While olive oil production is facing many challenges and prices are rising dramatically, the demand for non-invasive analytical methods is high. Spatially offset Raman spectroscopy (SORS) can be a potential method for sustainable food analysis as it can penetrate different types of containers while providing a good spectrum of the food in question. In this study, we have developed a SORS-based method for the authentication of olive oils that is also suitable for the high variability of packaging found in practice. Based on a dataset of verified oils from four sample groups and mixtures of this samples, we developed an analytical strategy using plotting, principal component analysis, and a classification and regression model. This analysis strategy was tested first using different mixtures of olive and sunflower oils. We were able to recognize 80 % of adulterated olive oil samples with 30 % added sunflower oil and 60 % of the samples adulterated with 10 % sunflower oil. Afterwards the strategy was tested in food inspections of different companies. The results show that our strategy for on-site analysis was successful and was positively evaluated by the parties involved. In an additional validation step, we analysed 30 samples from online retail where we were able to distinguish between actual adulterated and genuine olive oil samples.