System and method for recommending advertisement placements online in a real-time bidding environment

• US 10,002,368 B1
• Filed: 03/14/2013
• Issued: 06/19/2018
• Est. Priority Date: 04/06/2012
• Status: Active Grant

• Current Assignee: Valassis Digital Corp. (MX Holdings One LLC)
• Original Assignee: MaxPoint Interactive (MX Holdings One LLC)
• Inventors: Els, Michael, Postelnik, Igor
• Primary Examiner(s): Stoltenberg, David J
• Application Number:
  US13/831,252
• Time in Patent Office: 1,923 Days
• CPC Class Codes: G06Q 30/0243 Comparative campaigns
• CPC Class Codes: G06Q 30/0244 Optimization
• CPC Class Codes: G06Q 30/0249 based upon budgets or funds
• CPC Class Codes: G06Q 30/0254 based on statistics
• CPC Class Codes: G06Q 30/0275 Auctions

Abstract

A method and system for recommending advertisement placements based on scoring is disclosed. According to one embodiment, a computer-implemented method comprises receiving a real-time bidding (RTB) request for placing an online advertisement campaign. For each of a plurality of advertisement placements, a performance score is determined based on an estimated feedback parameter. The estimated feedback parameter is calculated from observed performance of the online advertisement campaign and similarity measures of other online advertisement campaigns. A first advertisement placement having a higher performance score is given more weight than a second advertisement placement having a lower performance score. A set of advertisement placements having their performance scores equal to or greater than the target rating is selected from the plurality of advertisement placements and provided for advertisement placements.

Claims

1. A computer-implemented method for improving automated real-time bidding for a particular campaign in an online network-based auction system by automatically identifying and prioritizing a group of high-performing online placements from a plurality of online placements, the method comprising:

calculating a similarity measure for each of a plurality of other campaigns, indicating a measure of similarity between each other campaign and the particular campaign, based on a measure of historical performance for each respective campaign with respect to a frequency of target action performance in response to campaign-related impressions delivered via one or more selected online placements;

selecting one or more of the other campaigns based on the calculated similarity measures of the respective campaigns;

scoring, by a bidding system, each of a plurality of online placements by executing a scoring algorithm including;

submitting a plurality of real-time bids for a particular online placement over time;

wherein at least a portion of the submitted bids comprise winning bids for which an impression is delivered via the particular online placement to an internet-connected device of a respective user;

monitoring user actions related to impressions delivered via the particular online placement, and generating observed performance data based on the monitored user actions, including determining whether an actual performance of the particular online placement meets a target performance threshold for online placement performance by;

determining a number of the impressions delivered via the particular online placement;

determining whether a particular user action is performed in association with each impression delivered via the particular online placement;

determining an actual number of the user action performances in association with the impressions delivered via the particular online placement;

accessing confidence level data defining, for the target performance threshold for online placement performance, a mapping between (a) a target number of user action performances and (b) a number of impressions delivered, wherein the target number of user action performances mapped to each reference number of impressions delivered indicates a number of user action performances required to ensure a defined probabilistic likelihood that the number of user action performances meets or exceeds the target performance threshold;

determining, based on the confidence level data, the target number of user action performances mapped to the actual number of the impressions delivered via the particular online placement, the target number of user action performances indicating the number of user action performances required to provide the defined probabilistic likelihood of meeting or exceeding the target performance threshold;

comparing the actual number of the user action performances with the target number of user action performances; and

in response to determining that the actual number of the user action performances meets or exceeds the target number of user action performances, selecting the particular online placement for further use in the particular campaign;

obtaining related-campaign performance data regarding the performance of the particular online placement for the selected one or more other campaigns;

calculating a performance score for the particular online placement based at least on (a) the observed performance data regarding the performance of the particular online placement for the particular campaign and (b) the related-campaign performance data regarding performance of the particular online placement for the one or more other campaigns;

receiving, at the bidding system, from a real-time online bidding exchange via a communications network, a series of bid requests, each identifying an online placement defined in digital content being loaded or rendered by an internet-connected device;

for each received bid request, executing, by the bidding system, an automated real-time bidding algorithm in real-time during the loading or rendering of the respective digital content, the automated real-time bidding algorithm including;

identifying, based on information contained in the bid request, the respective online placement;

determining the calculated performance score for the respective online placement;

determining whether to submit a real-time bid based at least on the calculated performance score for the respective online placement; and

in response to determining to submit a bid, submitting the bid, including a determined bid price, to the real-time online bidding exchange.

2. The computer-implemented method of claim 1, further comprising:

grouping a subset of the plurality of online placements based on their performance; and

calculating the performance score for each of the plurality of online placements based at least on performance scores of other online placements belonging to different groups.

3. The computer-implemented method of claim 2, wherein multiple groups of models selected from the group consisting of a collaborative filtering model, a popularity model, a related model, a speculative seen model, and a speculative unseen model are combined to provide an overall campaign recommendation model.

