Thermogravimetric analysis (TGA) is a fast, reagent-free way to evaluate the biochemical stabilisation of organic matter during composting, yet few studies combine a factorial design with multivariate statistics to separate the effects of substrate, initial carbon-to-nitrogen (C/N) ratio and heating rate. We analysed 160 thermogravimetric curves (29,567 mass-temperature points, 25–900 °C) from a balanced design of substrate (poultry litter, cattle manure), C/N (25, 35), heating rate (10, 200 °C/min), composting week (0–12) and replicate. Non-overlapping mass-loss zones and thermal stability indices were computed at nine diagnostic temperatures and analysed with four-way factorial ANOVA, Pearson correlation, principal component analysis (PCA) and k-means clustering. Composting week was the dominant source of variance across all thermal descriptors (η² between 0.70 and 0.75, p < 10⁻³²), with a pronounced transition between weeks 10 and 11. Substrate showed its strongest main effect on the recalcitrant fraction (η² = 0.069, p = 2.8×10⁻¹³), and heating rate significantly shifted the labile fraction (η² = 0.056, p = 8.8×10⁻⁹), indicating that substrate and initial stoichiometry modulate the pace, not the endpoint, of stabilisation. PCA condensed twenty variables into two components (93.6% of variance), and k-means clustering identified three thermal states, early, intermediate and mature (silhouette = 0.616). These results show that thermogravimetric zoning offers a fast, reproducible fingerprint of compost maturity, governed mainly by process time but fine-tuned by substrate chemistry and heating rate.