Sales of a product plays a vital role, towards profitability of any organisation. It helps in building trust and allegiance between customer and business. Sales department in any organisation help us to build a strategic decision for growing their business. In the process of building a strong relationship, there is chance of performing fraud, termed as sales fraud. Detecting such fraud can be of great help to the organization and taking appropriate steps to prevent the same in future. In this paper, we have implemented various clustering techniques like, k-means, k-modes, Hierarchical clustering, partitioning around medoids and also the Self organizing map technique in order to efficiently analyse and thus detect the sales fraud. The finding in our research, states that the self organizing maps and Hierarchical clustering provides good classification accuracy. These algorithms are strongly recommended in the field of fraud detection.