Header bidding is a way for publishers to solicit real-time bids from a multitude of programmatic partners on each and every piece of inventory by using the page header within a site’s source code. Basically, it’s a convoluted way to bypass the ad server waterfall and ensure inventory is valued fairly in real-time.
It works like this:
- When a user arrives on the page in question, the code calls out to the corresponding demand source (SSP, DSP, exchange) through the browser. (In desktop ad serving, the user’s browser is used as a communication hub for calls and redirects.)
- The demand source launches an auction or auctions for some or all of the page inventory (depending on a publisher’s preferences) within the browser.
- Winning bids are sent to the publisher’s ad server, where they are matched with the demand partner’s proper tags within “price buckets”—basically, giant sets of tags that cover a range of values, typically in $0.10 intervals.
- The publisher ad server determines which source gets the honor of serving up its ad, but rather than static bid ranges, header bidding has introduced dynamic bids from demand sources that more accurately represent what buyers are willing to pay for a specific impression at that time.
The widespread adoption of header bidding has resulted higher CPMs for publishers, greater transparency in the programmatic market, an uptick in display PMPs, and the removal of intermediaries and arbitragers in programmatic trading. However, header bidding does increase the potential for latency by throwing auctions before the page loads. The auctions also weigh heavy on the browser and usurp a great deal of data, which can become a drawback on the mobile web.