Abstract: Generating a user timeline for an electronic device (ED) userData is collected that includes: location data; application data; and activity data. Occurrences of predetermined types of observed events are detected based on the collected data. For each detected occurrence a respective observed event record is stored that includes information about a time and type of the observed event. Planned event records, each including information about a time and type of a respective planned event, are stored for planned events that the user is scheduled to participate in, the planned event records. Events are predicted based on the observed event records and the planned event records. Information about observed, planned and predicted events are output on a timeline user interface based on observed event records, planned event records and predicted event records stored for the predicted events, respectively.

Abstract: Systems and methods are described for providing user offers based on efficient iterative recommendation structures. In various aspects, a server invokes a bi-directional look-up interface via a lookup request, where the bi-directional look-up interface is exposed via an electronic recommendation structure. The lookup request causes the bi-directional look-up interface to return a bi-directional recommendation value. The bi-directional recommendation value indicates a likelihood of a first user selecting a first offer or a second offer. The bi-directional recommendation value is transmitted via a computer network to a client device associated with the first user upon a determination that the likelihood meets or exceeds a recommendation threshold. The client device is operative to display at least one of the first offer or the second offer.

Abstract: Systems, methods, and computer-readable media are disclosed for augmenting real-time bidding data with proprietary data. One method includes: receiving, at a server over an electronic communications network from a real-time impression bidder, a bid request or a request to augment a bid request with proprietary data; accessing, by the server from an internal database, proprietary data of a data augmenting service based on a user identifier of the bid request; determining, by the server, proprietary data to include in an augmented bid request based on at least one of the received bid request and the user identifier; formatting, by the server, the augmented bid request into a standardized, augmented bid request; and transmitting, by the server over the electronic communications network, the standardized, augmented bid request to the real-time impression bidder.

Abstract: The present disclosure relates to the field of emotion recognition technologies, and specifically, to a visual perception-based emotion recognition method. The method in the present disclosure includes the following steps: step 1: acquiring a face video by using a camera, performing face detection, and locating a region of interest (ROI); step 2: extracting color channel information of the ROI for factorization and dimension reduction; step 3: denoising and stabilizing signals and extracting physiological parameters including heart rate and breathing waveforms, and heart rate variability (HRV); step 4: extracting pre-training nonlinear features of the physiological parameters and performing modeling and regression of a twin network-based support vector machine (SVM) model; and step 5: reading a valence-arousal-dominance (VAD) model by using regressed three-dimensional information to obtain a specific emotional representation.

Abstract: A method includes receiving a first image; extracting a first set of features from the first image; executing a first machine learning model using the extracted first set of features as input to generate a first image performance score for the first image; iteratively executing, using the first set of features as input for each execution, a generative machine learning model to generate a plurality of generated images; extracting a second set of features from each of the plurality of generated images; iteratively executing the first machine learning model using a different second set of features as input for each execution to generate a generated image performance score for each of the plurality of generated images; and transmitting a defined number of the plurality of generated images to a computing device, the defined number of generated images determined based on the generated image performance scores.

Abstract: A system and method are presented that integrate date-related data from a variety of disparately defined data sources into a single user interface. Date-related data is defined in tactics that are associated with types. The types are associated with additional field definitions that allow for a single tactic data entity to be created for a variety of tactic types, with different tactic types being associated with different sets of additional fields. Cues are also used to automate the generation of date-related tactics. Cues can be triggered by the creation of trigger tactics of a predetermined type, or through an event in a subscribed database. Individual tactics record estimated and actual expenses for budgeting purposes. Finally, plan hierarchies allow for the integrated viewing of date-based tactics defined at different levels of the hierarchy.

Abstract: Testing of a computer-implemented dynamic content generator module operable within a computer networking environment is disclosed. According to an example, the dynamic content generator module executed by a computing system dynamically generates a content item for a test group based on a combination of test values of the test group as input to a template-rule framework. The dynamic content generator module outputs the content item for each test group. A testing module executed by the computing system associates the content item with the test values of the test group from which that content item was generated to obtain associated test result data, and outputs the associated test result data that includes the content item and associated test values.

