Online advertiser keyword valuation to decide whether to acquire the advertiser

• US 7,805,331 B2
• Filed: 11/20/2007
• Issued: 09/28/2010
• Est. Priority Date: 11/20/2007
• Status: Expired due to Fees

• Current Assignee: R2 Solutions LLC (Acacia Research Corporation)
• Original Assignee: Yahoo! Inc. (Apollo Global Management, Inc.)
• Inventors: Jin, Jiaqi, Chaudhury, Supratim Roy, Demir, Ramazan, Tomak, Kerem
• Primary Examiner(s): DIXON, THOMAS A
• Application Number:
  US11/943,192
• Publication Number:
  US 20090132336A1
• Time in Patent Office: 1,043 Days
• Field of Search: 705/10
• US Class Current: 705/14.4
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising
• CPC Class Codes: G06Q 30/0201 Market modelling; Market analysis; Collecting market data
• CPC Class Codes: G06Q 30/0241 Advertisements

Abstract

A method for valuating an advertiser considered for acquisition includes capturing click-related data by users with reference to a plurality of advertisements owned by a plurality of advertisers in a keyword market; calculating a marginal value of each of the plurality of advertisers as a function of a plurality of averaged market factors, wherein the marginal value of an additional advertiser is also calculated to result in at least an estimated displaced revenue if at least one advertisement of the additional advertiser for at least one keyword were competitively included in the keyword market; and deciding whether to pursue the additional advertiser based on the marginal value of the additional advertiser as associated with the at least one keyword.

Claims

1. A method for valuating an advertiser considered for acquisition, the method executed by a server having a processor and memory, the method comprising:

capturing, with the server, click-related data by users with reference to a plurality of advertisements owned by a plurality of advertisers in a keyword market;

calculating, by a marginal value indicator executed by the processor, a marginal value of each of the plurality of advertisers as a function of a plurality of averaged market factors, wherein the marginal value of an additional advertiser is also calculated to result in at least an estimated displaced revenue when at least one advertisement of the additional advertiser for at least one keyword is competitively included in the keyword market; and

deciding, by the server, whether to acquire the additional advertiser within the keyword market based on the marginal value of the additional advertiser as associated with the at least one keyword.

2. The method of claim 1, wherein the plurality of averaged market factors comprise averaged values for at least one selected from the group consisting of bid amounts, total advertising spending, an advertiser-specific quality score, a market page quality, a page bid, or a market maximum bid.

3. The method of claim 2, wherein calculating an average marginal value of each of the plurality of advertisers is further a function of at least one selected from the group consisting of a dispersion of quality score, a dispersion of bids, a total number of advertisers, and a user response by the users.

4. The method of claim 1, wherein the marginal value of the additional advertiser depends on at least one selected from the group consisting of a highest bid price attainable from the additional advertiser for the at least one keyword, a quality score dependant on relevancy of the at least one keyword to the at least one advertisement, and a combination thereof.

5. The method of claim 4, further comprising:

letting V(k) denote a marginal unit value for each keyword (k);

letting V(k, ∂

b, ∂

Q) denote a marginal value of an advertiser where b is an expected bid and Q is an expected quality score of an advertiser; and

calculating V(k), an expected marginal value of the additional advertiser, as determined by ∫

∫

∫

V(k, ∂

b, ∂

Q)∂

b∂

Q through integrating over substantially all possible bid and quality values of the additional advertiser.

6. The method of claim 5, further including change in display depth ∂

d within the calculation of the expected marginal value of the additional advertiser.

7. The method of claim 5, further comprising:

calculating a total value index of the additional advertiser by multiplying a search frequency, f(k), of the keyword times the expected marginal value V(k) of the additional advertiser.

8. The method of claim 4, further comprising:

calculating the quality score of the additional advertiser as a function of at least one selected from the group consisting of a ranking, a click-through-rate (CRT), and a click yield.

9. A method for developing a keyword value index in advertisement markets to enable adoption of advertiser acquisition strategies, the method executed by a server having a processor and memory, the method comprising:

tracking, with the server, click activity by users with reference to a plurality of advertisements owned by a plurality of advertisers in a keyword market, wherein the advertisements are displayed to the users through a set of hierarchal search results pages;

defining, by the server, a plurality of state dimensions with regards to at least one keyword related to at least one advertisement, including at least one selected from the group consisting of a page display depth, a page quality score, and a page bid;

capturing, with the server, a user response and market state changes to the at least one keyword and corresponding advertisement, including data representative of at least a number of clicks and a realized revenue therefrom;

based on the defined state dimensions and the user response, identifying, by the processor, a probability distribution that models changes in at least one state dimension;

calculating, with a value indices builder executed by the processor, value indices for the at least one selected from the group consisting of the page display depth, the page quality score, and the page bid from the probability models; and

aggregating, by the processor, a plurality of the value indices at a level of the at least one keyword for the advertiser that owns the at least one advertisement to adopt an advertiser acquisition strategy.

10. The method of claim 9, wherein the set of hierarchal search results pages further includes a plurality of display pages, together “

results pages.”

