Machine-learned model including incrementality estimation

A system comprising:

• a non-transitory computer-readable medium having instructions stored thereon, which, when executed by a processor, cause the system to perform operations comprising:
• obtaining a first piece of content from a first content poster and a first budget for the first piece of content, the first budget indicative of a first value of promoting the first piece of content over a particular time period;
• splitting the first budget asymmetrically into a high budget value and a low budget value, the high budget value being different and higher than the low budget value;
• randomly assigning each user in a first group of users into a first subgroup or a second subgroup;
• randomly assigning the high budget value to either the first subgroup or the second subgroup and assigning the low budget value to the subgroup to which the high budget value was not assigned;
• causing the first piece of content to be displayed to users in the first subgroup based on the budget value assigned to the first subgroup;
• causing the first piece of content to be displayed to users in the second subgroup based on the budget value assigned to the second subgroup;
• designating results produced by the causing display of the first piece of content to the users in the first subgroup and the causing display of the first piece of content to the users in the second subgroup as training data;
• training the first machine-learned model using a machine learning algorithm by feeding the training data into the machine learning algorithm;
• generating a web page comprising a web-based user interface, the web-based user interface including a first user interface element configured to enable selection of a second piece of content, a second engageable user interface element configured to adjust a budget for the second piece of content upwards or downwards responsive to user engagement therewith, and a third user interface element displaying a first forecasted value of a performance-based metric for the second piece of content;
• receiving an adjusted budget for the second piece of content responsive to a detected engagement with the engageable second user interface element of the web-based user interface;
• sending information about the second piece of content and sending the adjusted budget for the second piece of content to the first machine-learned model;
• generating an updated forecasted value of the performance-based metric by the first machine-learned model for the second piece of content based at least in part on the adjusted budget for the second piece of content, the updated forecasted value comprising an incremental increase or decrease in the first forecasted value;
• and
• dynamically updating the third user interface element in the web-based user interface to display the generated updated forecasted value of the performance-based metric.

A system comprising:

• means for obtaining a first piece of content from a first content poster and a first budget for the first piece of content, the first budget indicative of a first value of promoting the first piece of content over a particular time period;
• means for splitting the first budget asymmetrically into a high budget value and a low budget value, the high budget value being different and higher than the low budget value;
• means for randomly assigning each user in a first group of users into a first subgroup or a second subgroup;
• means for randomly assigning the high budget value to either the first subgroup or the second subgroup and assigning the low budget value to the subgroup to which the high budget value was not assigned;
• means for causing the first piece of content to be displayed to users in the first subgroup based on the budget value assigned to the first subgroup;
• means for causing the first piece of content to be displayed to users in the second subgroup based on the budget value assigned to the second subgroup;
• means for designating results produced by the causing display of the first piece of content to the users in the first subgroup and the causing display of the first piece of content to the users in the second subgroup as training data;
• means for training the first machine-learned model using a machine learning algorithm by feeding the training data into the machine learning algorithm;
• means for generating a web page comprising a web-based user interface, the web-based user interface including a first user interface element configured to enable selection of a second piece of content, a second engageable user interface element configured to adjust a budget for the second piece of content upwards or downwards responsive to user engagement therewith, and a third user interface element displaying a first forecasted value of a performance-based metric for the second piece of content;
• means for receiving an adjusted budget for the second piece of content responsive to a detected engagement with the engageable second user interface element of the web-based user interface;
• means for sending information about the second piece of content and sending the adjusted budget for the second piece of content to the first machine-learned model;
• means for generating an updated forecasted value of the performance-based metric by the first machine-learned model for the second piece of content based at least in part on the adjusted budget for the second piece of content, the updated forecasted value comprising an incremental increase or decrease in the first forecasted value;
• and
• means for dynamically updating the third user interface element in the web-based user interface to display the generated updated forecasted value of the performance-based metric.