with plant teams during the first months, translating data into practical choices rather than sending reports from headquarters.

Combining Data and Judgment

Supervisors know things data miss: a machine making unusual noises, an operator out sick, a supplier truck delayed by weather. Daily meetings will review dashboard flags and invite supervisors to add context, adjusting priorities when their knowledge points to risks the data do not show.

Guarding Against Bias

Decision routines will ask explicit questions: what does the data show, what would change our decision and what are we assuming. Capacity decisions will be checked against base rates from past peak seasons rather than anchored on optimistic forecasts.

When Data and Judgment Disagree

Sometimes the dashboard says an order is on track while a supervisor expects trouble. The routine is to treat disagreement as information: the supervisor explains the concern, the team checks quickly and the order is flagged if the concern holds. Over time, recording these cases shows whether supervisors' warnings or the model's predictions are more accurate for different types of problems, improving both.

Decision Governance

Rules for routine decisions, such as order priorities, will be documented and owned by the scheduling manager. Nonroutine decisions, such as capacity changes, will go to the operations leadership team with data and options. Decisions and their outcomes will be logged so the team can learn.

Measuring Decision Quality

Measures include on-time complete delivery, finishing wait times, changeovers, warning lead times and the accuracy of at-risk predictions. Reviewing outcomes against decisions helps distinguish good decisions with bad luck from poor decisions.

Research Questions

Open questions include how supervisors integrate dashboard information with tacit knowledge and whether simple priority rules outperform supervisors' discretion in semi-custom manufacturing.

Conclusion

Data will improve Tri-State's delivery only through better decisions. Research on management practices, judgment and operations science shows how. Due-date priorities, capacity that lowers utilization at the bottleneck, reduced variability, earlier warnings and routines that combine data with supervisors' knowledge turn visibility into results.

References

Bloom, N., & Van Reenen, J. (2007). Measuring and explaining management practices across firms and countries. The Quarterly Journal of Economics, 122(4), 1351-1408. https://doi.org/10.1162/qjec.2007.122.4.1351

Hopp, W. J., & Spearman, M. L. (2011). Factory physics (3rd ed.). Waveland Press.

Tversky, A., & Kahneman, D. (1974). Judgment under uncertainty: Heuristics and biases. Science, 185(4157), 1124-1131. https://doi.org/10.1126/science.185.4157.1124