You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Overview
Publishers have the difficult task of constantly balancing monetization and user experience. From Prebid’s own documentation on setting auction timeout: “...Determining this value is a delicate balance: too short, and header bidding won’t have enough time to take place; too long, and revenue may go down due to delaying the ad server call to the point where users have left the page. Publishers must determine the value that works for them, considering a balance of factors: average user time on page, direct sellthrough, value of different ad channels, and average user network delay.”
With the number of channels, devices, and ad formats ever expanding in programmatic, creating an optimal auction timeout strategy has become more difficult than ever. Most publishers simply choose a single setting for all their inventory and rarely, if ever, change it. A/B testing can help identify a better timeout setting, but that takes precious time and resources. Several publishers host multiple Prebid configurations to set different timeouts depending on the context of the request. However, context that could be critical to selecting the optimal auction timeout can change throughout the day (e.g. time of day), making even this strategy sub-optimal.
Given this situation, we are proposing a module that would allow a publisher to sign up with an auction timeout provider that would be able to set your auction timeout in real-time based on the context of the request and is continually optimized using machine learning. In addition, if a publisher prefers not to work with a provider, they could use the module to enact a rules-based approach that changes the auction timeout based on the contextual signals of their choice.
Auction Timeout Provider Intended Workflow
Publisher signs up with an auction timeout provider
Auction timeout provider will provide the publisher with a url to retrieve the timeout data
Publisher configures Prebid.js to resolve and enforce the timeout
This feature should be designed to work with Prebid.js currently, but support for Prebid Server, and Prebid SDK could come later.
Goals
Provide a Prebid unified auction timeout solution controlled by the publisher
Allow for an open framework to set auction timeout manually or to work with a provider
Reduce time spent by publishers on manually testing strategies to set more optimal timeout values
Encourage other providers to propose models that effectively optimize a publisher timeout for revenue and user experience
Type of issue
Proposal
Description
Overview
Publishers have the difficult task of constantly balancing monetization and user experience. From Prebid’s own documentation on setting auction timeout: “...Determining this value is a delicate balance: too short, and header bidding won’t have enough time to take place; too long, and revenue may go down due to delaying the ad server call to the point where users have left the page. Publishers must determine the value that works for them, considering a balance of factors: average user time on page, direct sellthrough, value of different ad channels, and average user network delay.”
With the number of channels, devices, and ad formats ever expanding in programmatic, creating an optimal auction timeout strategy has become more difficult than ever. Most publishers simply choose a single setting for all their inventory and rarely, if ever, change it. A/B testing can help identify a better timeout setting, but that takes precious time and resources. Several publishers host multiple Prebid configurations to set different timeouts depending on the context of the request. However, context that could be critical to selecting the optimal auction timeout can change throughout the day (e.g. time of day), making even this strategy sub-optimal.
Given this situation, we are proposing a module that would allow a publisher to sign up with an auction timeout provider that would be able to set your auction timeout in real-time based on the context of the request and is continually optimized using machine learning. In addition, if a publisher prefers not to work with a provider, they could use the module to enact a rules-based approach that changes the auction timeout based on the contextual signals of their choice.
Auction Timeout Provider Intended Workflow
This feature should be designed to work with Prebid.js currently, but support for Prebid Server, and Prebid SDK could come later.
Goals
Other information
I've uploaded the code that implements this rtd module in a branch here: master...openx:timeout-rtd-module
The text was updated successfully, but these errors were encountered: