Survey and collect data already available, generate data where it is not yet available, store it in databases, data warehouses and data lakes, make intelligent dashboards
Operational context
In an Italian plant of about 3,000 square meters, belonging to a multinational company specializing in the production of feed for livestock, an ambitious path toward the digitization of logistics-production processes has begun. The plant comprises four main departments:
- Filtration
- Mixing
- Microencapsulation
- Quality Control (QC) and Packaging
The production layout is articulated and includes automated packaging lines, where the operator only has to feed the machines with batches of mixture (powder or liquid) and the corresponding packages. The other plants, while having standardized operating cycles – similar in logic to those of a large industrial washing machine – still require considerable human effort in loading and unloading operations.
The workforce consists of:
- 48 operators
- 6 maintenance technicians: 2 mechanical/pneumatic, 2 electrical/electronic, 1 construction
- Two warehouses, one for raw materials (MP) and one for finished product (PF). Semi-finished products are stored by production operators in high-turnover department buffers.
Annual maintenance expenditure is around 2 million euros.
Instrumentation and data collection
There are already a number of measuring instruments in the plants, distributed in the various departments:
- Thermocouples
- Electronic pressure gauges and pressure transducers
- Flow meters
- Hygrometers
- Vibrometers and accelerometers
- Load Cells
All of these devices are already digital or equipped with electronic sensors capable of converting physical quantities into electrical/digital signals. However, until recently the data collected by these instruments were not systematically acquired, stored or integrated into a centralized information system.
Existing information systems
The IT organization provides:
- A central management system (ERP type) with the modules:
- Purchases
- Materials Management
- Personnel management
- Production
- Shipping
- Billing and accounting
- Finance and management control
- A warehouse application with barcode loading/unloading capabilities and dynamic location management
- Excel sheets for:
- Maintenance
- Quality Control
- Reporting with pivot tables and manual dashboards
The plant’s enterprise network architecture is based on a combination of traditional technologies and modern solutions, with the following main elements:
- Industrial WiFi
- Wired Ethernet network
- VPN (Virtual Private Network)
- On-premise server + hybrid cloud
- Firewall and network segmentation
- Industrial Access Points
The challenge: building a data infrastructure
The first step was to survey all the data already present in existing facilities and digital systems. This made it possible to:
- Identify active and digitizable data sources
- Map the physical variables measured by the instruments (temperature, pressure, vibration, flow rate, humidity, weight, etc.).
- Understanding where data were not yet being generated or collected in a structured way
Next, the data acquisition and centralization phase began, with the goal of:
- Update and integrate firmware and protocols of field devices to ensure interoperability and proper data transmission
- Normalize and standardize data from devices with different formats and protocols through dedicated middleware software
- Gather available information from the field in real time
- Generate information on manual machine loading and unloading operations by equipping operators with tablets or barcode/QR code readers to record activities performed and associate batches moved, with integration to management or a dedicated app.
- Historicize them in relational databases, data warehouses, and data lakes
- Analyze them through intelligent dashboards and business intelligence tools
Toward a “smarter” factory
Dashboards, built with simple and accessible tools such as Microsoft Power BI and Excel connected to corporate databases, now enable:
- Monitor plant efficiency in real time
- Identify anomalies or process drifts through analysis of historical data
- Support maintenance decisions based on objective indicators and KPIs
- Improve production scheduling according to plant availability and conditions

Results and prospects
Thanks to these first steps, the plant has been able to:
- Raising the level of digital awareness among staff
- Reduce plant downtime related to unplanned failures
- Initiating a transition to a predictive model of maintenance and production
- Simplify reporting activities, resulting in time savings and increased data reliability
Below are the main KPIs, before and after the intervention
KPI (Name and Unit of Measurement) | Value Before Intervention | Value After Intervention |
Lead Time Order-End Production (days) | 3.5 days | 2.0 days |
Plant Utilization Rate (%) | 68% | 82% |
Unplanned Plant Stop Index (no./month) | 9 stops | 3 stops |
% Orders Fulfilled on Time | 83% | 97% |
Accuracy of Warehouse Data (%) | 76% | 93% |
Average Manual Reporting Time (hours/week) | 11 hours | 3 hours |
The project now continues with the integration of machine learning algorithms to anticipate maintenance needs and optimize production cycles. This case shows that even in SME contexts, starting a path to digitize logistics-production processes can start from what already exists: raw data, partially digital tools and Excel sheets. All that is needed is to put things in order, integrate, collect, visualize and improve.