Ad serving with multiple goals using constraint error minimization
The present disclosure describes a system that attempts to reconcile diverse goals and re-cast the goals into something that is quantifiable and optimizable. One way to reconcile diverse goals is by converting these “constraints”—with the huge problems of feasibility—into errors that can be minimized. This disclosure also presents solutions for rate constraints which previously have not been dealt with. The resulting system enables advertisers to dynamically adjust their campaign based on the needs of the moment. Such a system can have advantages in terms of controllability, smoothness, as well as avoiding hard stop conditions that plague the constraint-based approach. In order to achieve this result, solutions are presented for problems of pacing, viewability prediction, and most particularly, error minimization.
1. A method for optimizing content delivery to achieve a plurality of objectives, the method comprising:
identifying two or more objectives to meet in delivering content to a plurality of users;
estimating a first bid price for a first objective of the two or more objectives;
estimating a second bid price for a second objective of the two or more objectives;
determining a first error for the first objective based on a difference between a first objective value of the first objective and a first target value of the first objective, wherein the first objective value is based on the first bid price;
determining a second error for the second objective based on a difference between a second objective value of the second objective and a second target value of the second objective, wherein the second objective value is based on the second bid price;
calculating a multi-objective bid price, associated with the first objective and the second objective, based on the first error and the second error; and
delivering the content to one or more users based on the multi-objective bid price.
2. The method of claim 1 , wherein the two or more objectives comprise viewability and smooth delivery.
3. The method of claim 1 , wherein the two or more objectives comprise viewability.
4. The method of claim 1 , wherein the two or more objectives comprise smooth delivery.
5. The method of claim 1 , wherein the two or more objectives comprise demographic in-target rate.
6. The method of claim 1 , comprising using a weighting scheme to assign a first weighting factor to each objective of the two or more objectives.
7. The method of claim 1 , wherein calculating the multi-objective bid price comprises combining bid prices for each objective of the two or more objectives.
8. The method of claim 1 , wherein at least one of the first error or the second error is modified using a penalty function.
9. The method of claim 1 , comprising
identifying completion of the first objective; and
changing a weighting factor of the first objective.
10. The method of claim 1 , comprising de-weighting the first objective responsive to determining that the first objective has been successful in reaching the first target value.
11. The method of claim 1 , comprising setting at least one of the first error or the second error based on a user-defined factor.
12. A method, comprising:
determining that pre-bid viewability information is unavailable;
responsive to determining that pre-bid viewability information is unavailable;
obtaining a historical viewability rate of content delivering to a user,
determining whether the historical viewability rate of the content is above a minimum threshold, and
responsive to determining that the historical viewability rate of the content is above the minimum threshold, using the historical viewability rate to predict viewability; and
delivering content to one or more users based on the viewability.
13. The method of claim 12 , comprising responsive to determining that the historical viewability rate is below the minimum threshold, using a logic regression of one or more viewability prediction factors.
14. The method of claim 13 , wherein the one or more viewability prediction factors comprise at least one of:
time of day;
operating system;
browser type;
video i/frame; or
player size.
15. The method of claim 12 , comprising selecting the content based on a determination that the content has a first minimum predicted viewability for display.
16. The method of claim 12 , comprising estimating an impression volume and a win probability.
17. The method of claim 12 , comprising using the viewability to provide one or more values to a multi-KPI predictor, wherein the multi-KPI predictor determines whether to buy traffic.
18. A method for optimizing content delivery to achieve a plurality of objectives, the method comprising:
identifying two or more objectives to meet in delivering content to a plurality of users;
estimating a first bid price for a first objective of the two or more objectives, wherein the first objective comprises at least one of viewability or smooth delivery;
estimating a second bid price for a second objective of the two or more objectives, wherein the second objective comprises demographic in-target rate;
determining a first error for the first objective based on a difference between a first objective value of the first objective and a first target value of the first objective, wherein the first objective value is based on the first bid price;
determining a second error for the second objective based on a difference between a second objective value of the second objective and a second target value of the second objective, wherein the second objective value is based on the second bid price;
calculating a multi-objective bid price, associated with the first objective and the second objective, based on the first error and the second error; and
delivering the content to one or more users based on the multi-objective bid price.
19. The method of claim 18 , comprising:
identifying completion of the first objective; and
changing a weighting factor of the first objective.
20. The method of claim 18 , comprising de-weighting the first objective responsive to determining that the first objective has been successful in reaching the first target value.