DATA-DRIVEN MANUFACTURING


consistent, and data from an unstable process is unreliable as a basis for decisions.


•


Digitising an inefficient process will simply accelerate the same problem: Skip process improvement and you get a faster, more connected version of existing inefficiencies.


•


A gap opens between shop-floor data and management decisions: Data is collected at machine level but never reaches those responsible for capacity or quality targets, so the infrastructure serves reporting rather than decision-making.


•


The state of the operational technology (OT) environment itself: A typical confectionery line includes machines from multiple manufacturers with different PLCs, some with no connectivity, and extracting reliable data requires IT and OT expertise that is often split across teams.


•


Larger manufacturers will often carry a patchwork of systems built-up through acquisitions and one- offline investments: Adding a new platform, without first addressing this fragmentation will increase complexity rather than clarity.


Making good use of data According to Dr Stuart Gilby, Director of FMCG at 42 Technology, a UK-based product development and manufacturing innovation consultancy,


DATA NEEDS TO BE TIED TO AN OUTCOME, BUT ALSO TO SOMETHING YOU CAN CONTROL


because it is important to capture all the relevant contributing variables. Are ambient conditions recorded? Do you


significantly improve yield, consistency, and cost efficiency. Stuart argued that one of the


key challenges facing confectionery manufacturers trying to optimise their processes today is the need to gain a good understanding of what levers affect critical process and quality parameters. “Data needs to be tied to an outcome, but also to something you can control,” he says. Offering an example, Stuart pointed out


the competitive advantage


lies not in having more data, but in making better use of what you have. Those who combine strong data foundations, targeted use cases, and operator engagement can


28 • KENNEDY’S CONFECTION • JULY 2026


that, if the optimisation objective is fewer customer returns, then it is necessary to have unique traceable batch codes on each roll of packaging so that a product identified in the field can be related back to a manufacturing run. “Tracking which batches have been returned is a start, but a correlation to why they are being returned and what process you can control to reduce those returns is essential,” he continued. When optimisation targets are closer to the production line – such as uptime, waste reduction or scrap rate – data can be collected in real-time, but there will still be complexities to navigate


know which mould/tool/carousel position a reject came from? Do you know exactly what the settings were when the product was made and, crucially, if the setpoints were changed, what was the observation that led to the operator making the change? “This data then needs to be analysed to find what actual changes made an impact,” says Stuart. “Using statistical process control (SPC) with machine learning/AI in the right places can help. SPC helps maintain consistency, while machine learning can uncover non-obvious correlations such as seasonal humidity that can affect sugar crystallisation. Now the data has provided you with a desired outcome, you need to provide clear, real-time visualisation and simple decision rules to frontline teams. But if this doesn’t align with operating experience it may be distrusted.” To overcome this possibility, Stuart advises


that operators are given training to help them understand the how and the why behind the decisions. Starting with high-impact cases first will help build trust across the team. To turn your data ambitions into reality,


Stuart advises focusing on the following: Think about your data governance as


soon as possible. A clean data structure with consistent labelling is essential before advanced analytics. Build slowly and adapt as you go. Start with pilot projects on one line or process to demonstrate ROI before scaling. Use data scientists to accelerate your


learnings. If you don’t employ any, consider technology providers or consultants but


Page 1  |  Page 2  |  Page 3  |  Page 4  |  Page 5  |  Page 6  |  Page 7  |  Page 8  |  Page 9  |  Page 10  |  Page 11  |  Page 12  |  Page 13  |  Page 14  |  Page 15  |  Page 16  |  Page 17  |  Page 18  |  Page 19  |  Page 20  |  Page 21  |  Page 22  |  Page 23  |  Page 24  |  Page 25  |  Page 26  |  Page 27  |  Page 28  |  Page 29  |  Page 30  |  Page 31  |  Page 32  |  Page 33  |  Page 34  |  Page 35  |  Page 36  |  Page 37  |  Page 38  |  Page 39  |  Page 40  |  Page 41  |  Page 42  |  Page 43  |  Page 44