Abstract: A method implements efficient counterfactual search. The method includes receiving a request corresponding to an input vector, processing the input vector with a model to generate an output vector that does not correspond to a selected class, and processing the input vector using a component, of a plurality of components, to generate a counterfactual vector to the selected class. The plurality of components includes a number of dimensions that is less than a number of features of the input vector. The method further includes processing the counterfactual vector to generate a recommendation and presenting the recommendation.

Abstract: The present disclosure relates to systems and methods for tracking content across multiple content outlets. Content items, characterized by first metadata, and content slots, characterized by second metadata, can be received. Third metadata for digital content slots can be generated. Fourth metadata can be received. A recommended content item package can be provided based on a comparison between a combination of the first metadata, the second metadata, and the third metadata and the fourth metadata. The recommended content item package can include a content file and a metadata file. A performance and a measurement of the recommended content item package can be determined by executing the metadata file to track the content file across one or more of the content slots and the potential digital content slots. The performance and the measurement can be provided.

Abstract: Aspects of the subject disclosure may include, for example, collecting television viewership information for viewers of television content items of a television advertising campaign, collecting digital viewership information for online activity by users, joining the television viewership information and the digital viewership information, wherein the joining comprises associating household identifiers of the television viewership information with device identifiers of the digital viewership information, building a content ranking of digital content items closest to the television content items and initiating, by the processing system, a digital advertising campaign with one or more digital content items of the content ranking of digital content items attached to the digital advertising campaign. Other embodiments are disclosed.

Abstract: The present teaching relates to exploring a personal interest space based on a decision tree. In one example, information related to a personal interest space associated with a user is obtained. The personal interest space includes at least one channel of content that the user is interested in. A decision tree created based on information related to a plurality of users with respect to a plurality of channels of content is obtained. Each node of the decision tree is associated with at least one of the plurality of channels of content. A first node of the decision tree is identified based on the information related to the personal interest space associated with the user. A question for the user in connection with the first node of the decision tree is determined. An answer to the question is received from the user. A second node of the decision tree is identified based on the answer.

Abstract: Example methods, apparatus, systems and articles of manufacture to implement under-addressable advertisement measurement are disclosed. Example apparatus disclosed herein determine a tuned under-addressable advertisement duration and a total tuned advertisement duration for a first television broadcast at a first monitored site, the tuned under-addressable advertisement duration and the total tuned advertisement duration determined based on (i) return-path tuning data associated with the first monitored site, (ii) linear advertisement log data associated with a broadcast area including the first monitored site, and (iii) addressable advertisement log data associated with the first monitored site.

Abstract: Systems, methods and media for adaptive real time modeling and scoring are provided. In one example, a system for automatically generating predictive scoring models comprises a trigger component to determine, based on a threshold or trigger, such as a detection of new significant relationships, whether a predictive scoring model is ready for a refresh or regeneration. An automated modeling sufficiency checker receives and transforms user-selectable system input data. The user-selectable system input data may comprise at least one of email, display or social media traffic. An adaptive modeling engine operably connected to the trigger component and modeling sufficiency checker is configured to monitor and identify a change in the input data and, based on an identified change in the input data, automatically refresh or regenerate the scoring model for calculating new lead scores. A refreshed or regenerated predictive scoring model is output.

Abstract: Systems and methods for a computer-based process that determines when a viewer is likely to skip over supplemental content and adjusts supplemental content presentation to compensate. Systems of embodiments of the disclosure may utilize various inputs to determine the likelihood of skipping supplemental content, including cursor position at or near specified icons or other UI elements, as well as user actions such as gaze direction, various motions or actions, controller manipulations, and the like. Once a likelihood of skipping supplemental content is determined, various actions may be taken in response, including without limitation selection of supplemental content that conveys its intended message prior to skipping, supplemental content that can be played at increased speed, and designation of supplemental content slots as skippable or non-skippable.

Abstract: Methods of preparing retail establishments are disclosed. An area for a retail location may be chosen and an individual may be selected to manage the retail location. The selection of the area may include an evaluation of geographical factors. From the area, a retail location may be selected based on an aggregate score of a set of characteristics of the retail location. The individual may remotely select the layout and furnishing of the retail location and may remotely monitor the progress of the preparation of the retail location. In an embodiment, the individual may be prepared to manage the retail location at a centralized location that is distinct from the area where the retail location is located.

Abstract: Methods and systems herein may help to provide an advertisement marketplace for advertisements that are valued according to gaze data from wearable computing devices.

