Content-modification system with client-side advertisement caching

1. A method comprising:

identifying, by a computing system, a plurality of upcoming content-modification opportunities for a content-presentation device, each respective content-modification opportunity corresponding to a respective underlying advertisement that is available for replacement by a respective supplemental advertisement;

using, by the computing system, the identified upcoming content-modification opportunities as a basis for determining expected values for a plurality of supplemental advertisements, wherein, for each supplemental advertisement, the expected value is based at least in part on a probability of performing a successful content-modification operation using the supplemental advertisement and an expected revenue gained from performing the successful content-modification opportunity using the supplemental advertisement;

selecting, by the computing system, a subset of supplemental advertisements from among the plurality of supplemental advertisements based on the subset having expected values above a threshold value;

in advance of the upcoming content-modification opportunities, sending, by the computing system, the subset of supplemental advertisements to be locally stored at the content-presentation device; and

upon occurrence of one of the content-modification opportunities, causing, by the computing system, the content-presentation device to replace one of the respective underlying advertisements with one of the locally stored supplemental advertisements.

2. The method of claim 1 , wherein using the identified upcoming content-modification opportunities as a basis for determining expected values for the plurality of supplemental advertisements comprises determining the expected values based on one or more factors including (i) demographic information of an expected audience of the content-presentation device during the upcoming content-modification opportunities, (ii) television channels expected to be presented by the content-presentation device during the upcoming content-modification opportunities, (iii) a number of available impressions for each supplemental advertisement of the plurality of supplemental advertisements, (iv) a number of the identified upcoming content-modification opportunities, or (v) creative separation or versioning rules for each supplemental advertisement of the plurality of supplemental advertisements.

3. The method of claim 2 , wherein using the identified upcoming content-modification opportunities as a basis for determining expected values for the plurality of supplemental advertisements further comprises:

using the one or more factors as inputs for an artificial neural network; and

receiving as an output from the artificial neural network, the expected values for the plurality of supplemental advertisements.

4. The method of claim 3 , further comprising training, by the computing system, the artificial neural network based on actual revenues from content-modification operations performed in the past.

5. The method of claim 4 , further comprising repeatedly retraining the artificial neural network based on actual revenues from additional content-modification operations performed in the past.

6. The method of claim 1 , wherein the expected value is determined by multiplying the probability by the expected revenue.

7. A non-transitory computer-readable storage medium, having stored thereon program instructions that, upon execution by a processor, cause performance of a set of operations comprising:

identifying, by a computing system, a plurality of upcoming content-modification opportunities for a content-presentation device, each respective content-modification opportunity corresponding to a respective underlying advertisement that is available for replacement by a respective supplemental advertisement;

using, by the computing system, the identified upcoming content-modification opportunities as a basis for determining expected values for a plurality of supplemental advertisements, wherein, for each supplemental advertisement, the expected value is based at least in part on a probability of performing a successful content-modification operation using the supplemental advertisement and an expected revenue gained from performing the successful content-modification opportunity using the supplemental advertisement;

selecting, by the computing system, a subset of supplemental advertisements from among the plurality of supplemental advertisements based on the subset having expected values above a threshold value;

in advance of the upcoming content-modification opportunities, sending, by the computing system, the subset of supplemental advertisements to be locally stored at the content-presentation device; and

upon occurrence of one of the content-modification opportunities, causing, by the computing system, the content-presentation device to replace one of the respective underlying advertisements with one of the locally stored supplemental advertisements.

8. The non-transitory computer-readable storage medium of claim 7 , wherein using the identified upcoming content-modification opportunities as a basis for determining expected values for the plurality of supplemental advertisements comprises determining the expected values based on one or more factors including (i) demographic information of an expected audience of the content-presentation device during the upcoming content-modification opportunities, (ii) television channels expected to be presented by the content-presentation device during the upcoming content-modification opportunities, (iii) a number of available impressions for each supplemental advertisement of the plurality of supplemental advertisements, (iv) a number of the identified upcoming content-modification opportunities, or (v) creative separation or versioning rules for each supplemental advertisement of the plurality of supplemental advertisements.

