In order to solve that the traditional structure decomposition method cannot consider the simplicity of formula and the uniformity of factor changes at the same time, in this study the structure decomposition method was improved and applied to the research on the water driving effect of Taizhou City, the representative city of water saving. The research results show that the scale of industrial water consumption in the region has dropped by 9. 14% in the past five years. Technical effects are the main contributor to Taizhou’s water-saving benefits. The scale effect is the main driving factor leading to the growth of industrial water consumption in Taizhou. Contrary to technical effects, structural effects that have the inhibitory effect of the growth of industrial water consumption in various regions are showing a steadily weakening trend, but there is still great potential for structural effects in economically underdeveloped regions. Based on the GDIM decomposition results, the two structural decomposition methods can draw similar conclusions as a whole, while the traditional structural decomposition method will produce excessively high evaluations of contributing factors for some areas where water use factors change drastically. The improved structural decomposition method is more accurate both in terms of calculation results and in economic sense, and is suitable for water-saving construction periods when water usage fluctuates volatility and economic influencing factors are complicated.