Optimization of a multi-channel system using a feedback loop
Methods, systems, and apparatus include computer programs encoded on a computer-readable storage medium, including a system that controls content distribution using a feedback loop. Content is distributed over multiple different online channels using a same initial selection value for distribution over each different online channel. An observed user actions required for distribution of the content over the multiple different online channels is received through a feedback loop and for multiple different distributions of the content. Based on the observed user actions received through the feedback loop, a predicted user action rate is determined for the multiple different distributions across the multiple different online channels. The selection value is adjusted based on a difference between the predicted user action rate and a reference distribution amount specified by a provider of the content. The content is distributed over the multiple different online channels using the adjusted selection value.
1. A method, comprising:
distributing, by one or more servers in a distributed computing environment, content multiple times over multiple different online channels using a same reference distribution amount specified by a provider of the content for distribution of the content, wherein a feedback loop is configured to obtain feedback about the distributions and adjust subsequent transmission of the content by adjusting a selection value used to distribute the content over the multiple different online channels;
receiving, through the feedback loop and for multiple different distributions of the content, observed user actions with the content distributed to multiple different client devices;
determining, by the one or more servers and based on the observed user actions received through the feedback loop, an observed distribution amount for the multiple different distributions of the content to the multiple different client devices;
adjusting, by the one or more servers, the selection value for subsequent distribution of the content based on a difference between the observed distribution amount for the multiple different distributions of the content to the multiple different client devices and a maximum distribution amount specified by the provider, wherein the maximum distribution amount specified by the provider is a total amount that the one or more servers are authorized to use for distribution of the content over a defined time period; and
distributing, by the one or more servers, the content using the adjusted selection value.
2. The method of claim 1 , further comprising:
determining a relationship between a predicted user action rate and the same reference distribution amount; and
adjusting the selection value based on the relationship between the predicted user action rate and the same reference distribution amount, including increasing the selection value when the predicted user action rate is less than a specified amount and decreasing the selection value when the predicted user action rate is greater than the specified amount.
3. The method of claim 2 , wherein adjusting the selection value based on a relationship between the predicted user action rate and the same reference distribution amount comprises adjusting the selection value by a capped adjustment amount when the predicted user action rate is greater than a specified amount.
4. The method of claim 2 , further comprising determining an average distribution amount for distributing the content over the multiple different online channels based on the observed user actions required for each distribution of the content over the multiple different online channels.
5. The method of claim 4 , wherein determining the average distribution amount comprises determining, for a given period of time, a ratio of a total observed user actions to a predicted user action rate for each distribution of the content during the given period of time.
6. The method of claim 1 , further comprising generating an initial instance of the selection value based on the same reference distribution amount and an initialization factor that causes the initial instance of the selection value to be greater than the same reference distribution amount.
7. The method of claim 1 , wherein distributing the content over multiple different online channels comprises distributing the content over each of a web page and a streaming media channel.
8. A system comprising:
a third-party content corpus database that stores information for various content that are available to be distributed to client devices; and
a distributed computing environment, including multiple computing devices and a feedback loop wherein the distributed computing environment is configured to interact with the third-party content corpus database and perform operations comprising:
distributing content multiple times over multiple different online channels using a same reference distribution amount specified by a provider of the content for distribution of the content, wherein a feedback loop is configured to obtain feedback about the distributions and adjust subsequent transmission of the content by adjusting a selection value used to distribute the content over the multiple different online channels;
receiving, through the feedback loop and for multiple different distributions of the content, observed user actions with the content distributed to multiple different client devices;
determining, based on the observed user actions received through the feedback loop, an observed distribution amount for the multiple different distributions of the content to the multiple different client devices and the same reference distribution amount specified by the provider;
adjusting the selection value for subsequent distribution of the content based on a difference between the observed distribution amount for the multiple different distributions of the content to the multiple different client devices and a maximum amount specified by the provider, wherein the maximum distribution amount specified by the provider is a total amount that the multiple computing devices are authorized to use for distribution of the content over a defined time period; and
distributing the content using the adjusted selection value.
