Scoping a Green Belt Project in a Distribution Centre
A Green Belt in a regional distribution centre is asked to 'reduce picking errors' after customer complaints rose from 1.2% to 2.4% of orders over six months. The sponsor wants a project charter by Friday. Available data: a monthly complaint count by customer, a list of 14 pickers, two shift patterns, and a warehouse layout with three pick zones. There is no defect-level data (which item was wrong, which picker, which zone). The operations manager says the cause is 'new agency staff who don't care'. The Green Belt has 4 hours per week of protected time and one part-time data analyst. The site runs 6 days a week and cannot stop picking for measurement.
1. Draft a problem statement and goal for the charter that is defensible given the data available, and explain what you deliberately exclude and why.
Read the worked answer
Problem statement (measurable, bounded, no cause): 'Between January and June, picking errors reported by customers rose from 1.2% to 2.4% of orders (roughly X orders per month at current volume), generating Y complaints. The mechanism, picker, and zone are not currently recorded.'
Goal: 'Reduce customer-reported picking errors to ≤1.2% of orders within 12 weeks, sustained for 8 weeks.'
Deliberately excluded:
- Any causal claim ('agency staff') - that is the operations manager's hypothesis, not evidence, and putting it in the charter pre-judges the Analyse phase.
- A target expressed in absolute counts only - percentages let you compare across volume changes.
- Scope creep into 'all warehouse quality' - bound it to picking errors reported by customers in the three zones.
The charter must also flag the measurement gap as a known risk: without defect-level data you cannot stratify, so a Measure-phase data collection plan is a charter commitment, not an afterthought.
2. The sponsor pushes back: 'Just fix the agency staff training, we don't have time for a project.' How do you respond in DMAIC terms?
Read the worked answer
Acknowledge the sponsor's hypothesis is plausible and worth testing, but explain the risk of acting on it now:
- No baseline by cohort. You cannot currently show whether errors concentrate among agency staff, new starters, or a specific zone. If the true driver is, say, a mislabelled shelf in Zone 2, retraining agency staff consumes the 4 hrs/week and the error rate stays at 2.4%.
- No way to prove improvement. Without a baseline split by cause, any drop after training is unattributable - the sponsor cannot claim the win, and if errors rise again there is no signal to act on.
- Cheap compromise. Propose a 2-week Measure sprint: add a defect code (item wrong / quantity wrong / wrong order) and picker ID to the existing scan-out step, costing the analyst ~6 hours. That converts the sponsor's hypothesis into a testable one and gives the project a control chart baseline.
Frame it as protecting the sponsor's investment, not blocking action.