Abstract: Requests for content are received from a user. For example, an HTTP request for a web page sent by a computing device of the user is received by a server. Each request has a landing context, which may include information about the user sending the request, the request itself, the referrer, or another subject. This information can be found in the requests or accessed from sources external to the requests. The information in the landing context is used to redirect the request to a particular web page or other piece of content. Information about a success or repel resulting from the redirection may be collected and used to inform later redirections.

Abstract: One embodiment of the present invention sets forth a technique for providing a variation of digital content that is based on a client-specified context token. Localized digital assets, e.g., graphics resources, video resource, audio resources, and text string resources, are stored as a collection of records and are dynamically linked to a assemble a software system based on the client-specified context token to generate the variation of digital content. The client receives the assembled variation of the digital content including the localized digital assets specified by the client-provided context.

Abstract: One embodiment of the present invention sets forth a technique for providing a variation of digital content that is based on a client-specified context token. Localized digital assets, e.g., graphics resources, video resource, audio resources, and text string resources, are stored as a collection of records and are dynamically linked to a assemble a software system based on the client-specified context token to generate the variation of digital content. The client receives the assembled variation of the digital content including the localized digital assets specified by the client-provided context.

Abstract: One embodiment of the present invention sets forth a technique for providing a variation of digital content that is based on a client-specified context token. Localized digital assets, e.g., graphics resources, video resource, audio resources, and text string resources, are stored as a collection of records and are dynamically linked to a assemble a software system based on the client-specified context token to generate the variation of digital content. The client receives the assembled variation of the digital content including the localized digital assets specified by the client-provided context.

Abstract: A facility for bidding on a pending advertising opportunity is described. The facility receives at an advertiser a notification from a publisher. The notification contains keywords relating to a content request received by the publisher, and is sent by the publisher between its receipt of the content request and its dispatch of a content response responsive to the content request. In response to receiving the notification, the facility sends to the publisher a dynamic bid. The dynamic bid specifies an advertising message and a bid amount that the advertiser is willing to pay to present the advertising message in connection with a content response responsive to the content request produced by the publisher.

Abstract: One embodiment of the present invention sets forth a technique for providing a variation of digital content that is based on a client-specified context token. Localized digital assets, e.g., graphics resources, video resource, audio resources, and text string resources, are stored as a collection of records and are dynamically linked to a assemble a software system based on the client-specified context token to generate the variation of digital content. The client receives the assembled variation of the digital content including the localized digital assets specified by the client-provided context.

Abstract: Requests for content are received, for example, an HTTP request for a web page is received. Each request has a landing context, which may include information about the user sending the request, the request itself, the referrer, or another subject. This information can be found in the requests or external to the requests. The information of the landing context is compared to appropriate triggering criteria. For requests that satisfy the appropriate triggering criteria, a destination is provided based on one or more weighted treatments. The request is directed to the provided destination. The results of the requests are subsequently analyzed and used to adjust the treatments in an automated fashion.

Abstract: An external traffic advertisement system is provided that generates advertisement sets based on analysis of visits to a web site that were referred by an external source. The advertisement system aggregates the referral information for each referral type. A referral type may be defined by one or more of keyword text derived from the query text of the referrals, landing page type, external source, product identifier, and so on. The advertisement system may, for each referral type, aggregate the total revenue from the visits of that referral type and may generate a count of the number of converting visits for that referral type. The advertisement system then identifies those referral types whose aggregated information satisfies an advertisement criterion and generates an advertisement set for each identified referral type with a keyword derived from keyword text and with a link based on the landing page type of the referral type.

Abstract: One embodiment of the present invention sets forth a technique for providing a variation of digital content that is based on a client-specified context token. Localized digital assets, e.g., graphics resources, video resource, audio resources, and text string resources, are stored as a collection of records and are dynamically linked to a assemble a software system based on the client-specified context token to generate the variation of digital content. The client receives the assembled variation of the digital content including the localized digital assets specified by the client-provided context.

Abstract: Requests for content are received, for example, an HTTP request for a web page is received. Each request has a landing context, which may include information about the user sending the request, the request itself, the referrer, or another subject. This information can be found in the requests or external to the requests. The information of the landing context is compared to appropriate triggering criteria. For requests that satisfy the appropriate triggering criteria, a destination is provided based on one or more weighted treatments. The request is directed to the provided destination. The results of the requests are subsequently analyzed and used to adjust the treatments in an automated fashion.

