Six Sigma DMAIC in DC

DMAIC (Define, Measure, Analyze, Improve, Control) is the right tool when: (1) you can define a defect, (2) count it reliably, and (3) the process runs >20 times/day. The statistical rigor pays off because you have enough data to find real root causes, not just plausible ones.

DMAIC Applied to Mis-Pick Reduction

The canonical DC application.

Define

Defect = a mis-pick (wrong item, wrong qty, wrong SKU). Business impact: $15–40 per mis-pick in re-pick, re-ship, and CS cost. Project goal: Reduce Zone 3 mis-pick rate from 1.8% to ≤0.5% in 90 days.

Measure

Pull 30 days of WMS pick data for Zone 3. Capture mis-picks by: picker, SKU, time of day, shift, pick method (RF vs. scan-verify vs. manual).

Baseline DPMO calculation:

• 1.8% error rate → DPMO = 18,000 → ~3.6 sigma
• Target ≤0.5% → DPMO = 5,000 → ~4.1 sigma
• True Six Sigma = 3.4 DPMO → essentially unattainable in a manual DC; 4.1 sigma is a realistic best-in-class target

Analyze

Run the Pareto. In this example: 80% of mis-picks trace to 6 SKUs slotted within 12 inches of each other with similar packaging. Secondary: third-shift error rate is 2.8× first-shift (lighting, training, or supervision factor).

Fishbone surfaces: slotting logic used in Q1 re-slot had no check for item similarity before finalizing adjacent assignments.

Improve

Three countermeasures, in order of impact:

1. Separate look-alike SKUs by ≥36 inches or use contrasting slot label colors
2. Add weight-check step at pack station — carton outside ±5% of expected weight → rejected for manual verification
3. Add look-alike item proximity check to the slotting process (prevents the Q1 situation from recurring in any zone)

Control

• Implement SPC on daily mis-pick rate; set control limits at ±3σ from new mean
• Any point outside limits → automatic root cause review
• Update slotting process SOP to include the look-alike check (permanent standard, not just a Zone 3 fix)
• Assign pack station weight gate as permanent standard work

Process Capability — Cpk Math

Cpk tells you whether your process is capable, not just whether the average looks acceptable.

Formula: Cpk = (Mean − Lower Spec Limit) ÷ (3 × Standard Deviation)

Worked example:

• Specification: ≥99.5% pick accuracy (lower spec limit = 99.5%)
• Process mean: 99.7%, standard deviation: 0.1%
• Cpk = (99.7 − 99.5) ÷ (3 × 0.1) = 0.2 ÷ 0.3 = 0.67

Cpk	Interpretation
<1.0	Process is not capable — will regularly produce defects
1.0–1.33	Marginally capable
≥1.33	Industry minimum for most processes
≥1.67	Safety-critical processes

Implication: Cpk 0.67 means regular misses even at 99.7% average. To reach Cpk 1.33 at 99.5% spec, standard deviation must drop to ~0.05% — requires tight process consistency, not just good averages.

Pick accuracy projects that focus on average performance miss the point. Days at 98.9% are costing real money even when the monthly average looks fine.

Statistical Process Control (SPC)

Control charts track daily performance against statistically derived limits (mean ± 3σ). A point outside the control limits is a signal — not noise — and triggers root cause review.

• UCL / LCL calculated from process data, not from target aspirations
• A process in statistical control is predictable — you know the range of outcomes
• A capable process (high Cpk) in statistical control reliably meets the spec

When DMAIC vs. Lean Kaizen

Situation	Use
Can define and count a defect; root cause is contested; 6–10 weeks acceptable	DMAIC
Problem is visible waste; root cause is obvious from waste walk; 3-day event can fix and validate	Lean Kaizen
Both defect and waste are present	Use VSM to find where; DMAIC where root cause is contested; Kaizen where solution is obvious

Running an 8-week DMAIC project on a problem you can solve with better slotting and a 5S event wastes everyone’s time and kills CI engineer credibility with operations.

Blended Approach

1. Lean VSM to find where the problems are
2. DMAIC for any problem where root cause is contested or statistical evidence is required to convince a skeptical operations team
3. Kaizen events for rapid implementation once root cause is clear