Speed zone thresholds in women's soccer are often based on thresholds derived from the men's game. This study applied a data-mining approach to generate (a) team-based speed zone thresholds, and (b) individualised speed zone thresholds for elite female soccer players. Activity data from 47 elite club-level female soccer players was collected using multi-constellation global navigation satellite systems technology during 76 competitive matches between 2020 and 2021. The elbow method was used to identify the number of thresholds in the dataset, and spectral clustering was applied to each player's instantaneous match-play speed data to determine the value of these thresholds. Team-based categories were then formed by calculating the mean value of each threshold from the individual players. Three speed zone thresholds were identified (1.12, 2.83, and 5.10 m·s-1). The newly-generated speed zone thresholds resulted in greater distance covered in the top three speed zones (all p < 0.05) compared to existing speed zone thresholds for women's and men's soccer. The between-player coefficient of variation for individualised speed zone thresholds was 13.0% for the lowest threshold, 12.9% for the middle threshold, and 8.5% for the highest threshold. Average maximum match-play running speed was a significant predictor of all individualised speed zone thresholds (lowest threshold: β = 0.13, p = 0.016; middle threshold: β = 0.44, p = 0.001; highest threshold: β = 0.74, p < 0.001). This study demonstrates a data-mining approach to establish team-based and individualised speed zone thresholds for women's soccer that could be applied within practical and academic settings.