4. The computer-implemented method of claim 1, further comprising automatically updating the performance scores of the plurality of online placements after delivering additional impressions via the online placements, monitoring additional user actions related to the additional delivered impressions, and generating additional observed performance data based on the monitored additional user actions.

5. The computer-implemented method of claim 1, further comprising comparing user traffic of unseen online placements with the performance of known online placements.

6. The computer-implemented method of claim 1, wherein calculating the performance score for each particular online placement comprises:

calculating an estimated feedback parameter based at least on (a) the observed performance data and (b) the related-campaign performance data regarding performance of the particular online placement, wherein the estimated feedback parameter is calculated using a variance of an estimable distribution, and

calculating the performance score for the particular online placement based on the estimated feedback parameter for the particular online placement.

7. The computer-implemented method of claim 6, wherein the estimable distribution is an estimable probability distribution.

8. The computer-implemented method of claim 1, wherein calculating the performance score for each particular online placement comprises:

determining a click-through-rate (CTR) for the particular online placement based at least on (a) the observed performance data and (b) the related-campaign performance data regarding performance of the particular online placement, and

calculating the performance score for the particular online placement based on the determined CTR for the particular online placement.

17. The computer-implemented method of claim 1, further comprising:

selecting a subset of the plurality of online placements based on the calculated performance scores for the online placements; and

wherein determining whether to submit a real-time bid for a particular bid request based at least on the calculated performance score for the respective online placement comprises determining whether to submit a real-time bid based at least on whether the placement identified in the particular bid request is in the selected subset of online placements.

18. The computer-implemented method of claim 1, wherein determining whether to submit a real-time bid for a particular bid request based at least on the calculated performance score for the respective online placement comprises comparing the performance score for the placement identified in the particular bid request to a defined threshold value, and determining to submit a bid only of the performance score is equal to or greater than the defined threshold value.

19. The computer-implemented method of claim 18, wherein the define threshold value comprises a dynamically adjusted pacing threshold value.

20. The computer-implemented method of claim 1, wherein the particular user action associated with each delivered impressions comprises a user click.

21. The computer-implemented method of claim 1, wherein the mathematical function between (i) the number of user action performances and (ii) the number of impressions delivered comprises a linear function.

9. A bidding system configured to provide improved automated real-time bidding for a particular campaign in an online network-based auction system by automatically identifying and prioritizing a group of high-performing online placements from a plurality of online placements, the bidding system comprising:

at least one processor; and

non-transitory computer readable medium having stored thereon computer-readable instructions, which instructions when executed by the at least one processor cause the at least one processor to;

calculate a similarity measure for each of a plurality of other campaigns, indicating a measure of similarity between each other campaign and the particular campaign, based on a measure of historical performance for each respective campaign with respect to a frequency of target action performance in response to campaign-related impressions delivered via one or more selected online placements;

select one or more of the other campaigns based on the calculated similarity measures of the respective campaigns;

determine whether to select each of a plurality of online placements for use with the particular campaign by executing a placement selection algorithm including;

submitting a plurality of real-time bids for a particular online placement over time;

wherein at least a portion of the submitted bids comprise winning bids for which an impression is delivered via the particular online placement to an internet-connected device of a respective user;

monitoring user actions related to impressions delivered via the particular online placement,

determining a number of the impressions delivered via the particular online placement;

determining whether a particular user action is performed in association with each impression delivered via the particular online placement;

determining an actual number of the user action performances in association with the impressions delivered via the particular online placement;

determining a target number of user action performances based on (a) the actual number of the impressions delivered via the particular online placement and (b) a mathematical function between (i) a number of impressions delivered and (ii) a number of user action performances required to ensure a defined probabilistic likelihood that the number of user action performances meets or exceeds a predefined target performance level for the particular online placement;

comparing the actual number of the user action performances with the target number of user action performances; and

determining to select the particular online placements for use with the particular campaign in response to determining that the actual number of the user action performances meets or exceeds the target number of user action performances;

receive from a real-time online bidding exchange, via a communications network, a series of bid requests, each bid request identifying an online placement defined in digital content being loaded or rendered by an internet-connected device;

for each received bid request, execute an automated real-time bidding algorithm in real-time during the loading or rendering of the respective digital content, the automated real-time bidding algorithm including;

identifying, based on information contained in the bid request, the respective online placement;

determining whether the respective online placement is selected for the particular campaign;

if the respective online placement is selected for the particular campaign, determining whether to submit a real-time bid based on one or more input variable; and

in response to determining to submit a bid, submitting the bid, including a determined bid price, to the real-time online bidding exchange.

10. The bidding system of claim 9, wherein the operations further comprise:

grouping a subset of the plurality of online placements based on their performance; and

calculating the performance score for each of the plurality of online placements based at least on performance scores of other online placements belonging to different groups.

11. The bidding system of claim 10, wherein the different groups include one or more of a collaborative filtering model, a popularity model, a related model, a speculative seen model, or a speculative unseen model.

12. The bidding system of claim 9, wherein the performance score for each of the plurality of online placements is determined based on a determined number of impressions served in association with the respective online placement.

