NextMille RTB 2.0: Next Gen Header Bidding

NextMille RTB 2.0: Next Gen Header Bidding

Header Bidding Background

It’s old news that header bidding revolutionized the ad tech industry back in 2015. The daisy chain waterfall model of selling ads was inefficient and losing publishers money and advertisers opportunity.

Header Bidding with its real time auction brought a new level of efficiency, transparency, and democracy to the ad-tech industry. Like most new products, it was better than its predecessor, but not perfect.

Current header bidding improved latency significantly. As you know latency is a vital factor in a website’s performance and any lag is undesired. 

Header Bidding Basics

The way it currently works is when a user comes onto a page, bid requests are sent to all ad partners in the publisher’s stack, multiplied by the number of ad units on the page. For the most sophisticated stacks that number can be significant.

What this often translates into is high latency. Because there are so many requests going out at once it takes longer for the servers to respond. This lag multiplied by thousands of users can equal a chunk of revenue lost for the publisher. 

Header Bidding Upgrade

NextMille RTB 2.0 developed by Next Millennium addressed these concerns with an elegant solution. Instead of bids requests being sent out to each partner for each ad unit, requests are bundled by for each ad partner, this minimizes the number of requests. The max number of requests is the number of ad partners.

For example, a simple page has a sticky top and stick bottom ad unit, and the publisher is partnered with two DSPs. In traditional header bidding that would mean four bid requests any time a user visited the page. With NextMille RTB 2.0, that would be halved, as the two requests per each partner would be sent as one. Most pages have more than two ad units and significantly more partners so the difference between bid requests sent with header bidding and header bidding 2.0 would be significant. All of this is done client side, in their browser. 

NextMille RTB 2.0 + AI

This of course leads to the question of, is it really faster to minimize requests sent? The answer is yes, but only if the bids come back right away. If one of the bid requests lags in getting a response, it holds up all the other bid responses from that ad partner from reporting back. This is similar when you’re ordering many items from one website, and you’re given the option to have it shipped all together or separately. If you choose all together and one item is back ordered, you’ll be waiting a while for some things that were already ready to go. In this analogy, traditional header bidding is the ship separately option, and comes out ahead.

Taking this design consequence into consideration, Next Millennium layered an AI algorithm to mitigate the concern. The AI is constantly optimizing user latency on each site, considering multiple factors, like speed, viewability, monetization, geos, time, and site history, as to whether it’s faster to send bid requests together or separate. Based on the AI evaluation the request will be sent in the manner that reduces latency. 

In this way publishers get to have their cake and eat it too: have sophisticated NextMille RTB 2.0 for latency optimization, and the AI to interject when 2.0 is not the solution. 

Wrap Up 

Header bidding revolutionized and democratized the ad -tech industry. As with all innovations, there are always ways to improve and optimize. Next Millennium has developed NextMille RTB 2.0 to address the latency when sending multiple bid requests. Using AI, it optimizes the way the bids are requested ensuring the least latency for every interaction.

To learn more about how NextMille RTB 2.0 can improve your site’s latency and bottom line, contact us at


Esther Kurtz