Numerous online communities and e-commerce sites provide users with crowd-based recommendations to influence decision making about products. Similarly, automated recommender systems often use social advice or curated knowledge provided by experts to give customers personalized product recommendations. Little, however, is known about the relative strengths of these approaches in repeated-decision scenarios. We used social comparison and an expert recommendation to examine the relative effectiveness of these methods of persuasion for users making repeated retirement saving decisions. We exposed 314 performance-incentivized experiment participants to a retirement saving simulator where they made 34 yearly asset allocation decisions in one of three user interface conditions. The gap between participants’ retirement goal and actual savings was smallest in the expert advice condition and significantly better than the social comparison condition. Both conditions were significantly better than the control condition. In non-control conditions, users adjusted their behavior and achieved their saving goal more effectively.