The Problem
A global contact centre running 160+ agents across four regional sites generates a lot of process variation. Different centres had developed their own handling practices, escalation habits, and documentation standards. The result: inconsistent customer outcomes, tickets bouncing between agents unnecessarily, and no structured way to capture what customers were actually telling us.
Two problems needed solving simultaneously. First, operationally: defect rates — defined as tickets requiring rework, incorrect routing, or unnecessary escalation — were too high and increasing as the operation scaled. Second, strategically: the business had no Voice of Customer programme. Customer feedback existed, but it lived in ticket notes and agent memories.
"We had 160 people talking to customers every day and no system for turning those conversations into intelligence. That’s a data problem as much as a process problem."
The Six Sigma Approach
We applied DMAIC across all four centres. Define: A defect was any ticket requiring more than one agent touch before resolution due to incorrect initial handling, routing, or documentation. Measure: Three months of ticket data, segmented by region, product line, and category. Defect rates varied significantly by centre — meaning process variation, not a systemic product issue. Analyse: Three primary root causes: inconsistent initial triage questions, unclear escalation thresholds, and inadequate handoff documentation. All fixable. Improve: Standardised triage scripts, rebuilt escalation decision trees, mandatory handoff documentation. Control: Monthly defect rate reporting by centre, visible to all regional managers.
Defect Reduction by Phase
The Voice of Customer Programme
We built a lightweight tagging system requiring agents to tag tickets with a secondary reason code from 18 categories covering product gaps, documentation deficiencies, and support experience failures. This added roughly 30 seconds per ticket. The return: a monthly VOC report distributed to product, engineering, and senior leadership showing the actual distribution of customer pain points. Three product-level changes were prioritised based directly on VOC data within the first year.
"VOC isn’t a survey. It’s a data collection discipline built into the operation. If your support agents are tagging tickets correctly, you already have more customer intelligence than most product teams see in a year."
What We Learned
Six Sigma works in contact centres if you define defects carefully. The methodology requires adapting the defect definition to something operationally meaningful and measurable.
The control phase is where most programmes fail. DMAIC without the C produces a one-time improvement followed by drift. Published monthly defect rates by centre — with names on them — were what made the improvement hold.
VOC data is only valuable if it reaches decision-makers. We formatted it for product and leadership audiences and put it on their desks monthly. Data that never leaves the operations team changes nothing.