11. The method of claim 10, wherein identifying probability distribution that models changes in at least one state dimension comprises letting ∂

R_k,t(∂

d, ∂

PQ, ∂

b) and ∂

C_k,t(∂

d, ∂

PQ, ∂

b) comprise, respectively, revenue and click state-response changes between t and t+1 in time, wherein ∂

R_k and ∂

C_k comprise response probabilities based on market changes, including ∂

d (change in page depth), ∂

PQ (change in page quality), and ∂

b (change in page bid).

12. The method of claim 11, further comprising:

calculating a marginal value of each of the plurality of advertisers as a function of a plurality of averaged market factors, wherein the marginal value of an additional advertiser is also calculated to result at least an estimated displaced revenue if at least one advertisement of the additional advertiser for at least one keyword were competitively included in the keyword market;

wherein the marginal value of the additional advertiser depends on at least one selected from the group consisting of a highest bid price attainable from the additional advertiser for the at least one keyword, a page quality score dependant on relevancy of the at least one keyword to the at least one advertisement, and a combination thereof.

13. The method of claim 12, wherein the plurality of averaged market factors comprise averaged values for at least one selected from the group consisting of bid amounts, total advertising spending, an advertiser-specific quality score, a market page quality, a page bid, and a market maximum bid.

14. The method of claim 12, further comprising:

letting V(k) denote a marginal unit value for each keyword (k);

letting V(k, ∂

d, ∂

PQ, ∂

b) denote a marginal value of an advertiser where d is an expected display depth, PQ is an expected page quality score, and b is an expected page bid; and

calculating V(k), an expected marginal value of the additional advertiser, as determined by ∫

∫

∫

V(k, ∂

d, ∂

PQ, ∂

d)∂

d∂

PQ∂

b through integrating over substantially all possible bid amounts and page quality values of the additional advertiser.

15. The method of claim 14, further comprising:

determining a total expected marginal value by multiplying an expected frequency of searches for the keyword, f(k), by the marginal value of the additional advertiser V(k).

16. The method of claim 11, wherein calculating the value indices comprises taking a mean of a revenue state-response change (∂

R_k) given by E[∂

R_k] and a dispersion of a revenue state-response change (∂

R_k) given by D[∂

R_k], and calculating f(E[∂

R_k], D[∂

R_k]), a function of the mean and dispersion revenue state-response changes.

17. The method of claim 16, further comprising:

detecting a change in the user response; and

updating the expected page bid, the expected page quality score, and the dispersion in response to the detected change.

18. The method of claim 16, wherein the dispersion D[∂

R_k] is estimated with at least one selected from the group consisting of a GINI index and another variance.

19. The method of claim 9, wherein aggregating the plurality of the value indices is performed on at least a subset of all the keywords owned by the advertiser.

20. A method for valuating an advertiser and associated keywords considered for acquisition, the method executed by a server having a processor and memory, the method comprising:

tracking, with the server, click activity by users with reference to a plurality of advertisements owned by a plurality of advertisers in a keyword market;

identifying, by the server, a plurality of keywords as a set of relevant keywords for a particular advertiser of the plurality of advertisers within the keyword market;

identifying, by the processor, a subset (L) of the set of relevant keywords as those keywords that the advertiser can efficiently track;

calculating, by a marginal value indicator executed by the processor, a marginal value of each of the keywords within the subset L of keywords for the particular advertiser, wherein each marginal value is based on at least an estimated displaced revenue when at least one advertisement of the particular advertiser for each keyword is competitively included in the keyword market; and

deciding, by the server, whether to acquire at least some of the subset L of keywords of the set of relevant keywords for the particular advertiser in the keyword market based on the marginal values of each respective keyword.

21. The method of claim 20, wherein each incremental marginal value depends on at least one selected from the group consisting of a highest bid price attainable from the particular advertiser for each respective keyword within the subset L of keywords, a quality score of the particular advertiser dependant on relevancy of each respective keyword to the at least one advertisement, and a combination thereof.

22. The method of claim 21, further comprising:

letting V(k) denote a marginal unit value for each keyword (k);

letting V(k, ∂

b, ∂

Q) denote a marginal value of an advertiser where b is an expected bid and Q is an expected quality score of an advertiser in relation to each keyword (k); and

calculating V(k), an expected marginal value of each keyword, as determined by ∫

∫

∫

V(k, ∂

b, ∂

Q)∂

b∂

Q through integrating over substantially all possible bid and quality values of the additional advertiser.

23. The method of claim 22, further including change in display depth ∂

d within the calculation of the expected marginal value of each keyword.

24. The method of claim 22, further comprising:

letting y_i,k comprise a decision variable for whether each keyword (k) of the subset L of the relevant keywords is to be pursued for inclusion in the keyword market; and

executing the decision step of whether to pursue at least some of the subset L of the keywords by solving;

25. The method of claim 22, further comprising:

calculating the quality score as a function of at least one selected from the group consisting of a ranking, a click-through-rate (CRT), and a click yield.