Tern

From Bitnami MediaWiki
Jump to navigation Jump to search

Adroll's Tern proposal provides the same functionality as Google's Turtledove, but allows for multiple organizations to provide this trusted "gatekeeper" service rather than just Google.

Adroll's Tern addresses marketers need for accuracy in the audiences they engage. Tern relies on the Turtledove auction mechanism, that separates current context from audience-based bidding.[1]

Tern is designed to address the following short comings of Turtledove:

  • Clarifying the necessary inputs to participate in an auction
  • Clarifying how to deal with multiple ad formats
  • Reducing networking overhead by streamlining the data which flows through SSPs
  • Eliminating the need for thresholds on interest group sizes
  • Reducing the time to potentially deliver an impression
  • Better supporting dynamic creative and product recommendation use cases
  • Supporting some functionality of third-party tags
  • Further specifying auditability of delivered ad creatives
  • Enabling brand safety for both advertisers and publishers
  • Creating a mechanism for trackability metrics
  • Adding a privateData object that can never escape the browser, but improve bidding signals (among potential other use cases)
  • Specifying a mechanism to support creating interest groups based on publisher browsing behavior
  • Allowing publishers to retain control of auction dynamics, while encouraging second-price auctions
  • Specifying frequency capping and optimization


Impact

Tern, like Sparrow, relies on a trust gatekeeper to operate the auction.

Open Questions

  • If different buyers send their models to different gatekeepers, how many gatekeepers can one publisher query to retrieve auction information?
  • What is the time-delay in providing the data necessary for publishers to optimize their revenue?
  • What is the time-delay in providing the data necessary for marketers to optimize their media spend?


References