9. The non-transitory computer-readable storage medium of claim 8 , wherein using the identified upcoming content-modification opportunities as a basis for determining expected values for the plurality of supplemental advertisements further comprises:

using the one or more factors as inputs for an artificial neural network; and

receiving as an output from the artificial neural network, the expected values for the plurality of supplemental advertisements.

10. The non-transitory computer-readable storage medium of claim 9 , the set of operations further comprising training, by the computing system, the artificial neural network based on actual revenues from content-modification operations performed in the past.

11. The non-transitory computer-readable storage medium of claim 10 , the set of operations further comprising repeatedly retraining the artificial neural network based on actual revenues from additional content-modification operations performed in the past.

12. The non-transitory computer-readable storage medium of claim 7 , wherein the expected value is determined by multiplying the probability by the expected revenue.

13. A computing system comprising:

a processor; and

a non-transitory computer-readable storage medium, having stored thereon program instructions that, upon execution by the processor, cause performance of a set of operations comprising:

identifying, by the computing system, a plurality of upcoming content-modification opportunities for a content-presentation device, each respective content-modification opportunity corresponding to a respective underlying advertisement that is available for replacement by a respective supplemental advertisement;

using, by the computing system, the identified upcoming content-modification opportunities as a basis for determining expected values for a plurality of supplemental advertisements, wherein, for each supplemental advertisement, the expected value is based at least in part on a probability of performing a successful content-modification operation using the supplemental advertisement and an expected revenue gained from performing the successful content-modification opportunity using the supplemental advertisement;

selecting, by the computing system, a subset of supplemental advertisements from among the plurality of supplemental advertisements based on the subset having expected values above a threshold value;

in advance of the upcoming content-modification opportunities, sending, by the computing system, the subset of supplemental advertisements to be locally stored at the content-presentation device; and

upon occurrence of one of the content-modification opportunities, causing, by the computing system, the content-presentation device to replace one of the respective underlying advertisements with one of the locally stored supplemental advertisements.

14. The computing system of claim 13 , wherein using the identified upcoming content-modification opportunities as a basis for determining expected values for the plurality of supplemental advertisements comprises determining the expected values based on one or more factors including (i) demographic information of an expected audience of the content-presentation device during the upcoming content-modification opportunities, (ii) television channels expected to be presented by the content-presentation device during the upcoming content-modification opportunities, (iii) a number of available impressions for each supplemental advertisement of the plurality of supplemental advertisements, (iv) a number of the identified upcoming content-modification opportunities, or (v) creative separation or versioning rules for each supplemental advertisement of the plurality of supplemental advertisements.

15. The computing system of claim 14 , wherein using the identified upcoming content-modification opportunities as a basis for determining expected values for the plurality of supplemental advertisements further comprises:

using the one or more factors as inputs for an artificial neural network; and

receiving as an output from the artificial neural network, the expected values for the plurality of supplemental advertisements.

16. The computing system of claim 15 , the set of operations further comprising training, by the computing system, the artificial neural network based on actual revenues from content-modification operations performed in the past.

17. The computing system of claim 16 , the set of operations further comprising repeatedly retraining the artificial neural network based on actual revenues from additional content-modification operations performed in the past.

18. The method of claim 1 , wherein the expected revenue is based on a cost-per-mille of the supplemental advertisement and an opportunity cost associated with performing the successful content-modification opportunity using the supplemental advertisement.

19. The non-transitory computer-readable storage medium of claim 7 , wherein the expected revenue is based on a cost-per-mille of the supplemental advertisement and an opportunity cost associated with performing the successful content-modification opportunity using the supplemental advertisement.

20. The computing system of claim 13 , wherein the expected revenue is based on a cost-per-mille of the supplemental advertisement and an opportunity cost associated with performing the successful content-modification opportunity using the supplemental advertisement.