Company: Novo Nordisk
Role: Lean Partner / Transformation Lead
Context: Regulated pharmaceutical manufacturing
Scope: End-to-end deviation handling across Production, QA, QC, and support functions
Context
In regulated pharmaceutical manufacturing, deviations are unavoidable. What determines operational stability is how effectively deviations are investigated, aligned, and closed.
At site level, deviation handling was compliant and thorough, but closure lead time was long, and the majority of deviations were overdue, directly affecting batch release, audit readiness, and operational focus.
A widely accepted belief existed about the root cause — but it had never been tested.
The Dominant Myth
The prevailing assumption was:
“QC lead time is the main driver of late deviation closure.”
The logic made sense:
- QC was a central function
- QC activities were outside local control
- Samples and data often came late
As a result, improvement discussions focused away from the site and toward external constraints.
What the Data Actually Showed
I tested this assumption using deviation lifecycle data.
- Analysed QC response lead time vs total deviation lead time
- Tested correlation across deviation categories
- Compared deviations with and without QC involvement
Result:
👉 No meaningful correlation between QC lead time and overall deviation lead time.
QC was not the constraint.
This shifted the focus from assumptions to evidence.
The Real Problem
The data revealed a different, internal driver:
- Significant misalignment between Production and QA supporters on:
- What constituted a “good” deviation
- Level of detail and evidence required
- Inconsistent expectations within QA itself
- Deviations were frequently returned for rework
- In some cases up to 8 times
- Each return added ~1 week on average to overall deviation lead time
Deviation handling was not slow because of investigation complexity —
it was slow because of decision and expectation churn.
My Mandate & Authority
I was asked to reduce deviation lead time without compromising quality or compliance, with responsibility to:
- Identify true delay drivers using data
- Eliminate rework caused by unclear expectations
- Align QA and Production on decision quality
- Reduce overdue deviations structurally, not temporarily
I worked hands-on with QA, Production, and support functions, aligning leadership around shared evidence and operating rules.
What I Did
1. Replaced assumptions with hypothesis testing
- Tested the QC-delay hypothesis using real data
- Made correlation (or lack thereof) visible to stakeholders
- Shifted improvement focus from external constraints to internal controllables
2. Clarified expectations and decision quality
- Aligned QA supporters on what “good” looks like for deviation investigations
- Clarified acceptance criteria to reduce interpretation-driven returns
- Established right-first-time expectations between Production and QA
This reduced iteration loops dramatically.
3. Introduced flow and transparency
- Implemented KPIs for deviation lead time, returns, and overdue rate
- Made rework loops visible instead of hidden
- Used data to intervene before deviations became overdue
Deviation handling moved from reactive escalation to managed execution.
Results & Impact
Hard outcomes
- Overdue deviations reduced from ~95% to ~20%
- Average deviation lead time significantly reduced
- Fewer release delays caused by open deviations
- Reduced rework cycles and decision churn
Structural outcomes
- QA and Production aligned on investigation quality expectations
- Deviation handling governed as a decision system, not a document exercise
- KPI-driven follow-up embedded into routines
- Improvements sustained without increasing audit risk
Key Insight
In regulated environments, deviation delays are rarely caused by external bottlenecks.
They are caused by misaligned expectations, decision ambiguity, and hidden rework loops.
By testing assumptions and fixing decision quality at the interface between functions, speed and compliance improved together.
Öner TANK 