**3. Techniques to optimise quality**

Research findings suggest the basis for quality improvement needs to be understood by everyone at all levels of production ([24], p. 460; [17, 28]), for instance in a milk production plant, from milk intake all the way through to finished product. Dr. Kaoru Ishikawa defined '*the seven improvement tools*' in response to the realisation that everyone in a company had to become involved in improvement works, thus they needed to be simple and effective tools [27]. The set of tools within the toolbox methodology can vary across the literature; however, the end objective is always the same whereby they are systematically designed for problem solving and optimisation. As the 'Define' phase of the Six Sigma methodology states, the collection of data is a fundamental step in the road to quality improvements [29], Ishikawa's Seven Improvement Tools also demonstrated the importance of this step [27]. Histograms

and Pareto charts, can be used as visualisation tools, as an important part of data analytics is to represent data efficiently and that can help in the decision of which problem should be solved first [30]. Once data has been analysed the root cause of problems within a process has to be defined. This can be achieved through the use of a cause and effect diagram [27, 29]. A tool that allows the description of causes that can possibly lead to quality problems and allows for the investigation of each cause, one problem at a time. One common problem however, with the use of cause and effect diagrams, is that a quality team can start with the major problems while the root causes of the minor ones become harder to solve [24, 31]. Stratification is another problem solving tool within the seven improvements toolbox, based on classifying data from different sources into subgroups and obtaining important information for improvement work [27]. Scatter plots can be used when stratification cannot, to identify process or product variations due to explanatory variables [32]. Finally, the use of control charts in quality optimisation has two main objectives, the first being to identify causes of variations and secondly to quickly detect when variations occur in a stable process [31].

#### **4. Laboratory activities**

A vital part of any quality system are laboratory quality standards [33, 34], and an increasing number of laboratories within the dairy industry are operating to accreditation standards. Calibration and testing activities can either be partially or fully accredited [35]. In Ireland, accreditation can be granted by the Irish National Accreditation Board (INAB), a voluntary scheme open to laboratories within the country, which offers accreditation to the international standard outlaying requirements on testing, personnel, confidentiality, laboratory facilities and environmental conditions [36]. In recent times, clinical laboratories have considered laboratory turnaround time (TAT) an important key performance indicator (KPI) of laboratory performance [37], defined as *the time it takes for a test to be ordered to the time it takes for a result to be delivered*. This is a topic however, that has not been widely researched in relation to food, and more so dairy manufacturing. Activities are carried out within dairy manufacturing plants with little consideration to downtime or the effect of sampling and analysis on personnel labour hours [34]. A study carried out in 2014, on methods of identifying unnecessary laboratory testing in a tertiary care hospital deduced that a user approach as well as a systems approach would ultimately reduce over-testing [38]. Within Dairy manufacturing plants however, over testing is more times than not, due to out of specification test results (OOS), defined as a test result that falls outside of the established testing criteria [39]. While an initial OOS result does not mean a failed batch, the concept of 'testing into specification' rather than applying an investigative approach to determine the root cause of such a result, often leads to unnecessary over analysis, ultimately negatively affecting cost and laboratory TAT. Investigating OOS test results should be thorough and timely and include the use of scientific decisions, justifications and risk based analyses [40]. Furthermore, applying a strategic in-process sampling system would minimise the level of OOS in final product. In-process sampling should always start at the beginning of the process to allow for early detection of processing issues. Finally, managing test results can be a complex process if it is not performed in an accurate manner. Improving how test results are recorded and communicated can achieve savings in

staff time and improve the overall reliability and communication of the results. OOS results can be due to a number of reasons including failed equipment or reagents, in process issues and human error.
