Recent years have shown a large increase in the usage of content creation platforms—blogs, community QA sites, forums, etc.—aimed at the general public. User generated data contains emotional, opinionated, sentimental, and personal posts. This characteristic makes it an interesting data source for exploring new types of linguistic analysis, as is demonstrated by research on, e.g., sentiment analysis [4], opinion retrieval [3], and mood detection [1]. We introduce the task of experience mining. Here, the goal is to gain insights into criteria that people formulate to judge or rate a product or its usage. These criteria can be formulated as the expectations that people have of the product in advance (i.e., the reasons to buy), but can also be expressed as reports of experiences while using the product and comparisons with other products. We focus on the latter: reports of experiences with products. In this paper, we define the task, describe guidelines for manual annotation and analyze linguistic features that can be used in an automatic experience mining system.