Abstract: Systems and methods for a computer-based process that determines when a viewer is likely to skip over supplemental content and adjusts supplemental content presentation to compensate. Systems of embodiments of the disclosure may utilize various inputs to determine the likelihood of skipping supplemental content, including cursor position at or near specified icons or other UI elements, as well as user actions such as gaze direction, various motions or actions, controller manipulations, and the like. Once a likelihood of skipping supplemental content is determined, various actions may be taken in response, including without limitation selection of supplemental content that conveys its intended message prior to skipping, supplemental content that can be played at increased speed, and designation of supplemental content slots as skippable or non-skippable.

Abstract: In accordance with some embodiments of the disclosed subject matter, methods, systems, and media for setting and using an advertisement frequency cap based on causal conversions or impact of advertisements are provided.

Abstract: Briefly, example methods, apparatuses, and/or articles of manufacture are disclosed that may be implemented, in whole or in part, using one or more computing devices to facilitate and/or support one or more operations and/or techniques for machine learning (ML) classification of digital content for mobile communication devices, such as implemented in connection with one or more computing and/or communication networks and/or protocols.

Abstract: A method for pushing information, and an electronic device are disclosed. The method includes: obtaining a plurality of initial pages containing document content; predicting, based on the document content in the plurality of initial pages, an effective rate for pushing information using each initial page; selecting, based on the effective rate, at least one candidate page from the plurality of initial pages and extracting a first keyword from the document content of the at least one candidate page; and determining a target page for mounting information to be pushed from the at least one candidate page, based on a matching degree between a second keyword of the information to be pushed and the first keyword.

Abstract: A system for maintaining product networks to provide recommendations for bid amounts to third party entity devices. A service provider entity device tracks product purchase history for customers and determines relationships associated with products purchased by the customers. When a webpage is loaded by a customer device, a third party entity device receives a bid request from a publisher device and transmits a bid recommendation request to the service provider entity device. The service provider entity device determines input features including a base bid, a propensity score, and a pacing score based on the relationships. The service provider entity device transmits a bid recommendation response with a recommended bid amount to the third party entity device, based on the input features.

Abstract: The present disclosure relates to systems and methods for automatic attribution modeling and measurement. In aspects, a system may receive identification information associated with profiles and electronic devices that were exposed to a certain piece of targeted content. The demographic and device data associated with the individuals who were exposed to the targeted content are used to create a control group of individuals who were not exposed to that targeted content. The real-world visit rates of the exposed group and the control group to one or more locations may be monitored over a period of time (or campaign) and evaluated to assess the effectiveness of the targeted content.

Abstract: A method includes: receiving a query from a computing device accessed by a recipient; accessing an identifier, associated with an advertisement card, including a pointer to an electronic document associated with the advertisement card and a recipient code associated with the recipient; accessing a set of recipient profiles associated with target recipients of a set of advertisement cards.

Abstract: The present disclosure relates to systems, methods, and computer-readable media for selectively identifying lodging content from listings provided by different listing providers and generating a composite listing that includes content selected across the different providers. For example, systems described herein may include content selection models that are trained to select different content types from listings that are provided by different listing providers based on features of the content and/or users. The composite listing may enable a user to book a stay with a lodging facility based on content presented via different third-party content providers for the same facility.

Abstract: The disclosed embodiments relate to a computer implemented recommender system/method which enables the computer to provide more “emotionally” relevant/connected, and therefore more likely to be successful, recommendations while minimizing dependency on historical data as well susceptibility to bias, e.g. monetary and/or sponsor based. The disclosed embodiments segment a target set of known users into groups based on similar characteristics and then associate each group with one or more unique sets of affinities derived from the social media interactions of a general population having similar affinities. Similarly, a unique set of affinity characteristics is derived, and stored in a data structure, for each of a set of recommendations. The unique sets of affinities associated with the groups are compared with the set of affinity characteristics associated with each recommendations to derive a subset thereof relevant to the users of the particular group.

Abstract: Embodiments of the present disclosure relate to data correlation of data pertaining to content distributed through distinct content delivery networks including offline networks and data related to user interaction with an online presence on the distributed computer network and uses for such correlated data, including to measure and quantify latent effects of the content distributed through a distinct content delivery network, such as an offline network, on user's interactions with the online presence on the distributed computer network.

