Abstract: Methods, computer systems, and computer-storage medium for generating an advertisement-targeting profile based on learned user preferences are provided. An indication that a user is interacting with marketing messages on one or more Web pages is received. Each marketing message comprises a plurality of attributes. A set of attributes that is shared across the marketing messages is determined; the set of attributes comprises, in part, a subset of incentives. An advertisement-targeting profile comprising user preferences related to the set of attributes is generated. The advertisement-targeting profile is used, in part, to generate predefined marketing-message elements that may be used to modify marketing messages to increase user responsiveness to the messages.

Abstract: Methods, systems, and computer-readable media for generating a real-time advertisement are provided. An advertisement candidate that accepts one or more items of real-time data is generated, having one or more advertiser extensions associated with the advertisement candidate. The advertisement extensions accept items of real-time data that are retrieved from an advertiser data provider, according to a mechanism for retrieval of real-time data items. A real-time advertisement is generated, having at least one of the retrieved items of real-time data and the advertisement candidate. Upon receiving a search request from a user, a real-time advertisement may be presented. The real-time advertisement may include items of real-time information incorporated into the advertisement candidate, as well as other advertisement content responsive to parameters of the advertisement API call.