The integration and application of causal machine learning models in mining and mineral extraction processes have always been a major problem in the mining industry due to the uncertainties in the geological models such as grade variations, geotechnical issues, mine operations challenges, and metal price fluctuations. Current conventional methods such as the use of stability charts, numerical models and correlation based machine learning are employed in controlling unplanned ore dilution. But the limitations, complexity and computational challenges in the conventional methods result in poor interpretation of outcomes. There are currently contradictory views on the model that are effective for the mining industry in mitigating unplanned ore dilution. Paucity of knowledge, therefore, exists on the specific machine learning model that can be practically applied in mining setting to minimise unplanned ore dilution considering all the dynamic planning parameters, in an efficacious manner for maximum return. The key research questions and hypothesis are: 1. How effective is causal models compared with the other conventional mine planning techniques to obtain high cash flows and maximise output, taking keynote of historical realisations? 2. What fundamental principles limit the transition of most mining companies/operations from the conventional ore dilution control methods to causal models? 3. How can the limitation be eliminated for improved returns? The objective of this study is to investigate the application of causal model in mitigating unplanned dilution in mining for maximum return while addressing the above research questions. Links to relevant literature will be highlighted.