Abstract: An influence system for predicting advertisement impact for advertising response selection. A treatment group of entities is selected which have received an advertising treatment of a campaign. A control group of entities is selected which excludes the treatment group entities is selected which have not received the advertising treatment of the campaign. An influence model is created for each campaign by comparing features of the treatment group converters to features of the control group converters. A campaign is selected for an opportunity to expose a specified entity to advertising based on the result of applying each respective campaign's influence model to features of the specified entity.

Abstract: The present disclosure provides for management of impressions in advertisement campaigns. Impressions may be moved between different impression media based on performance metrics and historical data. Impression budgets may be modified in an active campaign based on forecasts determined using current performance data. Impression budgets may be reallocated between simultaneously active advertisement campaigns.

Abstract: Devices for symbiotic display of ads on mobile devices are provided. A mobile device has a display screen that includes an ad display for displaying ad content and an operating system display for operating system and applications. Ads are constantly displayed on the ad display independent to the content displayed on the operating system display. The display screen may have an aspect ratio of 18:9 with a 2:9 ad display and a 16:9 operating system display. The ad display and operating system display may be separate touch screens that are contiguous or a single touch screen that is divided into two or more distinct displays as a function of an operating system or application. The ad display may be in an auxiliary display device connected to the mobile device. A system and method for applying symbiotic targeted advertising is also provided.

Abstract: Methods, systems, and media for managing online advertising campaigns based on causal conversion metrics are provided. In some embodiments, the method comprises: receiving conversion information corresponding to test group including consumers that were presented with an advertisement using an advertising channel; receiving advertisement viewability information indicative of a probability that each of the consumers viewed the advertisement; determining that a subset of the consumers did not view the advertisement based on the probability; placing the consumers into a control group and a test group based on the probability corresponding to each of the consumers; calculating a causal conversion metric based on a comparison of the conversion information corresponding to consumers of the control group and conversion information corresponding to consumers of the test group; and determining whether to place an advertisement using the advertising channel based on the causal conversion metric.

Abstract: A wireless communication device is disclosed. The wireless communication device has a near field communications transmitter system. The device has an outer chassis that can be sized, dimensioned and decorated to resemble common household items that lack any apparent value. The wireless communication device can also include an indicia that is invisible to an unaided human eye.

Abstract: In accordance with embodiments, there are provided mechanisms and methods for advertisement distribution using a multi-tenant on-demand database system. These mechanisms and methods for distributing advertisements using a multi-tenant on-demand database system can integrate an advertisement distribution service with the services provided by the multi-tenant on-demand database system. The ability to provide an advertisement distribution service with the services provided by the multi-tenant on-demand database system can enable the distribution of advertisements to take advantage of other data stored by and/or services provided by the multi-tenant on-demand database system.

Abstract: A method begins by monitoring exchange item transactions occurring in an exchange item marketplace network. The method continues with generating trend information based on the monitored exchange item transactions and determining attributes for a targeted marketing program based on the trend information. The method continues with identifying a subset of buyer computing devices based on the attributes. The method continues by generating offer information based on the targeted marketing program and sending the offer information to the subset of buyer computing devices. When receiving a purchase request from a buyer computing device, the method continues by generating an exchange item request regarding an exchange item and sending the exchange item request to a branded server associated with the exchange item. The method includes receiving offer exchange item information from the branded server and sending at least a portion of the offer exchange item information to the buyer computing device.

Abstract: A method for online advertising, is provided, including: receiving a request for a web page from a client device; in response to the request, selecting an advertisement for presentation on the web page, and transmitting the web page to the client device; determining a duration of exposure of the advertisement when the web page is presented on the client device; determining a level of performance of a guaranteed presentation amount associated with the advertisement, the level of performance based on the duration of exposure of the advertisement. In another embodiment, a plurality of candidate advertisements are identified for presentation on a requested web page; for each candidate, an expected cost per time unit (eCPTU) of exposure is determined; one candidate advertisement is selected for presentation on the web page based on the eCPTU's; the web page is transmitted to a client device.

Abstract: Example methods, apparatus, systems and articles of manufacture to implement under-addressable advertisement measurement are disclosed. Example apparatus disclosed herein determine a tuned under-addressable advertisement duration and a total tuned advertisement duration for a first television broadcast at a first monitored site, the tuned under-addressable advertisement duration and the total tuned advertisement duration determined based on (i) return-path tuning data associated with the first monitored site, (ii) linear advertisement log data associated with a broadcast area including the first monitored site, and (iii) addressable advertisement log data associated with the first monitored site.

