Quality Defect Investigation
A food manufacturer received a customer complaint
about slight texture variation between batches
A yoghurt manufacturer faced a critical customer complaint after two failed investigations. The retail contract was at risk.
The situation
The same defect had returned two times. Nobody knew why.
A mid-sized yoghurt manufacturer in Germany had been supplying a major retail chain for four years. In February, the retail buyer raised a formal complaint: texture inconsistency across multiple batches of their 200ml product line. Returns were increasing, and a formal CAPA report was required and an external audit scheduled for the following month.
Two previous investigations had both concluded the cause was supplier milk fat variation. Both times, the supplier corrected and the case was closed. Both times, the defect returned. A third failure would almost certainly put the contract at risk.
The investigation process was entirely informal - no structured cause framework, no evidence trail, and no way to prove what had been ruled out and why.
Define
The problem was narrower than anyone had realised.
The Data Wizard chatbot guided the quality manager through a structured problem definition before the investigation could proceed. This narrowed it down to particular batches. Pulling the production schedule revealed something nobody had spotted before: every affected batch had been produced after a maintenance cleaning window.
This ruled out supplier variation as the cause. This was a process issue. The team just didn't know which one yet.
Identify
A handfull of promising causes was generated. All relevant to the case.
The AI-assisted 6M Ishikawa diagram surfaced 5 promising cause candidates. Previous investigations focused on a total of 25 causes, of which none was the root cause.
Among the causes that surfaced for the first time: post-cleaning temperature recovery time, time between CIP completion and batch start, and batch start sequence adherence on high-output days.
Evaluate
Fact based likelihood assessment narrowed down the focus to three high runner.
Likelihood scoring flagged a cluster of high-probability causes around the CIP cycle and batch start sequence.
On high-output days, batches were starting within 18–22 minutes of CIP completion well below the SOP-required 35-minute hold. Equipment surface temperature was still below 68°C. When a batch started cold, milk didn't reach the correct temperature in the first 40 minutes resulting in lower viscosity. The correlation with the affected batches was 100% consistent.
Root cause confirmed: post-cleaning temperature insufficiency at batch start, caused by consistent SOP deviation on high-output production days. The supplier had nothing to do with it.
Resolve
Three corrective actions. All closed in 8 days.
A temperature interlock was installed on the production line start sequence the batch cannot begin until equipment reaches ≥68°C. The SOP was revised to replace the ambiguous time-based hold with a temperature-based condition. All shift operators were briefed and retrained on the revised procedure within five working days.
The results
Root cause found. Customer retained. Defect eliminated.
The investigation that had stumped the team for weeks was resolved in a single day with a level of rigour and documentation the team had never previously produced. The retail buyer formally confirmed satisfaction with the outcome and renewed the supply contract for a further two years.
No further texture complaints have been received in the past months since the corrective actions were closed. Following this case, the quality director standardised Causetec as the team's primary investigation tool across all production lines.
AI-guided Root Cause Analysis for quality and operations teams. Structured investigations, consistent quality, fewer repeat incidents.
Causetec OY 2026© | All rights reserved.