Aiming at the issues of long data analysis cycle and slow market response in the analysis and application of marketing big data, this article applies predictive analysis, data mining and other methods based on cloud-based back propagation neural network (BPNN), so as to solve the problems in the traditional marketing big data analysis, aiming to improve the effectiveness and implementation efficiency of marketing strategies. In this article, it is found that optimization of marketing strategies and precision marketing can be achieved by combining the learning capabilities of BPNN with the efficient processing capabilities of cloud algorithms. The resource consumption of the cloud-based BPNN used in the study is 10% higher than that of the autoencoder model; the accuracy of the cloud-based BPNN is 25% higher than that of the autoencoder model; the resource utilization of cloud-based BPNN is 15% higher than that of autoencoder models. Therefore, the cloud-based BPNN has the advantages of low resource consumption and high accuracy, which can be more effectively applied in marketing big data analysis and decision-making.