Abstract: An apparatus, method, and computer program product are provided to adjust and modify input signals used in connection with predictive models by detecting events, such as changes in operating parameters of data objects and/or related systems and calculating adjusted decay rates to be applied to time-series data associated with times prior to an occurrence of an event. In some example implementations, an indication of an event associated with a given datastream is received, in a manner which indicates the change in an operating parameter and the time at which the change occurred. Based at least in part on the indication of the event associated with the datastream, a second decay rate associated with the set of time-series data is determined and applied to the set of time-series data, such that an updated future performance level can be calculated by a predictive model.

Abstract: In some embodiments, an intervention evaluation system estimates counterfactual metric for a focal online platform based on an assessment model built using performance data of the focal online platform and control online platforms. The intervention evaluation system accesses performance data of the focal online platform that has been subject to a temporary intervention and performance data of control online platforms that are not subject to the temporary intervention. The intervention evaluation system determines estimation weights for these control online platforms based on the performance data in a pre-intervention period. Based on the estimation weights, the intervention evaluation system computes a counterfactual metric indicating the performance of the focal online platform in a post-intervention period in the absence of the temporary intervention.

Abstract: A non-transitory computer-readable recording medium stores therein a prediction program that causes a computer to execute a process including obtaining a machine learning model trained by using training data, the machine learning model predicting presence or absence of purchase actions or predetermined actions of users corresponding to feature information on the users, the training data including feature information on users and information indicating the presence or the absence of the purchase actions of commercial products or the predetermined actions, receiving input of a budget amount in an entry field displayed on a display in association with a user group including users including a common feature in the feature information, and displaying, in association with the budget amount on the display, feature information on the users included in the user group corresponding to the entry field.

Abstract: Systems and methods for a computer-based process that determines when a viewer is likely to skip over supplemental content and adjusts supplemental content presentation to compensate. Systems of embodiments of the disclosure may utilize various inputs to determine the likelihood of skipping supplemental content, including cursor position at or near specified icons or other UI elements, as well as user actions such as gaze direction, various motions or actions, controller manipulations, and the like. Once a likelihood of skipping supplemental content is determined, various actions may be taken in response, including without limitation selection of supplemental content that conveys its intended message prior to skipping, supplemental content that can be played at increased speed, and designation of supplemental content slots as skippable or non-skippable.

Abstract: An automated advertising scheduling and distribution process reacts to the effectiveness of sales data. A hosted platform creates location-specific playlists based on key consumer variables that impact buying behavior, and dynamically performs data analytics. Utilizing a programmatic system and machine learning algorithmic methodology, the platform gathers data from the retailer's data warehouse and automatically pulls location-by-location sales data while simultaneously collecting playback data. If sales are not being affected on the particular item that is being promoted, then the platform may be configured to replace that message with a promotional message for another product with a higher likelihood of engagement and conversion. This virtual feedback loop ensures that the platform is optimizing the most effective series of promotional messages for any given location.

Abstract: A computer-based system and method to assign patients to providers. In one embodiment, the invention predicts the performance of provider-patient pairs in terms of different outcome-, financial- and satisfaction-related metrics across providers using advanced machine learning methodologies to develop distinct models for each of these metrics across each of the providers using a historical database. In another embodiment, patients are assigned to providers in a batch or online manner using this information through an optimization framework that looks to maximize or minimize arbitrary combinations of the outcome-, financial-, and satisfaction-related metrics subject to practical operational constraints.

Abstract: An analysis system may receive a request to determine a cause of underdelivery for an ad (“query content item”). Properties of the query content item can be processed through a matching model to determine properties of a bid request, thereby generating a bid request. The generated bid request also includes a request for the content server to return a report in response to the bid request regarding ad selection decisions. The bid request may be submitted to the content server, which causes the content server to define a set of candidate content items and select at least one of the candidate content items to submit bids on behalf of in response to the bid request. The content server also provides a report in regarding ad selection decisions. Thus, if the query content item is one of the ads that were not selected, the report may include factors contributing to the query content item not being selected.

