In the current period, patterns of buying by individuals show increasing difficulty and separation, which makes traditional approaches to marketing that use broad methods less effective and not possible to maintain. Marketing that targets specific groups, which finds individuals with high potential and improves the use of resources through approaches using data, shows development as a main strategy for organizations to increase competitive position. The traditional RFM model shows wide use for dividing customers into groups, but the static features of this model have limitations in capturing changes over time in customer patterns, which limits the degree of accuracy in predicting responses of individuals. To provide a solution for this issue, this study presents an improved RFMTC framework for predicting response that includes measures of time span and the rate of leaving into the main dimensions, which allows modeling of customer states that is both following changes over time and more detailed. To further improve the performance of the model, an improved approach using methods of optimization that change to conditions is used to optimize important parameters of the model across all possibilities, which addresses the limitation of traditional methods that tend to result in solutions that are not the best possible. This study provides support in theory and direction in practice for organizations to develop strategies for marketing using data, and it shows significant importance in practice and value in application.