The invention discloses an APP user behavior analysis method. The method comprises the following steps of: obtaining a user click behavior log and a corresponding history transaction record of an APPin a preset time period; dividing users into three types: new users, active users and old users; obtaining user behavior data corresponding to each type of users according to the user click behavior log and a preset burying point; extracting a user behavior path corresponding to each type of users and then determining a key transition point of changing users from new users to active users and changing the users from active users to old users; and displaying the key transition point. By adoption of the method, pointed intervene can be subsequently carried out on the users according to the key transition point, so that the conversion rate of changing users from new users to active users and changing the users from active users to old users is improved. Moreover, the invention discloses an APP user behavior analysis device.