9. The system of claim 8 , wherein the distributed computing environment is configured to perform operations further comprising:
determining a relationship between a predicted user action rate and the same reference distribution amount; and
adjusting the selection value based on the relationship between the predicted user action rate and the same reference distribution amount, including increasing the selection value when the predicted user action rate is less than a specified amount and decreasing the selection value when the predicted user action rate is greater than the specified amount.
10. The system of claim 9 , wherein adjusting the selection value based on a relationship between the predicted user action rate and the same reference distribution amount comprises adjusting the selection value by a capped adjustment amount when the predicted user action rate is greater than a specified amount.
11. The system of claim 9 , wherein the distributed computing environment is configured to perform operations further comprising determining an average distribution amount for distributing the content over the multiple different online channels based on the observed user actions required for each distribution of the content over the multiple different online channels.
12. The system of claim 11 , wherein determining the average distribution amount comprises determining, for a given period of time, a ratio of a total observed user actions to a predicted user action rate for each distribution of the content during the given period of time.
13. The system of claim 8 , wherein the distributed computing environment is configured to perform operations further comprising generating an initial instance of the selection value based on the same reference distribution amount and an initialization factor that causes the initial instance of the selection value to be greater than the same reference distribution amount.
14. The system of claim 8 , wherein distributing the content over multiple different online channels comprises distributing the content over each of a web page and a streaming media channel.
15. A non-transitory computer readable medium storing instructions that, when executed by one or more data processing apparatus, cause the one or more data processing apparatus to perform operations comprising:
distributing, in a distributed computing environment, content multiple times over multiple different online channels using a same reference distribution amount specified by a provider of the content for distribution of the content, wherein a feedback loop is configured to obtain feedback about the distributions and adjust subsequent transmission of the content by adjusting a selection value used to distribute the content over the multiple different online channels;
receiving, through the feedback loop and for multiple different distributions of the content, observed user actions with the content distributed to multiple different client devices;
determining, based on the observed user actions received through the feedback loop, an observed distribution amount for the multiple different distributions of the content to the multiple different client devices;
adjusting the selection value for subsequent distribution of the content based on a difference between the observed distribution amount for the multiple different distributions of the content to the multiple different client devices and a maximum distribution amount specified by the provider, wherein the maximum distribution amount specified by the provider is a total amount that the one or more data processing apparatus are authorized to use for distribution of the content over a defined time period; and
distributing the content using the adjusted selection value.
16. The non-transitory computer readable medium of claim 15 , wherein the instructions cause the one or more data processing perform operations further comprising:
determining a relationship between a predicted user action rate and the same reference distribution amount; and
adjusting the selection value based on the relationship between the predicted user action rate and the same reference distribution amount, including increasing the selection value when the predicted user action rate is less than a specified amount and decreasing the selection value when the predicted user action rate is greater than the specified amount.
17. The non-transitory computer readable medium of claim 16 , wherein adjusting the selection value based on a relationship between the predicted user action rate and the same reference distribution amount comprises adjusting the selection value by a capped adjustment amount when the predicted user action rate is greater than a specified amount.
18. The non-transitory computer readable medium of claim 16 , wherein the instructions cause the one or more data processing perform operations further comprising determining an average distribution amount for distributing the content over the multiple different online channels based on the observed user actions required for each distribution of the content over the multiple different online channels.
19. The non-transitory computer readable medium of claim 18 , wherein determining the average distribution amount comprises determining, for a given period of time, a ratio of a total observed user actions to a predicted user action rate for each distribution of the content during the given period of time.
20. The non-transitory computer readable medium of claim 15 , wherein the instructions cause the one or more data processing perform operations further comprising generating an initial instance of the selection value based on the same reference distribution amount and an initialization factor that causes the initial instance of the selection value to be greater than the same reference distribution amount.