Predicting keyword monetization

• US 8,682,839 B2
• Filed: 06/02/2008
• Issued: 03/25/2014
• Est. Priority Date: 06/02/2008
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Li, Hua, Chen, Zheng, Wang, Jian
• Primary Examiner(s): Chang, Edward
• Application Number:
  US12/131,125
• Publication Number:
  US 20090299855A1
• Time in Patent Office: 2,122 Days
• Field of Search: 707/3, 705/35
• US Class Current: 707/3
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising
• CPC Class Codes: G06Q 30/0256 User search
• CPC Class Codes: G06Q 30/0267 Wireless devices
• CPC Class Codes: G06Q 30/08 Auctions

Abstract

Embodiments of the claimed subject matter provide a method and system for predicting bidding keyword monetization. The claimed subject matter provides a method and system with which the value of a keyword for the purpose of relevant online advertisement may be evaluated according to various metrics to determine a bidding landscape for use in advertising campaigns. The value of the keyword considers certain attributes related to the monetization of the keyword.

One embodiment of the claimed subject matter is implemented as a method for predicting keyword monetization for one or more keyword-advertisement relationships. Historical data for the one or more keyword-advertisement relationships is referenced and used to generate a global model of the one or more keyword-advertisement relationship. The relationships are then evaluated according to a time-series analysis, which parses the data from the historical data and the global model to create predictions for the keyword monetization according to the keyword-advertisement relationships.

Claims

1. A method of predicting keyword monetization for a period of time for one or more advertisements corresponding to a keyword in one or more keyword-advertiser relationships, the method comprising:

fetching historical data for the one or more keyword-advertiser relationships;

generating a global model incorporating the one or more keyword-advertiser relationships from the historical data; and

predicting the keyword monetization based on the historical data for the one or more keyword-advertiser relationships based on the generated global model, the keyword monetization comprising a prediction of performance for the one or more advertisements corresponding to the keyword,

wherein instructions to perform the fetching, the generating, and the predicting are comprised on computer readable media and, the method comprises executing the instructions with a computing device.

2. The method of claim 1, wherein the predicting further comprises:

evaluating the one or more keyword-advertiser relationships according to a time-series analysis.

3. The method of claim 2, wherein the generating comprises:

calculating an average cost per click rate for the keyword from the historical data; and

calculating an average clickthrough rate for the keyword from the historical data.

4. The method of claim 2, wherein the evaluating comprises:

predicting a number of impressions for an individual advertisement from an individual keyword-advertiser relationship that will be displayed during the period of time;

predicting a cost for an advertiser to advertise with the keyword during the period of time;

predicting a number of clicks to be generated by the individual advertisement from the indvidual keyword-advertiser relationship during the period of time;

predicting a bidding price for the keyword for the period of time; and

predicting a number of conversions to be generated by the individual advertisement from an individual keyword-advertiser relationship during the period of time.

5. The method of claim 2, wherein the evaluating further comprises predicting an average position in which an individual advertisement from an individual keyword-advertiser relationship will be displayed in a search return list generated by a search engine in response to a search query which includes the keyword, the search return list being generated during the period of time.

6. The method of claim 5, wherein the predicting the average position comprises:

calculating a historical average position in which the individual advertisement is displayed in other search return lists generated by the search engine in response to other search queries that include the keyword; and

applying one or more regression models to the historical average position.

7. The method of claim 6, wherein the one or more regression models comprise:

a linear regression model;

a least-square regression model;

or

a slope-based regression model.

8. The method of claim 5 wherein the predicting the average position further comprises finding a corresponding period of time to be predicted from the historical data of the individual keyword-advertiser relationship.

9. The method of claim 4, wherein the predicting the number of impressions for the individual advertisement from the individual keyword-advertiser relationship further comprises applying constraints from the generated global model.

10. The method of claim 4, wherein the predicting the number of conversions further comprises calculating an average value of the performance.

21. The method of claim 1, wherein the fetching comprises referencing the historical data, the historical data corresponding to advertising between a plurality of advertisers and a plurality of advertiser publishers with respect to the keyword.

22. The method of claim 1, wherein the fetching comprises referencing the historical data, the historical data corresponding to a plurality of advertising campaigns engaged in by the one or more keyword-advertiser relationships.

11. A method for generating a keyword performance landscape for one or more keyword-advertiser relationships representing one or more advertisers with one or more advertisements corresponding to a keyword, the method comprising:

connecting to a data resource containing data for the one or more keyword-advertiser relationships, the data including historical data of the one or more keyword-advertiser relationships;

fetching the historical data for the one or more keyword-advertiser relationships, the historical data including a historical keyword monetization property for the one or more keyword-advertiser relationships;

predicting a future keyword monetization property comprising a prediction of a performance for the one or more advertisements corresponding to the keyword for a period of time, wherein the predicting the future keyword monetization property includes generating a global model incorporating the one or more keyword-advertiser relationships and evaluating the one or more keyword-advertiser relationships according to a time-series analysis; and

generating the keyword performance landscape for the one or more keyword-advertiser relationships from the historical data and the predicted future keyword monetization property,

wherein instructions to perform the connecting, the fetching, the predicting, and the generating are comprised on computer readable media and the method comprises executing the instructions with a computing system.

12. The method of claim 11, wherein the predicted future keyword monetization property comprises:

an average bidding price for an individual advertisement corresponding to the keyword, the individual advertisement being from an individual keyword-advertiser relationship;

a number of impressions for the individual advertisement when displayed during the period of time;

a number of clicks generated by the individual advertisement during the period of time;

a cost incurred by an individual advertiser of the individual advertisement to advertise with the keyword during the period of time; and

a number of conversions generated by the individual advertisement during the period of time.

13. The method of claim 12, wherein the predicted future keyword monetization property further comprises an average position in which the individual advertisement is displayed in a search return list generated by a search engine in response to a search query which includes the keyword, the search return list being generated during the period of time.

14. The method of claim 11, wherein the period of time is a day.

15. The method of claim 11 further comprising outputting the keyword performance landscape.

23. The method of claim 11, wherein the fetching comprises referencing the historical data, the historical data being for advertising between a plurality of advertisers and a plurality of advertiser publishers with respect to the keyword.

24. The method of claim 11, wherein the fetching comprises referencing the historical data, the historical data corresponding to a plurality of advertising campaigns engaged by the one or more keyword-advertiser relationships.

16. A system for generating a keyword performance landscape including one or more keyword-advertiser relationships, the system comprising:

a computer system having a processing unit coupled to a memory, the memory having computer readable code, which when executed by the processing unit causes the computer system to implement;

an advertiser production data resource; and

a keyword monetization engine, wherein the keyword monetization engine predicts monetization by predicting performance of advertisements corresponding to a keyword for a duration of time by generating a global model incorporating the one or more keyword-advertiser relationships and evaluating the one or more keyword-advertiser relationships according to a time-series analysis, generating the keyword performance landscape from the global model and time series analysis, and outputting the keyword performance landscape.

17. The system according to claim 16, wherein the advertisements are displayed on a remote computing device.

18. The system according to claim 17, wherein the remote computing device is another computer system.

19. The system according to claim 17, wherein the remote computing device is a hand-held computing device.

20. The system according to claim 17, wherein the advertisements are displayed in a search result list generated by a search engine that is accessed by a user over the Internet.