Singapore · Packaging & Food Processing
Tetra Pak
“Every plant had its own truth. None of them matched.”
Disconnected production data across multiple plants unified into a single real-time operational layer - giving plant managers and regional leadership a shared source of truth for the first time.
Key Results
Multi-site
Unified data
Single source of truth
Real-time
Cross-plant visibility
Was weekly manual
Standardised
KPI definitions
Cross-facility benchmarking enabled
Minutes
Deviation alerting
Was hours in shift-end report
Tech Stack
The Situation
At a global packaging and food-processing operation the scale of Tetra Pak, "cross-facility visibility" sounds simple. In practice, it means reconciling data from plants in multiple geographies, each running its own system stack, each with its own definition of the same KPIs, each reporting on a different cadence. When regional leadership asked "how is OEE trending across the network this quarter?" - the answer required an email to four plant operations leads, four different spreadsheets, three days of manual alignment, and a final number that everyone had slightly different confidence in. By the time any cross-plant insight was available, the operational window had passed.
In multi-site manufacturing and packaging operations, this is the standard, not the exception:
- ✓
Your regional operations view is assembled manually from plant-level exports - every single period
- ✓
The same KPI (yield, OEE, reject rate) is calculated differently in each facility - making benchmarking meaningless
- ✓
Plant managers have their own reporting - but there's no single view across the network for leadership
- ✓
Identifying your best and worst performing lines requires a multi-day data reconciliation exercise
- ✓
Production deviations are only visible after a shift-end report - not in time to intervene
- ✓
"We have all the data - we just can't see it all in one place" - the most common sentence in multi-site manufacturing
If three or more of these describe your operation, you're looking at the right case study.
The Root Problem
- 1
Multiple manufacturing facilities operating with separate, incompatible data systems and different KPI definitions
- 2
No consolidated cross-facility view - regional leadership relied on manual email-based data requests from each plant
- 3
OEE, yield, and throughput were calculated differently per site - rendering cross-plant benchmarking unreliable
- 4
Reporting cycles were week-long reconciliation exercises; by the time data was aligned, the operational window had passed
- 5
No alerting or deviation detection - issues were discovered in retrospective reports, not in real time
How We Fixed It
Define a common production data model across all facilities
Before building anything, we established a unified production ontology - standardising metric definitions for OEE, yield, throughput, quality rate, and downtime classification across all facilities. This common data model became the semantic foundation everything else was built on. Without agreed definitions, unified reporting is just a faster way to get the wrong answer.
Build a multi-plant integration layer on Microsoft Fabric
We built CDC-enabled Azure Data Factory pipelines pulling structured production data from each facility's source systems - MES, ERP, and SCADA feeds - into a centralised Delta Lake on Microsoft Fabric. Change data capture ensured near real-time freshness without disrupting operational systems at each site. All raw data now lands in OneLake with full lineage tracking.
Unified operations dashboard - plant and regional views
Built a Power BI semantic model and dashboard suite giving plant managers individual facility views and regional leadership cross-plant benchmarking - all from a single, governed dataset. Drill-through from regional → plant → production line level in one click. The same number, the same calculation, regardless of which view you're in.
Production deviation alerting
Configured threshold-based alerting for production deviations - line stoppages, OEE dips below target, quality reject spikes - routed to the relevant operations lead via Teams notifications. The window between a deviation occurring and the right person knowing about it dropped from hours to minutes.
Measured Outcomes
Cross-facility visibility
Weekly manual email requests
Real-time unified dashboard
↑ Key win
KPI definitions
Different per facility - inconsistent
Standardised across all sites
↑ Key win
Reporting cycle
Week-long reconciliation exercise
Automated, always current
Deviation detection
Shift-end retrospective report
Real-time alert to operations lead
Benchmarking
Meaningless - incomparable definitions
Genuine cross-plant comparison
What This Means For You
What this means for multi-site manufacturing and packaging operations
The inability to see across facilities isn't a data problem - it's a data model problem. Each plant accumulates data. The gap is the governed, unified layer that sits above all of them with agreed definitions and a single calculation logic. Microsoft Fabric provides the architecture for this at enterprise scale without requiring a multi-year transformation programme. If your regional leadership is still assembling cross-plant performance views manually, the fix is structural, not technological - and it's achievable in a quarter.
Next Step
Is this your situation?
Book a 30-minute call. No slides, no pitch. We'll look at your specific setup, tell you what's causing the problem, and what a realistic fix looks like - including timeline and cost range.