Systems and methods for a television scoring service that learns to reach a target audience
View Patent ↗Television is the largest advertising category in the United States with over 65 billion spent by advertisers per year. A variety of different targeting algorithms are compared, ranging from the traditional age-gender targeting methods employed based on Nielsen ratings, to new approaches that attempt to target high probability buyers using Set Top Box data. The performance of these different algorithms on a real television campaign is shown, and the advantages and limitations of each method are discussed. In contrast to other theoretical work, all methods presented herein are compatible with targeting the existing 115 million Television households in the United States and are implementable on current television delivery systems.
1. A computer-implemented method comprising:
receiving, at a server, one or more heterogeneous sources of media data, the media data including television viewing events;
generating, by the server, a plurality of media asset patterns from the one or more heterogeneous sources of media data, the plurality of media asset patterns being possible media placements which are represented as conjunctive expressions;
calculating, by the server, one or more heterogeneous advertisement effectiveness measures for each media asset pattern;
calculating, by the server, for a plurality of pairs of an advertisement and a media instance, a number of previously placed airings of the advertisement in the media instance;
generating, by the server, a model to predict advertisement effectiveness for each pairing of an advertisement and a media instance based on a combination of the advertisement effectiveness measures and the number of previously placed airings of the advertisement in the media instance; and
defining, by the server, for each of the one or more heterogeneous advertisement effectiveness measures, a default value and a minimum participation threshold,
wherein the model adjusts a value of a first advertisement effectiveness measure for a first particular pairing of an advertisement and a media asset to the default value of the first particular advertisement effectiveness measure if a number of impressions for the first particular pairing of the advertisement and the media asset is below the minimum participation threshold for the first advertisement effectiveness measure, and
wherein the model includes one or more model parameters that are automatically generated by minimizing an error in predicting historical advertisement effectiveness measures.
2. The method of claim 1 , wherein the one or more heterogeneous advertisement effectiveness measures include one or more of phone responses, demographic similarity, set top box buyers, and web responses.
3. The method of claim 1 , wherein the number of previously placed airings of the advertisement in the media instance is estimated based on a number of historical airings and co-viewing probabilities from set top box data.
4. The method of claim 1 , wherein weights for different heterogeneous media sources are based on one or more of a count of persons, and other data sufficiency statistics.
5. The method of claim 1 , wherein the model disregards a second advertisement effectiveness measure for a second particular pairing of an advertisement and a media asset if a number of impressions for the second particular pairing of the advertisement and the media asset is below the minimum participation threshold for the second advertisement effectiveness measure.
6. The method of claim 1 , further comprising:
applying the model to a plurality of media assets for a particular advertisement to assist in selection of one or more of the plurality of media assets for airing the advertisement.
7. The method of claim 1 , wherein one of the media asset patterns is identified as media asset patterns with at least a predetermined number of observed viewers over an expected number of viewers for a predetermined time and a predetermined network.
8. The method of claim 1 , further comprising:
creating, by the server, a media asset pattern of same-time-last-week; and
calculating, by the server, an advertisement effectiveness measure for the media asset pattern.
9. The method of claim 1 , wherein the media asset patterns include one or more of station, program, station-program, station-day-hour, station-day-hour-program-market.
10. The method of claim 1 , wherein each advertisement effectiveness measure for media asset pattern predictor is standardized so that heterogeneous media asset patterns are directly comparable with each other.
11. The method of claim 10 , wherein the standardized values are combined along with historical airing count to predict a standardized advertisement effectiveness.
12. The method of claim 11 , wherein the predicted standardized advertisement effectiveness is converted into native units including buyers per million and phone responses per million.
13. A system for generating a model to predict advertisement effectiveness, the system comprising:
a data storage device that stores instructions for generating a model to predict advertisement effectiveness; and
a processor configured to execute the instructions to perform a method including: receiving one or more heterogeneous sources of media data, the media data including television viewing events;
generating a plurality of media asset patterns from the one or more heterogeneous sources of media data, the plurality of media asset patterns being possible media placements which are represented as conjunctive expressions;
calculating one or more heterogeneous advertisement effectiveness measures for each media asset pattern;
calculating for a plurality of pairs of an advertisement and a media instance, a number of previously placed airings of the advertisement in the media instance;
generating a model to predict advertisement effectiveness for each pairing of an advertisement and a media instance based on a combination of the advertisement effectiveness measures and the number of previously placed airings of the advertisement in the media instance; and
defining, by the server, for each of the one or more heterogeneous advertisement effectiveness measures, a default value and a minimum participation threshold,
wherein the model adjusts a value of a first advertisement effectiveness measure for a first particular pairing of an advertisement and a media asset to the default value of the first advertisement effectiveness measure if a number of impressions for the first particular pairing of the advertisement and the media asset is below the minimum participation threshold for the first advertisement effectiveness measure, and
wherein the model includes one or more model parameters that are automatically generated by minimizing an error in predicting historical advertisement effectiveness measures.
14. The system of claim 13 , wherein the one or more heterogeneous advertisement effectiveness measures include one or more of phone responses, demographic similarity, set top box buyers, and web responses.
15. The system of claim 13 , wherein the number of previously placed airings of the advertisement in the media instance is estimated based on a number of historical airings and co-viewing probabilities from set top box data.
16. The system of claim 13 , wherein weights for different heterogeneous media sources are based on one or more of a count of persons, and other data sufficiency statistics.
17. The system of claim 13 , wherein the model disregards a second advertisement effectiveness measure for a second particular pairing of an advertisement and a media asset if a number of impressions for the second particular pairing of the advertisement and the media asset is below the minimum participation threshold for the second advertisement effectiveness measure.