Recent research has identified a positivity effect in consumers’ evaluations of agents, such as friends and professional critics, who provide word-of-mouth evaluations and recommendations. Specifically, agreement with an agent on previously loved alternatives is perceived as more diagnostic of the agent’s suitability than agreement on previously hated alternatives. This article argues that the positivity effect arises from greater ambiguity about attribute ratings of hated versus loved alternatives. Three studies support this by showing that the effect is moderated by the number of attributes, the number of alternatives, and the revelation of an agent’s attribute ratings, and is mediated by attribute ambiguity.