The question of why individuals choose to explore or exploit as their routines accumulate remains largely unexplored in the organizational literature in a strict causal sense. To bridge this gap, we conducted an experimental laboratory study of individual decision-making sequences using a real-effort task and a two-phase procedure involving a training phase and an active phase. The former involved a learning process in which participants used their own skills to solve the same task across eight rounds to simulate the development of individual-level routines. In the active phase, we observed sequential choices to exploit an established routine, explore a new routine, or exploit a new routine. Participants were financially incentivized to abandon an established routine by providing higher performance-related payoffs for exploring unknown task environments. Our findings show that initial conditions related to task environment complexity matter; we found that when individuals are initially exposed to simpler tasks, they were more likely to continue exploiting an established routine. When initially exposed to more complex tasks, they were more likely to explore new and more profitable tasks, and then to continue exploiting the new routines they learned. Interestingly, we find not only that different kinds of performance feedback lead the exploration-exploitation choice, but our results also show how the feedback-choice linkage is contingent upon the task environment, i.e. the context (simple or complex) in which initial routines are formed. These findings contribute to the literature on individual search by demonstrating the important role of routinization and initial conditions in exploration and exploitation behavior.