Process experiments that reduce waste and increase throughput.
Maintenance schedules, tooling parameters, quality control methods, shift structures — manufacturing improvements are often opinions. Decision Process turns them into controlled experiments with measurable outcomes.
Experiment Templates
Ready-to-run experiments
Preventive Maintenance Schedule A/B
“Weekly PM reduces unplanned downtime more than bi-weekly PM at acceptable cost.”
Conditions
- Bi-weekly PM (control)
- Weekly PM
- Daily inspection only
Metrics
- Unplanned downtime (hrs/mo)
- Defect rate (%)
- PM labor cost ($)
Cutting Speed Optimization
“A 15% increase in cutting speed maintains surface finish quality while improving throughput.”
Conditions
- Standard speed (control)
- +10% speed
- +15% speed
- +20% speed
Metrics
- Parts per hour
- Surface finish Ra (µm)
- Tool life (parts/edge)
Quality Control Method Comparison
“Statistical process control catches defects earlier than end-of-line inspection.”
Conditions
- End-of-line inspection (control)
- SPC — every 20 units
- SPC — every 10 units
Metrics
- Defect escape rate (%)
- Scrap cost per 1,000 units ($)
- Inspection labor (min/unit)
Shift Structure Comparison
“Three 8-hour shifts produce higher throughput than two 12-hour shifts due to fatigue effects.”
Conditions
- Two 12-hr shifts (control)
- Three 8-hr shifts
Metrics
- OEE %
- Units per labor hour
- Error rate in last 2 hrs of shift
Assembly Jig Comparison
“A new pneumatic jig reduces assembly cycle time without affecting defect rate.”
Conditions
- Manual jig (control)
- Pneumatic jig
Metrics
- Cycle time (sec/unit)
- Defect rate (%)
- Operator fatigue rating (1–10)
Worked Example
Maintenance schedule experiment across 6 production lines
A plant manager tests two preventive maintenance schedules across 6 identical production lines over 3 months. 3 lines stay on bi-weekly PM; 3 switch to weekly PM. Unplanned downtime and defect rate are tracked weekly.
Results: unplanned_downtime (hrs/month)
Bi-weekly PM (control)
mean: 14.2 hrs/month
95% CI: 11.8–16.6
Weekly PM
mean: 9.8 hrs/month
95% CI: 7.9–11.7
P(weekly PM better) = 94%
Weekly PM reduces unplanned downtime by 31% (d = −0.83, large effect). At 94% posterior probability this exceeds the plant's 90% threshold. The added PM labor cost ($420/line/month) is offset 3:1 by reduced downtime cost ($1,280/line/month saved). Recommend switching all 12 lines to weekly PM.
Explore other domains
Run your first manufacturing experiment
Private beta — tell us about your use case and we'll get you set up.