Summary of "SESSION 02"

Session 02 — Yellow Belt / Lean Six Sigma training (Summary)

Key frameworks, processes and playbooks

Voice of the Customer (VOC)
- Translate VOC (spoken and unspoken) into measurable CTQs (Critical To Quality).
- Use VOC to set problem and goal statements and to prioritize stakeholder alignment.

Problem statement structure (pyramid): Intro → negative (financial) consequences → root-cause lead-in.

Goal statement guidance (one per CTQ): SMART — Specific, Measurable, Achievable, Relevant, Time‑bound.

Problem & goal framing
- Connect problems to financial impact to secure stakeholder buy-in.
- Use one SMART goal per CTQ; recommended initial timeline ≈ 3 months (original Six Sigma guidance was ~6 months; adjust per project).
Benchmarking and targets
- Use comparable internal/local benchmarks.
- If the gap to benchmark is large, consider setting the initial target = benchmark gap × ~70% (round as appropriate).
- Renegotiate targets only with the process owner and Master Black Belt.
Project governance and process mapping
- Name the process as verb + noun.
- High-level mapping using SIPOC (presenter also referred to it as “CYP O / supplier-input-process-output-customer”).
- Create detailed process maps at L1–L4 granularity.
Measurement System Analysis (MSA / R&R)
- Assess repeatability (same operator), reproducibility (different operators), and accuracy/bias vs true value.
- If MSA issues exist, standardize measurement before deep analysis.
Standardization playbook for measurement & operations
- SOPs: atomic-level, step-by-step, include photos.
- Formal training and audits.
- Recognition approach: public praise for compliance; private coaching for non-compliance.
Control and capability analysis sequence
- Measure → check stability (control charts) → then capability (vs customer spec).
- Do not run capability analysis on an unstable process — first investigate and resolve special causes.
Statistical tools referenced
- Continuous data: I‑chart (individual control chart).
- Proportions/discrete (good vs bad): P‑chart.
- Capability commonly expressed via sigma level (Z).
Certification / AI guidance
- Typical pathway: white → yellow → green → black belts (content overlaps; tests are separate).
- Use AI tools only after full domain knowledge (e.g., after Black Belt + AI belt) to avoid incorrect outputs.

Key metrics, KPIs, targets & timelines

CTQs from the case study
- CTQ1: Delivery time (example customer target used = 10 minutes).
- CTQ2: Pizza quality (defect / bad pizza rate).
Delivery time sample metrics (example)
- Sample mean delivery time ≈ 14.73 minutes (sample ~30 observations).
- Customer upper spec = 10 minutes → process is stable but not capable.
- Example calculated sigma level for delivery ≈ −4.52 (indicates a large portion of outputs outside spec).
Quality / defects metrics (example)
- Aggregate defective rate ≈ 10.2% (0.1022) across sample days.
- Quality sigma level shown ≈ 1.27 (positive but low; not capable by typical industry expectations).
Industry references for sigma levels
- Four‑sigma: reasonable market reference for good performance.
- Six‑sigma: aspirational / near-zero defects (~2 parts per billion).
Recommended project duration
- General guidance: ~3 months per Yellow Belt project (some projects may require up to 1 year).

Concrete examples & case study takeaways (Pizza case study)

VOC → identified two CTQs: delivery time and pizza quality.
SIPOC example: named process = prepare/produce pizza; output = vegetarian pizza; key inputs = flour, tomato, cheese, oven, thermostat, chef, recipe.
Process mapping: high-level then detailed maps; sampling and measurement checks implemented.
MSA illustration: different observers counted letters differently — example used to show repeatability, reproducibility, and bias.
Stability analysis: I‑chart for delivery time showed a stable process with mean > spec (predictable but poor performance).
Capability analysis: histogram + spec limits demonstrated distribution not within spec — major opportunity for improvement (car-in-garage analogy).
For quality (discrete defects) used a P‑chart to show control chart behavior for proportions.
Business tradeoff example: ultra‑low‑cost airline sells cheap tickets but accepts poor reliability — acceptable low capability if customer expectations and price align.

Actionable recommendations / tactical checklist

Before using historical or process data
- Verify measurement system reliability (repeatability, reproducibility, bias).
- If MSA issues exist: standardize measurement (SOP + training + audit) before analysis.
Writing problem & goal statements
- Use the pyramid for problem statements; include financial negative consequences.
- Use SMART for goals — one goal per CTQ; make them timebound (start with ~3 months).
Target setting
- Use comparable benchmarks (avoid unrealistic superstars).
- If the gap is large, consider an initial target equal to ~70% of the gap.
- Negotiate targets only with the process owner and Master Black Belt.
Process mapping & SOPs
- Map at L1–L4 levels.
- SOPs: atomic steps, pictures, plain language, and tested by non‑experts (avoid curse of knowledge).
Data sampling
- Use sufficient, randomly selected samples from correct segments.
- Sampling is acceptable and powerful when done correctly.
Control / capability analytics
- First perform stability analysis (control chart over time); investigate special causes before capability.
- Continuous measures → I‑chart; binary defectives → P‑chart.
- Translate results to sigma level and compare to VOC/spec limits.
Behavior & leadership
- Translate operational problems into financial impact to align stakeholders.
- Combine standardization with respectful coaching and negotiation.
- Use public praise for compliance and private, constructive feedback for non-compliance.
Use of AI
- Only after domain mastery — premature use risks incorrect outputs.

Other operational / organizational notes

Certification expectations: instructor expects ~20 hours classroom + ~10 hours self‑study; tests are proctored and tracked.
Marketing/channel: recruitment/advertising cost for the course increased due to platform/tax changes; choice of platform (Meta vs Google) involves practical tradeoffs.
Upcoming topics: root cause analysis planned for the next session (priority after the Measure phase).

Presenters / sources

Presenter: Marcelo Machado Fernandez (Marcelo). Support: session team (notably his niece Lara).
Referenced organizations / tools / authors: ASQ (American Society for Quality), Minitab (statistical software), Edward Deming, Joseph Juran, and a referenced author noted as Gigu/Gijo Anthony.