The feedback score is roughly the number of users with whom they have had successful transactions. Let's assume two users have only had one transaction between them. Positive feedback adds one point, as you would expect. A user does not receive credit for a neutral transaction, so the score does not change. Negative feedback subtracts one point, taking away credit for a successful transaction. If multiple transactions take place between two users, the feedback is averaged so that one user always has exactly one point of influence over another. For example, if a user receives one negative and one positive review from the same person, they will average out to a neutral review and have no effect on the feedback score. Feedback percentage is a different matter: it's much less forgiving. It's designed to represent, roughly, a satisfaction rate. To achieve 100%, you would need all positive reviews. Neutral reviews decrease your percentage; negative reviews decrease it even more. Again, one person has a fixed amount of influence over another, with multiple reviews for the same person by the same person being averaged. This system is somewhat comparable to those seen on other sites, but we've made adjustments to account for differences in our industry and community. For example, eBay displays the number of total transactions a user has performed alongside their percentage of positive feedback. We display a score instead of a total, and our percentages give a little more influence to negative reviews. Most notably, users can't abuse the system by leave a large number of reviews for one user: such activity will have little to no impact on the score and percentage. As an unintended but potentially positive side effect, users are rewarded for their ability to conduct business with a wide range of customers, rather than just a handful of close friends.