13. The bidding system of claim 9, wherein the computer-readable instructions are further executable to automatically update the performance scores of the plurality of online placements after delivering additional impressions via the online placements, monitoring additional user actions related to the additional delivered impressions, and generating additional observed performance data based on the monitored additional user actions.

14. The bidding system of claim 9, wherein the operations comprise comparing user traffic of unseen online placements with the performance of known online placements.

15. The bidding system of claim 9, wherein calculating the performance score for each particular online placement comprises:

calculating an estimated feedback parameter based at least on (a) the observed performance data and (b) the related-campaign performance data regarding performance of the particular online placement, wherein the estimated feedback parameter is calculated by incorporating a variance of an estimable distribution, and

calculating the performance score for the particular online placement based on the estimated feedback parameter for the particular online placement.

16. The bidding system of claim 9, wherein calculating the performance score for each particular online placement comprises:

determining a click-through-rate (CTR) for the particular online placement based at least on (a) the observed performance data and (b) the related-campaign performance data regarding performance of the particular online placement, and

calculating the performance score for the particular online placement based on the determined CTR for the particular online placement.

22. A computer-implemented method for improving automated real-time bidding for a particular campaign in an online network-based auction system by automatically identifying and prioritizing a group of high-performing online placements from a plurality of online placements, the method comprising:

identifying, by a bidding system, a plurality of online placements for consideration for used with the particular campaign;

for each of the plurality of online placements, determining whether to select that online placement for use with the particular campaign by;

submitting a plurality of real-time bids for a particular online placement over time;

wherein at least a portion of the submitted bids comprise winning bids for which an impression is delivered via the particular online placement to an internet-connected device of a respective user;

monitoring user actions related to impressions delivered via the particular online placement,

determining a number of the impressions delivered via the particular online placement;

determining whether a particular user action is performed in association with each impression delivered via the particular online placement;

determining an actual number of the user action performances in association with the impressions delivered via the particular online placement;

determining a target number of user action performances based on (a) the number of the impressions delivered via the particular online placement and (b) a mathematical function between (i) a number of impressions delivered and (ii) a number of user action performances required to ensure a defined probabilistic likelihood that the number of user action performances meets or exceeds a predefined target performance level for the particular online placement; and

comparing the actual number of the user action performances with the target number of user action performances; and

determining to select the particular online placements for use with the particular campaign in response to determining that the actual number of the user action performances meets or exceeds the target number of user action performances;

receiving, at the bidding system, from a real-time online bidding exchange via a communications network, a series of bid requests, each identifying an online placement defined in digital content being loaded or rendered by an internet-connected device;

for each received bid request, executing, by the bidding system, an automated real-time bidding algorithm in real-time during the loading or rendering of the respective digital content, the automated real-time bidding algorithm including;

identifying, based on information contained in the bid request, the respective online placement;

determining whether the respective online placement is selected for the particular campaign;

if the respective online placement is selected for the particular campaign, determining whether to submit a real-time bid based on one or more input variable; and

in response to determining to submit a bid, submitting the bid, including a determined bid price, to the real-time online bidding exchange.

23. A computer-implemented method for improving automated real-time bidding for a particular campaign in an online network-based auction system by automatically identifying and prioritizing a group of high-performing online placements from a plurality of online placements, the method comprising:

identifying, by a bidding system, a plurality of online placements for consideration for used with the particular campaign;

for each of the plurality of online placements, determining whether to select that online placement for use with the particular campaign by;

submitting a plurality of real-time bids for a particular online placement over time;

wherein at least a portion of the submitted bids comprise winning bids for which an impression is delivered via the particular online placement to an internet-connected device of a respective user;

monitoring user actions related to impressions delivered via the particular online placement,

determining a number of the impressions delivered via the particular online placement;

determining whether a particular user action is performed in association with each impression delivered via the particular online placement;

determining a number of the user action performances in association with the impressions delivered via the particular online placement;

determining a confidence level for performance of the particular online placement based on (a) the number of the impressions delivered via the particular online placement, (b) the number of the user action performances, and (c) mathematical functions for each of a plurality of different confidence levels, the mathematical function for each confidence level defining a mathematical relationship between (a) a number of delivered impressions and (b) a number of user action performances required to ensure a defined probabilistic likelihood that the number of user action performances meets or exceeds a predefined target performance level for the particular online placement; and

determining whether to select the particular online placements for use with the particular campaign based on the determined confidence level for performance of the particular online placement;

receiving, at the bidding system, from a real-time online bidding exchange via a communications network, a series of bid requests, each identifying an online placement defined in digital content being loaded or rendered by an internet-connected device;

for each received bid request, executing, by the bidding system, an automated real-time bidding algorithm in real-time during the loading or rendering of the respective digital content, the automated real-time bidding algorithm including;

identifying, based on information contained in the bid request, the respective online placement;

determining whether the respective online placement is selected for the particular campaign;

if the respective online placement is selected for the particular campaign, determining whether to submit a real-time bid based on one or more input variable; and

in response to determining to submit a bid, submitting the bid, including a determined bid price, to the real-time online bidding exchange.