Abstract: Techniques to measure consumption of content pages comprising a plurality of distinct content assets are disclosed. In various embodiments, content consumption signal data gathered by a plurality of clients, the content consumption signal data reflecting for at least a subset of content pages user engagement by content asset comprising the content page, is received. The received content consumption signal data and content attribute data associated with each respective content page are used to compute for each content page a content consumption metric reflecting an amount of content determined to have been consumed.

Abstract: Application personalization techniques and systems are described that leverage an embedded machine learning module to preserve a user's privacy while still supporting rich personalization with improved accuracy and efficiency of use of computational resources over conventional techniques and systems. The machine learning module, for instance, may be embedded as part of an application to execute within a context of the application to learn user preferences to train a model using machine learning. This model is then used within the context of execution of the application to personalize the application, such as control access to digital content, make recommendations, control which items of digital marketing content are exposed to a user via the application, and so on.

Abstract: In various embodiments described herein, a visualization system receives message requests from client devices, wherein the message requests comprise at least location data that identifies a location of a client device, and media content, wherein the media content includes at least one of image data, audio data, and video data. In response to receiving the message requests that comprise the media content, the visualization system analyzes and parses the media content to detect one or more tags embedded within the media content. In response to detecting the tag, the visualization system identifies a campaign or account referenced by or associated with the tag. Based on the identification of the campaign based on the tag within the media content, the visualization system determines a distribution of the campaign based on the location data from the message request.

Abstract: A computer-driven apparatus coupled to a network receives data from metadata sources and consumers' display devices. From the sources, the apparatus collects metadata concerning characteristics of a given item of host video and a proposed class of advertisements for web delivery proximate or embedded in the given item. The apparatus continually monitors actual delivery of the given item by receiving transmissions from consumers' display devices. The apparatus develops initial forecasting inputs based on historical data from similar host videos, and then adjusts the forecasting inputs based on the actual deliveries. These forecasting inputs are used to compute an interim supply of advertising opportunities associated with the given item, which is reduced by relevant factors to provide a net available supply forecast. The apparatus provides a human-readable display of information including the net available supply forecast.

Abstract: A method for finding a solution to a problem is provided. The method includes storing candidate individuals in a candidate pool and evolving the candidate individuals by performing steps including (i) testing each of the candidate individuals to obtain test results, (ii) assigning a performance measure to the tested candidate individuals, (iii) discarding candidate individuals from the candidate pool in dependence upon their assigned performance measure, and (iv) adding, to the candidate pool, a new candidate individual procreated from candidate individuals remaining in the candidate pool.

Abstract: In one embodiment, a system for predicting imputed revenue of inquiries includes: one or more computing devices comprising computer hardware and configured to: obtain data relating to a plurality of inquiries. Each of the plurality of inquiries is (a) indicative of a request for information about one or more programs providing a service, and (b) received from a particular channel of a plurality of online channels for presenting the one or more programs. The plurality of inquiries are represented using one or more data structures in one or more data sources. The computing devices are further configured to determine a model for predicting enrollment rates of the plurality of inquiries based at least in part on historical enrollment data of at least one program of the one or more programs, the model specifying a predicted enrollment rate value for one or more characteristics associated with the plurality of inquiries.

Abstract: A method, system and apparatus for inter-parties network computing. A policy management method can include establishing a set of policy terms for receiving content from a content provider based upon preferences selected by a content recipient externally to the content recipient and content provider. The method further can include enforcing the policy terms externally to the content recipient and content provider for content provided by the content provider for receipt by the content recipient. Notably, the establishing step can include mapping the selected preferences to the set of policy terms. The establishing step also can include exposing an interface to the content recipient, and accepting the selected preferences through the interface.

Abstract: Methods, systems, and media for managing online advertising campaigns based on causal conversion metrics are provided. In some embodiments, the method comprises: receiving conversion information corresponding to test group including consumers that were presented with an advertisement using an advertising channel; receiving advertisement viewability information indicative of a probability that each of the consumers viewed the advertisement; determining that a subset of the consumers did not view the advertisement based on the probability; placing the consumers into a control group and a test group based on the probability corresponding to each of the consumers; calculating a causal conversion metric based on a comparison of the conversion information corresponding to consumers of the control group and conversion information corresponding to consumers of the test group; and determining whether to place an advertisement using the advertising channel based on the causal conversion metric.