Modifying advertisement bids using predicted advertisement performance
An advertising system provides advertisements to client devices. To select advertisements, the advertising system identifies previously selected advertisements to determine which presentations of the advertisement are still in-flight and have not yet resulted in conversion event. The advertising system predicts total advertising spend based on the in-flight advertisements, and adjusts a paced bid for the advertisement by determining whether the estimated total advertising spend, reflecting predicted in-flight advertisements, is above or below an expected spending to reach a budget for the advertising campaign, which may increase or decrease the paced bid.
1. A method comprising:
receiving an advertising campaign from an advertiser, the advertising campaign designating a candidate advertisement, an advertising budget, and one or more conversion events, each conversion event designating desired actions that cause payment by the advertiser for the candidate advertisement;
receiving an advertising request for an advertisement from a user of an online system;
identifying one or more in-flight candidate advertisements of the advertising campaign previously sent to other users of the online system for which no conversion event of the one or more conversion events has occurred and for which an interaction with the one or more candidate advertisements has occurred;
estimating a value of in-flight spend of the advertising campaign by predicting a likelihood that each of the in-flight candidate advertisements will result in a conversion event of the one or more conversion events, wherein the value is predicted based on a time elapsed since the interaction, and
wherein predicting the likelihood comprises:
generating, in real-time or updating periodically, a decay curve for the advertising campaign by performing a linear regression on recorded historical data associated with the advertising campaign; and
determining a conditional probability of receiving a conversion event associated with the in-flight candidate advertisement from the advertisement campaign, wherein the conditional probability is determined by the decay curve;
determining a paced bid amount based on the advertising budget, a verified spend, and the estimated in-flight spend of the advertising campaign, wherein the verified spend accounts for payments made by the advertiser for completed conversion events; and
adding the candidate advertisement to an auction of advertisements for the advertising request, wherein the auction determines a selected advertisement presented to the user based on the determined paced bid amount.
2. The method of claim 1 , wherein determining the paced bid amount comprises determining a bidding multiplier and multiplying a campaign bid provided by the advertiser by the bidding multiplier.
3. The method of claim 2 , wherein the bidding multiplier is inclusively within the range [0.0-1.0].
4. The method of claim 1 , wherein determining the paced bid amount comprises determining a total spend for the candidate advertisement using the estimated in-flight spend and the verified spend, and comparing the total spend to a spending path for the advertising campaign, wherein the paced bid is increased when the total spend is below the spending path and the paced bid is decreased when the total spend is above the spending path.
5. The method of claim 1 , wherein the in-flight candidate advertisement is associated with at least one action of a user responsive to the candidate advertisement, and the conditional probability determines the conditional probability of the conversion event given the action.
6. The method of claim 1 , wherein the predicted likelihood that an in-flight candidate advertisement will result in a conversion is based on network bandwidth or network quality associated with a client device of the user on which the in-flight candidate advertisement was presented.
7. A system comprising:
a processor; and
a computer-readable storage medium having instructions encoded thereon that when executed by the processor cause the processor to perform steps of:
receiving an advertising campaign from an advertiser, the advertising campaign designating a candidate advertisement, an advertising budget, and one or more conversion events, each conversion event designating desired actions that cause payment by the advertiser for the candidate advertisement;
receiving an advertising request for an advertisement from a user of an online system;
identifying one or more in-flight candidate advertisements of the advertising campaign previously sent to other users of the online system for which no conversion event of the one or more conversion events has occurred and for which an interaction with the one or more candidate advertisements has occurred;
estimating a value of in-flight spend of the advertising campaign by predicting a likelihood that each of the in-flight candidate advertisements will result in a conversion event of the one or more conversion events, wherein the value is predicted based on a time elapsed since the interaction, and
wherein predicting the likelihood comprises:
generating, in real-time or updating periodically, a decay curve for the advertising campaign by performing a linear regression on recorded historical data associated with the advertising campaign; and
determining a conditional probability of receiving a conversion event associated with the in-flight candidate advertisement from the advertisement campaign, wherein the conditional probability is determined by the decay curve;
determining a paced bid amount based on the advertising budget, a verified spend, and the estimated in-flight spend of the advertising campaign, wherein the verified spend accounts for payments made by the advertiser for completed conversion events; and
adding the candidate advertisement to an auction of advertisements for the advertising request, wherein the auction determines a selected advertisement presented to the user based on the determined paced bid amount.
8. The system of claim 7 , wherein determining the paced bid amount comprises determining a bidding multiplier and multiplying a campaign bid provided by the advertiser by the bidding multiplier.
9. The system of claim 8 , wherein the bidding multiplier is inclusively within the range [0.0-1.0].
10. The system of claim 7 , wherein determining the paced bid amount comprises determining a total spend for the candidate advertisement using the estimated in-flight spend and the verified spend, and comparing the total spend to a spending path for the advertising campaign, wherein the paced bid is increased when the total spend is below the spending path and the paced bid is decreased when the total spend is above the spending path.
11. The system of claim 7 , wherein the in-flight candidate advertisement is associated with at least one action of a user responsive to the candidate advertisement, and the conditional probability determines the conditional probability of the conversion event given the action.
12. The system of claim 7 , wherein the predicted likelihood that an in-flight candidate advertisement will result in a conversion is based on network bandwidth or network quality associated with a client device of the user on which the in-flight candidate advertisement was presented.
13. A computer program product comprising a non-transitory computer-readable storage medium having instructions encoded thereon that when executed by a processor cause the processor to perform steps of:
receiving an advertising campaign from an advertiser, the advertising campaign designating a candidate advertisement, an advertising budget, and one or more conversion events, each conversion event designating desired actions that cause payment by the advertiser for the candidate advertisement;
receiving an advertising request for an advertisement from a user of an online system;
identifying one or more in-flight candidate advertisements of the advertising campaign previously sent to other users of the online system for which no conversion event of the one or more conversion events has occurred and for which an interaction with the one or more candidate advertisements has occurred;
estimating a value of in-flight spend of the advertising campaign by predicting a likelihood that each of the in-flight candidate advertisements will result in a conversion event of the one or more conversion events, wherein the value is predicted based on a time elapsed since the interaction, and
wherein predicting the likelihood comprises:
generating, in real-time or updating periodically, a decay curve for the advertising campaign by performing a linear regression on recorded historical data associated with the advertising campaign; and
determining a conditional probability of receiving a conversion event associated with the in-flight candidate advertisement from the advertisement campaign, wherein the conditional probability is determined by the decay curve;
determining a paced bid amount based on the advertising budget, a verified spend, and the estimated in-flight spend of the advertising campaign, wherein the verified spend accounts for payments made by the advertiser for completed conversion events; and
adding the candidate advertisement to an auction of advertisements for the advertising request, wherein the auction determines a selected advertisement presented to the user based on the determined paced bid amount.
14. The computer program product of claim 13 , wherein determining the paced bid amount comprises determining a bidding multiplier and multiplying a campaign bid provided by the advertiser by the bidding multiplier.
15. The computer program product of claim 14 , wherein the bidding multiplier is inclusively within the range [0.0-1.0].
16. The computer program product of claim 13 , wherein determining the paced bid amount comprises determining a total spend for the candidate advertisement using the estimated in-flight spend and the verified spend, and comparing the total spend to a spending path for the advertising campaign, wherein the paced bid is increased when the total spend is below the spending path and the paced bid is decreased when the total spend is above the spending path.
17. The computer program product of claim 13 , wherein the in-flight candidate advertisement is associated with at least one action of a user responsive to the candidate advertisement, and the conditional probability determines the conditional probability of the conversion event given the action.