May 10, 2016

Load DFP Lightning Fast with Prebid

Prebid is introducing a new way of loading GPT and sending GPT requests. Compared to the previous mechanism, the new method has the below advantages.


  • Further reduces latency: the new method loads ads 40-80ms faster than the old method on fast Internet connections, and more (go up to 150ms faster) on slower Internet connections, such as those of mobile cellular. Reasons:

    • The new method requires no need to delay the loading of the GPT library. The GPT library can be loaded upfront.

    • Instantly send out ad server ad requests when bids are ready or timeout is hit. No need to wait until the GPT library finishes loading.

  • Easier implementation and maintenance: The new method requires no change to your pages’ original GPT implementation. It’s a simple code snippet insertion now. This can be done by the ad ops teams within a tag management system.


GPT Instant Load (right) sends GPT ad requests about 100ms faster than the previous method (left):

Prebid Instant Load Benefits

What is GPT Instant Load?

GPT Instant Load leverages the googletag.pubads().disableInitialLoad()call to disable GPT from sending out ad requests out immediately. This call informs GPT to wait until the first googletag.pubads().refresh() is executed. GPT Instant Load triggers the refresh() call when all requested bids are back, or when timeout is hit, whichever happens faster.


  bidsBackHandler: function() {

How to implement it?

Please follow the line by line code example documented here.


Credits to Bart Van Bragt in the Github suggestion.