Abstract: A new release advertisement system generates advertisement sets for new release items that have a release date. To generate the advertisement sets, the system identifies items with release dates within a “new release advertising window.” The new release advertisement system then identifies item and keyword pairs from the text within the attributes of the items. After identifying item and keyword pairs for each of the new release items, the system removes pairs that do not satisfy a filtering criterion. The system then generates advertisements from the text of the attributes of the items remaining after the removing based on the filtering criterion. The system then creates advertisement sets for the remaining pairs that include the advertisement generated for the item of the pair and the keyword of the pair.

Abstract: A new release advertisement system generates advertisement sets for new release items that have a release date. To generate the advertisement sets, the system identifies items with release dates within a “new release advertising window.” The new release advertisement system then identifies item and keyword pairs from the text within the attributes of the items. After identifying item and keyword pairs for each of the new release items, the system removes pairs that do not satisfy a filtering criterion. The system then generates advertisements from the text of the attributes of the items remaining after the removing based on the filtering criterion. The system then creates advertisement sets for the remaining pairs that include the advertisement generated for the item of the pair and the keyword of the pair.

Abstract: Requests for content are received, for example, an HTTP request for a web page is received. Each request has a landing context, which may include information about the user sending the request, the request itself, the referrer, or another subject. This information can be found in the requests or external to the requests. The information of the landing context is compared to appropriate triggering criteria. For requests that satisfy the appropriate triggering criteria, a destination is provided based on one or more weighted treatments. The request is directed to the provided destination. The results of the requests are subsequently analyzed and used to adjust the treatments in an automated fashion.

Abstract: A method and system for augmenting keywords, submitted to advertisement placement services, that are not performing as expected is provided. An advertisement system analyzes the performance of keywords of advertisement sets submitted with advertisement placement services, such as search engine services. Based on the analysis, the advertisement system identifies keywords that are candidate keywords for augmentation. The advertisement system then identifies augmentation terms for the candidate keywords. The advertisement system then augments the candidate keyword of an advertisement set with the augmentation term and submits that augmented advertisement set with the augmented keyword to an advertisement placement service.

Abstract: A calculate bid system calculates bid amounts based on category-specific conversion information and category-specific advertising expense factors. The calculate bid system collects conversion information for keywords used in advertisements for categories of items purchased during converting sessions. The calculate bid system uses collected category-specific conversion information along with category-specific advertising expense factors to calculate a bid amount for a keyword. The calculate bid system may calculate a bid amount by generating a product for each category of the financial benefit for the category multiplied by the advertising expense factor for that category. The calculate bid system may then aggregate the products and normalize the aggregated product by dividing by a collected number of conversions. The calculate bid system may then multiply the normalized aggregated product by a forecast conversion rate to give the bid amount.

Abstract: A facility for bidding on a pending advertising opportunity is described. The facility receives at an advertiser a notification from a publisher. The notification contains keywords relating to a content request received by the publisher, and is sent by the publisher between its receipt of the content request and its dispatch of a content response responsive to the content request. In response to receiving the notification, the facility sends to the publisher a dynamic bid. The dynamic bid specifies an advertising message and a bid amount that the advertiser is willing to pay to present the advertising message in connection with a content response responsive to the content request produced by the publisher.

Abstract: One embodiment of the present invention sets forth a technique for providing a variation of digital content that is based on a client-specified context token. Localized digital assets, e.g., graphics resources, video resource, audio resources, and text string resources, are stored as a collection of records and are dynamically linked to a assemble a software system based on the client-specified context token to generate the variation of digital content. The client receives the assembled variation of the digital content including the localized digital assets specified by the client-provided context.

Abstract: A user of a computing device may see an item of interest she would like to remember for future reference. The user captures data of the item of interest and submits it to a memory enhancement service for enhancement and storage. The service submits the captured data to a human interaction task system which distributes the captured data to one or more human workers which to identify the item of interest, determine the user's interest in the item, and provide information regarding the item based on this determined interest. To facilitate the enhancement process, the user may add indications to the captured data prior to submission. Alternatively or additionally, the service may electronically submit queries to the user. The enhanced data returned from the human interaction task system is then stored by the memory enhancement service for subsequent recall by the user and possible use by the user.