Wood properties arise from a complex hierarchical organisation of lignocellulosic material and porosity spanning several length scales. This complexity is responsible for the significant variation in wood properties as it enables wood to adapt to a wide range of growth conditions by varying its microstructure. Understanding the linkage between microstructure and mechanical properties across different length scales is essential for improved material and product development, ranging from novel wood-based bio-composites and targeted wood modifications to tailored optimisation of glulam (GLT) and cross-laminated timber (CLT). In this work, a multiscale continuum micromechanics framework for stiffness homogenisation across different length scales, tailored to softwood, is developed. Whereas wood is modelled as a series of linked representative volume elements across distinct observation scales. While analytical micromechanics is computationally efficient, it cannot capture certain geometrical complexities of wood microstructure, particularly at the wood cell scale. Thus, we augment this analytical approach by incorporating a unit cell finite-element model at the porous cell wall scale, thereby enabling a more comprehensive consideration of morphological factors. Though at an increased computational cost. The predictions of the combined analytical–numerical model are validated against tensile and shear stiffness data reported in the literature, and the model is then used to generate a large parametric dataset for training a feedforward neural network surrogate. This surrogate is significantly faster than the combined analytical–numerical model, thereby enabling efficient large-scale sensitivity analyses. This sensitivity analysis reveals the relative importance of selected morphological parameters on the clearwood engineering constants, quantified by the relative Coefficient of Prognosis (COP), which measures the predictive relevance of each parameter (Figure 1). Ultimately, this surrogate framework opens new possibilities for integrated micromechanics and board-scale FE modelling.