To realise the goals of programmatic, the ad tech industry has agreed on co-locating exchange and buy side trading technologies in the same physical locations. This has afforded us a latency advantage that allows a buyer to return a bid to an exchange in as little as a few milliseconds.
By offloading this intensive work from the browser to the data center, we have been able to connect to sixty or more buying platforms in parallel, all while incurring a single network request in the browser. This reduces the burden on the web page and allows for more rapid rendering times, without compromising the liquidity advantage afforded by including an ever-increasing number of demand sources.
“Time and Distance. If you’re not optimising for this, you won’t be able to remain competitive. Users are becoming increasingly more demanding of instantaneous digital experiences.”
With the recent uptick in excitement around placing tags in the header of web pages, we have to remember the reasons why programmatic was even able to scale to the place it exists today. It’s important that we not find ourselves in a position where there are so many heavy requests being made on the page that we run up against the law of diminishing returns, especially as it relates to fostering favorable end user experiences. Just as many exchanges and buying platforms have cooperated on the goal of reducing latency, a similar dialogue is ongoing between exchanges. The same technology that we are utilising to conduct low latency transactions with buying platforms can be applied to aggregate liquidity from other exchanges.
For us to move back to a model where we can reduce the touch points on the page (as it was prior to the uptick in header tag bidding), a few key factors need to be considered:
- Trade by trade transparency: The exchange that intermediates requests between buying platforms and third-party exchanges must offer 100% transparency into all bids and cleared transactions, including trade codes explaining the rationale for why an impression cleared with one participant or another.
- Co-location: Exchanges that do not have a physical presence in key liquidity hubs would need to migrate to those data centers in order to participate, since making a request to an external data center would place them at an undesirable latency disadvantage in each auction.
- Pass through signaling models: 100% of bids received in the allotted timeout window must be returned to the ad server. These bids would constitute the result of a third- party exchange’s auction.
- No load fee structure: The third-party exchange would be able to push the full agreed upon net CPM value to the publisher without fees being charged by the liquidity aggregating exchange.
- Decentralised clearing structure: The third-party exchange would continue to reconcile delivery and process payments directly with the publisher.
It’s understandable that all exchanges would ideally like to sit directly on the page. Until recently, this key positioning was only afforded to Google due to the Ad Exchange being embedded within the DFP ad server, which exists directly on the page, and is activated through a header tag call. This highly sought after positioning allows the exchange to not only see all available opportunity, but also to do it in a way that is fully independent of any intermediary. With the framework described above, and closer collaboration between exchanges, we hope to bring some of the best practices forged through working with buying platforms to help reduce page load, without degradation to market participation and yield.