We present a cognitive model that demonstrates how the recall accuracy of letter positions on a cell phone keypad effects the transition from novice to expert behavior. In particular, we target multi-tap text entry methods and focus on the process of visually searching versus selecting (i.e. deciding) a letter on the keypad. The model predicts the probability of letter location recall by novice users through a cognitive architecture named ACT-R, and models learning as the user gradually gains cognitive expertise with practice, session after session. We then employ this probability within a model of strategy adaptation that encapsulates the effect of different visual exploration strategies: novice users search for a letter while the behavior of advanced users is modeled by the Hick-Hyman Law. The final output of our cognitive model is the entry speed for the key press for a letter in a letter-group containing multiple distinct letters.