Capturing a cluster effect with targeted digital-content exposures
This disclosure covers methods, non-transitory computer readable media, and systems that determine a cluster effect representing the impact that a user's digital-content exposure has on other users' conversion actions. The disclosed methods, non-transitory computer readable media, and systems detect the downloads, purchases, or other forms of consumption of a featured item that result from users within the same cluster viewing digital content featuring the item. In some embodiments, the methods, non-transitory computer readable media, and systems apply the cluster effect by, for example, generating a conversion report representing the cluster effect or by providing tools that exploit the cluster effect in distributing digital content.
1. A method comprising:
generating clusters of users, each cluster having one or more users of a social networking system;
separating the clusters of users into a first group of clusters and a second group of clusters;
assigning particular clusters of users from the first group of clusters to a first test group and a first control group;
determining a first incremental lift between the first test group and the first control group from the first group of clusters;
assigning individual users from the second group of clusters to a second test group and a second control group;
determining a second incremental lift between the second test group and the second control group from the second group of clusters; and
comparing the first incremental lift to the second incremental lift to determine a cluster effect.
2. The method of claim 1 , wherein generating the clusters of users comprises generating the clusters of users based on one or more of an affinity coefficient, a designated relationship, a commonly used Internet Protocol (“IP”) address, a common physical address, a common device identifier, tagged users within images, a common employer, or a common educational institution.
3. The method of claim 1 , further comprising modifying distribution of digital content based on the cluster effect.
4. The method of claim 3 , wherein modifying the distribution of the digital content based on the cluster effect comprises increasing distribution of the digital content to users within clusters comprising multiple users having one or more common attributes.
5. The method of claim 1 , further comprising:
generating a conversion report comprising a representation of the cluster effect; and
providing the conversion report to a client device.
6. The method of claim 5 , wherein the conversion report comprises a selectable option to adjust distribution of digital content to users within clusters comprising multiple users having one or more common attributes.
7. The method of claim 6 , wherein the conversion report comprises selectable options to adjust distribution of the digital content to users within clusters of a target demographic or target location.
8. A system comprising:
at least one processor; and
at least one non-transitory computer readable storage medium storing instructions that, when executed by the at least one processor, cause the system to:
generate clusters of users, each cluster having one or more users of a social networking system;
separate the clusters of users into a first group of clusters and a second group of clusters;
assign particular clusters of users from the first group of clusters to a first test group and a first control group;
determine a first incremental lift between the first test group and the first control group from the first group of clusters;
assign individual users from the second group of clusters to a second test group and a second control group;
determine a second incremental lift between the second test group and the second control group from the second group of clusters; and
compare the first incremental lift to the second incremental lift to determine a cluster effect.
9. The system of claim 8 , further comprising instructions that, when executed by the at least one processor, cause the system to:
generate the clusters of users by assigning user identifiers and cluster identifiers to the users of the social networking system; and
separate the clusters of users into the first group of clusters and the second group of clusters by separating the clusters of users into the first group of clusters and the second group of clusters based on the cluster identifiers.
10. The system of claim 9 , further comprising instructions that, when executed by the at least one processor, cause the system to assign the cluster identifiers to the users of the social networking system by applying a clustering algorithm to assign the cluster identifiers to the users of the social networking system.
11. The system of claim 10 , further comprising instructions that, when executed by the at least one processor, cause the system to separate the clusters of users into the first group of clusters and the second group of clusters based on the cluster identifiers by randomly assigning a particular cluster of users to the first group of clusters or the second group of clusters based on randomly associating a cluster identifier corresponding to the particular cluster of users with either the first group of clusters or the second group of clusters.
12. The system of claim 9 , further comprising instructions that, when executed by the at least one processor, cause the system to:
assign the particular clusters of users from the first group of clusters to the first test group and the first control group by randomly assigning the particular clusters of users from the first group of clusters to the first test group or the first control group based on cluster identifiers; and
assign the individual users from the second group of clusters to the second test group and the second control group by randomly assigning users from the second group of clusters to the second test group or the second control group based on user identifiers.
13. The system of claim 8 , further comprising instructions that, when executed by the at least one processor, cause the system to:
determine the first incremental lift between the first test group and the first control group by delivering one or more instances of digital content to the first test group but not to the first control group; and
determine the second incremental lift between the second test group and the second control group by delivering the one or more instances of digital content to the second test group but not to the second control group.
14. The system of claim 8 , further comprising instructions that, when executed by the at least one processor, cause the system to generate clusters of users by applying a clustering algorithm to assign one or more users to a particular cluster of users based on attributes of the one or more users.
15. The system of claim 14 , further comprising instructions that, when executed by the at least one processor, cause the system to:
apply an alternative clustering algorithm to generate alternative clusters of users, each alternative cluster having one or more users of the social networking system;
separate the alternative clusters of users into an alternative first group of clusters and an alternative second group of clusters;
determine an alternative first incremental lift between an alternative first test group and an alternative first control group each comprising alternative clusters of users from the alternative first group of clusters;
determine an alternative second incremental lift between an alternative second test group and an alternative second control group each comprising individual users from the alternative second group of clusters;
compare the alternative first incremental lift to the alternative second incremental lift to determine an alternative cluster effect; and
based on a comparison of the cluster effect and the alternative cluster effect, identify either the clustering algorithm or the alternative clustering algorithm as generating clusters that capture a larger cluster effect.
16. A non-transitory computer readable medium storing instructions thereon that, when executed by at least one processor, cause a computer system to:
generate clusters of users, each cluster having one or more users of a social networking system;
separate the clusters of users into a first group of clusters and a second group of clusters;
assign particular clusters of users from the first group of clusters to a first test group and a first control group;
determine a first incremental lift between the first test group and the first control group from the first group of clusters;
assign individual users from the second group of clusters to a second test group and a second control group;
determine a second incremental lift between the second test group and the second control group from the second group of clusters; and
compare the first incremental lift to the second incremental lift to determine a cluster effect.
17. The non-transitory computer readable medium of claim 16 , further comprising instructions that, when executed by the at least one processor, cause the computer system to generate the clusters of users by generating a first cluster comprising a single user and a second cluster comprising multiple users having one or more common attributes.
18. The non-transitory computer readable medium of claim 16 , further comprising instructions that, when executed by the at least one processor, cause the computer system to generate the clusters of users by generating clusters of one or more classmates, club members, coworkers, households, neighbors, organizational members, or social-network friends.
19. The non-transitory computer readable medium of claim 16 , further comprising instructions that, when executed by the at least one processor, cause the computer system to:
determine the first incremental lift between the first test group and the first control group by determining a first average-spend lift, a first conversion lift, a first incremental-sales lift, a first incremental-consumption lift, a first incremental-spend-amount lift, or a first total-sales lift; and
determine the second incremental lift between the second test group and the second control group by determining a second average-spend lift, a second conversion lift, a second incremental-sales lift, a second incremental-consumption lift, a second incremental-spend-amount lift, or a second total-sales lift.
20. The non-transitory computer readable medium of claim 16 , further comprising instructions that, when executed by the at least one processor, cause the computer system to compare the first incremental lift to the second incremental lift to determine the cluster effect by determining a divergence between the first incremental lift and the second incremental lift.