This research has proposed two approaches that are the efficient classification approach and approach based on threshold value for improving the design of sites. The classification algorithm uses LCS to categorize the browsing patterns and based on the classified patterns website improvement algorithm is proposed. Factors being used to judge the quality of the prediction are accuracy and coverage. The second algorithm proposes a mathematical model based on threshold value to improve design of sites. It states that if the Average click ratio is greater that threshold (eg. threshold=2) it means web personalization or web transformation is required i.e. either improve the linkage on web page or reshuffle the web pages. AMPL that is a mathematical programming language using CPLEX and MINOS 5.5 solver being used to generate the code for the proposed approach. And calculated the necessary clicks needed to attain the target page.