Considering its high specific modulus, nanocellulose, including rigid cellulose nanocrystals (CNCs) and semiflexible cellulose nanofibrils (CNFs), is widely used as a nano-reinforcing filler for polymeric-based composites, which is regarded as the most promising application of these biomass nanoparticles. The quantitative evaluation of the compatibility and dispersion/aggregation state of nanocellulose in polymeric matrices is a critical issue, as it conditions the efficient stress transfer from the matrix to the filler and effective mechanical reinforcement effect. This study reports a comprehensive set of theories and methods to directly evaluate the compatibility and dispersibility of CNCs and CNFs in four polymer matrices with different polarities, where the compatibility was assessing by Hansen solubility and dispersibility by Raman mapping and cluster analysis. Triple-bond modification on the surface of nanocellulose is a promising approach for accurate recognition in composites, exhibiting the individual signal located in the Raman-silent regions of various polymeric matrices. Based on the discussion of the quantitative dispersion factor, a multiscale percolation model is proposed to better predict the mechanical properties of nanocellulose-reinforced composites based on Raman mapping results, in order to update traditional percolation models.