In an industrial context, a data culture is the organizational ability to transform process measurements into repeatable and auditable decisions. Consolidating this practice requires structured routines, clearly defined roles, and integration standards widely accepted by the market. This ensures that data collection, contextualization, and analysis do not depend on individual knowledge or parallel spreadsheets.
The ISA-95 standard remains the main reference for structuring integration between the shop floor and corporate systems. The standard establishes functional models and clear boundaries between automation levels, enabling organized communication between industrial control systems, MOM platforms, and corporate systems such as ERP. This structure offers several advantages:
- Reduces integration risks;
- Lowers interface costs;
- Reinforces the separation between the manufacturing domain and the business domain (a principle reiterated in the most recent updates of the standard).
From an analytical maturity perspective, it is worth recalling Monica Rogati’s synthesis of the AI Hierarchy of Needs. She organizes data and artificial intelligence initiatives into hierarchical layers, where infrastructure and data quality precede more advanced analytical applications. The model functions as a pragmatic guide for organizations seeking to scale analytics without skipping essential steps (Rogati, 2017).
Cybersecurity must also be part of the design of industrial integrations. The IEC 62443 series provides the foundation for this approach by structuring the security architecture into zones and conduits. In addition, it establishes requirements for asset owner security programs and for the development and integration of systems and components. Recommended practices include:
- Structured patch management;
- Network segmentation through an industrial DMZ;
- Periodic validation of backup and restore routines;
- Essential measures to reduce risks and ensure operational continuity.
Data Governance Applied to Smart Dairy Manufacturing
For a data culture to move beyond discourse and begin guiding operational routines, organizations must structure a simple and coherent information flow. In this context, data governance is consolidated through:
- Clear definition of responsibilities for each dataset;
- Adoption of a common data dictionary;
- Formalization of rules for access, sharing, and information retention.
Recent studies in smart manufacturing indicate that transmission failures and data heterogeneity, when not prevented, compromise operational reliability and may lead to incorrect decisions. For this reason, it is more effective to monitor and correct information quality at the point of capture than to attempt to fix it later in the data flow.
In contexts involving food safety and traceability, standards such as ISO 22000 and FSSC 22000 provide the management system framework necessary to ensure information consistency and traceability. As a result, the same data used in operations can also support audits, batch genealogy, and quality release processes.
In the dairy industry, this logic translates into design decisions that treat recipes and process states as controlled configurations, with version and change history aligned with the principles of the ISA-88 standard. In practice, this means defining which variables must be measured and which criteria guide decisions at each stage of cheese production:
- During pasteurization, consistent collection of outlet temperature and holding time data is directly associated with batch release;
- In the coagulation stage, pH, temperature, and process time determine the curd cutting point;
- During brining, parameters such as salinity, temperature, and immersion time define the desired salt content in the product;
- During maturation, chamber temperature and humidity must remain recorded and auditable against the profile specified for each cheese.
CIP in Cheese Production: Data-Driven Cleaning Control
In the cheese manufacturing process, equipment sanitation typically employs Clean-in-Place (CIP) systems guided by process measurements. In this approach, cleaning endpoints replace fixed time buffers, allowing stages to be determined by variables monitored in real time. Conductivity is widely used to distinguish between product, water, caustic, and acid, as well as to confirm the end of rinse stages.
In situations that require greater sensitivity in the transition between product and water, optical turbidity sensors can be used to detect this change more precisely. This strategy is often described in supplier application notes and represents an established practice that enables controlled solution reuse while reducing process time, water consumption, and chemical inputs.
When this data begins to be collected continuously and in a structured manner, it becomes possible to observe operational patterns that previously remained invisible in day-to-day production. Variations in parameters such as process temperature, curd pH, cutting time, or brine concentration can be correlated with yield, texture, and the sensory profile of the cheese. As a result, historical analysis ceases to be merely a record and begins to support more consistent process adjustments across shifts, lines, and production campaigns.
Consolidating this information also expands visibility into operational efficiency and process stability. By integrating data from different stages, from milk reception to maturation, it becomes possible to:
- Identify recurring deviations;
- Understand their causes;
- Prioritize improvement actions based on evidence.
In increasingly connected industrial environments, a data culture ceases to be only an organizational concept and begins to be directly reflected in process performance. The ability to measure, contextualize, and interpret critical variables throughout cheese manufacturing creates the conditions for more stable, traceable, and efficient processes.
Learn more about ST-One.