We analyze a model in which an uninformed decision maker (DM) makes a series of decisions after receiving unverifiable information from an expert before each decision. The novelty in our model is that DM and the experts might have a conflict of interest and whether this is the case is private information of DM. We characterize conditions under which “reputational information gathering” occurs, i.e., DM makes a series of decisions that are not optimal in the beginning in order to receive truthful information from the experts in the future. We show that this happens only if the potential bias is large, prior expected bias is small, and the relative importance of the future decisions is big enough. We also analyze the implications of our results for optimal organizational design. We show that if the initial expected bias is small enough, then DM prefers a centralized decision making regime as opposed to delegating to the experts, whereas if the initial expected bias is large, she prefers centralization only when the potential bias is large enough. We also show that if the potential bias is large but the initial expected bias is small enough, then the best thing for DM is to order the decisions from the most important to the least. This implies that if DM could choose the order of decisions herself, then reputational information gathering would never occur.