Automated campaign configuration switching framework
Dynamic campaign optimization systems and methods may be used to continuously test many alternative campaign configurations while allowing all configurations, including configurations formerly identified as successful and unsuccessful, to be re-tested in order to identify successful configurations that may previously have been identified as unsuccessful.
1 . A method comprising:
receiving, over a network connection and at a computing device, user requests originating from respective client devices, for a content element;
identifying, by the computing device, configuration alternatives for the content element;
in response to each of the user requests:
selecting one of the configuration alternatives based on initial weights of the configuration alternatives;
configuring, by the computing device, an alternative-configured impression of the requested content element according to the selecting; and
presenting the alternative-configured impression to the respective user request, wherein presenting includes transmitting the alternative-configured impression of the content element to a respective originating client device;
updating, by the computing device, each respective initial weight of the configuration alternatives to a respective second weight based on subsequent requests for the content element corresponding to the presented alternative-configured impression;
automatically selecting, by the computing device and after the updating, a configuration alternative, the configuration alternative having a second weight that is below a threshold; and
based on the automatically selecting, increasing, by the computing device the second weight of the selected configuration alternative to a third weight by adding a random amount to the second weight.
2 . The method of claim 1 , further comprising:
calculating a success metric value for each configuration alternative based on the subsequent requests from users to whom each corresponding alternative-configured impression is presented; and
wherein the updating is in proportion to the success metric each configuration alternative.
3 . The method of claim 2 , wherein the updating is further based on a member of the group comprising: statistical significance intervals of the success metric, a number of trials in which each configuration alternative has been tested, a total number of trials in which all configuration alternatives in a campaign have been tested, a frequency of configuration alternative testing trials, timing history of configuration alternative testing trials, dates configuration alternative test impressions, and a geographical location of users viewing a tested configuration alternative.
4 . The method of claim 1 , wherein the automatic selecting comprises selecting a configuration alternative with a statistical confidence interval wider than a threshold confidence interval.
5 . The method of claim 1 , wherein the automatic selecting comprises selecting a configuration alternative with a number of impressions during a past period of time below a threshold number of impressions.
6 . The method of claim 1 , wherein identifying the configuration alternatives comprises identifying information describing a user and selecting a plurality of configuration alternatives based on the information describing the user.
7 . The method of claim 6 , wherein identifying information describing the user comprises identifying a user cohort of which the user is a member.
8 . A non-transitory computer-readable medium comprising instructions, which when executed by at least one processor, configure the processor to perform operations comprising:
receiving, over a network connection and at a computing device, user requests originating from respective client devices, for a content element;
identifying, by the computing device, configuration alternatives for the content element;
in response to each of the user requests:
selecting one of the configuration alternatives based on initial weights of the configuration alternatives;
configuring, by the computing device, an alternative-configured impression of the requested content element according to the selecting; and
presenting the alternative-configured impression to the respective user request, wherein presenting includes transmitting the alternative-configured impression of the content element to a respective originating client device;
updating, by the computing device, each respective initial weight of the configuration alternatives to a respective second weight based on subsequent requests for the content element corresponding to the presented alternative-configured impression;
automatically selecting, by the computing device and after the updating, a configuration alternative, the configuration alternative having a second weight that is below a threshold; and
based on the automatically selecting, increasing, by the computing device the second weight of the selected configuration alternative to a third weight by adding a random amount to the second weight.
9 . The non-transitory computer-readable medium of claim 8 , the operations further comprising:
calculating a success metric value for each configuration alternative based on the subsequent requests from users to whom each corresponding alternative-configured impression is presented; and
wherein the updating is in proportion to the success metric each configuration alternative.
10 . The non-transitory computer-readable medium of claim 9 , wherein the updating is further based on a member of the group comprising: statistical significance intervals of the success metric, a number of trials in which each configuration alternative has been tested, a total number of trials in which all configuration alternatives in a campaign have been tested, a frequency of configuration alternative testing trials, timing history of configuration alternative testing trials, dates configuration alternative test impressions, and a geographical location of users viewing a tested configuration alternative.
11 . The non-transitory computer-readable medium of claim 8 , wherein the automatic selecting comprises selecting a configuration alternative with a statistical confidence interval wider than a threshold confidence interval.
12 . The non-transitory computer-readable medium of claim 8 , wherein the automatic selecting comprises selecting a configuration alternative with a number of impressions during a past period of time below a threshold number of impressions.
13 . The non-transitory computer-readable medium of claim 8 , wherein identifying the configuration alternatives comprises identifying information describing a user and selecting a plurality of configuration alternatives based on the information describing the user.
14 . The non-transitory computer-readable medium of claim 13 , wherein identifying information describing the user comprises identifying a user cohort of which the user is a member.
15 . A system comprising:
at least one processor;
at least one storage device comprising instructions, which when executed by the at least one processor, configure the at least one processor to perform operations comprising:
receiving, over a network connection, user requests originating from respective client devices, for a content element;
identifying configuration alternatives for the content element;
in response to each of the user requests:
selecting one of the configuration alternatives based on initial weights of the configuration alternatives;
configuring an alternative-configured impression of the requested content element according to the selecting; and
presenting the alternative-configured impression to the respective user request, wherein presenting includes transmitting the alternative-configured impression of the content element to a respective originating client device;
updating each respective initial weight of the configuration alternatives to a respective second weight based on subsequent requests for the content element corresponding to the presented alternative-configured impression;
automatically selecting, after the updating, a configuration alternative, the configuration alternative having a second weight that is below a threshold; and
based on the automatically selecting, increasing, the second weight of the selected configuration alternative to a third weight by adding a random amount to the second weight.
16 . The system of claim 15 , the operations further comprising:
calculating a success metric value for each configuration alternative based on the subsequent requests from users to whom each corresponding alternative-configured impression is presented; and
wherein the updating is in proportion to the success metric each configuration alternative.
17 . The system of claim 16 , wherein the updating is further based on a member of the group comprising: statistical significance intervals of the success metric, a number of trials in which each configuration alternative has been tested, a total number of trials in which all configuration alternatives in a campaign have been tested, a frequency of configuration alternative testing trials, timing history of configuration alternative testing trials, dates configuration alternative test impressions, and a geographical location of users viewing a tested configuration alternative.
18 . The system of claim 15 , wherein the automatic selecting comprises selecting a configuration alternative with a statistical confidence interval wider than a threshold confidence interval.
19 . The system of claim 15 , wherein the automatic selecting comprises selecting a configuration alternative with a number of impressions during a past period of time below a threshold number of impressions.
20 . The system of claim 15 , wherein identifying the configuration alternatives comprises identifying information describing a user and selecting a plurality of configuration alternatives based on the information describing the user.
21 . The system of claim 20 , wherein identifying information describing the user comprises identifying a user cohort of which the user is a member.