This study aimed to understand the distribution of lithium and other elements in the Beauvoir granite, which is a granite enriched in lithium (Li), tin (Sn), niobium (Nb) and tantalum (Ta), and develop a reliable method for lithium quantification. These granite samples underwent various mineral processing techniques to produce a large dataset with different lithium concentrations and different mineral matrices. Thin sections were analysed using optical microscopy and micro X-ray fluorescence imaging to determine the distribution of potassium (K), rubidium (Rb), and cesium (Cs) in different mineral phases. Both energy dispersive X-ray fluorescence (ED-XRF) and laser induced breakdown spectroscopy (LIBS) measurements were conducted using handheld instruments on the same samples. The different resulting spectra were used to develop a predictive model for lithium content according to the whole rock estimates. Principal component analysis (PCA) and biplot analysis were used to identify correlations between elemental composition and lithium content; multiple linear regression was employed to establish a regression model for lithium quantification. Residual analysis and various statistical metrics, including R², root mean squared error, and mean absolute percentage error, were used to assess the accuracy and performance of the models. In summary, this study highlights that for high lithium grade samples, ED-XRF can be preferred over LIBS as it requires less time for both sample preparation and analysis. However, this lithium assay is indirect, whereas for low lithium grade samples (<0.8 wt% Li₂O), LIBS should be preferred to ensure a robust lithium assay method.