Disclosed are computer systems and methods for contextually targeted advertising using a regular periodicity of information derived from a user's messaging behavior. The disclosed systems and methods enable the prediction of future purchases based on a periodicity analysis of a user's purchase history, whereby advertisements can be targeted to the user based on the user's determined purchase habits. The disclosed systems and methods analyze a user's inbox by mining for purchase receipt messages, and determine a frequency of purchases associated with such receipts. Based on the determined frequency, the disclosed systems and methods can predict when subsequent like or similar purchases are to occur, whereby relevant advertisements or coupons may be served to the user in advance of the projected purchases.