#OPSPOV: Header Bidding’s Not Dead… Yet

Header Bidding's Not Dead... Yet

All right, everyone – it was fun while it lasted, but it’s time to admit the party’s over. Like the cops shutting down a suburban rager, Google has shown up to end all of our header bidding fun by opening up Dynamic Allocation to those filthy third-party demand sources. So clean up your source codes and tell your demand partners that they don’t have to go home, but they can’t stay in your header.

No, no, no, no – don’t kick out the guests quite yet. Even if industry luminaries swear this is the end of header bidding, rumors of its impending demise are greatly exaggerated. 

Yes, for many publishers and demand sources, header bidding was an end-run around Dynamic Allocation. Before third-party demand sources were stuck with average-bid tags in Doubleclick for Publishers (DFP) that Google could supersede thanks to Dynamic Allocation, which took in real-time bids from Ad Exchange (AdX). Since a demand source’s average bid could be lower than the actual amount a their bidder might pay for an impression, publishers were leaving money on the table.

Dynamic Allocation was potentially costing publishers revenue while earning Google extra because the ad server and exchange were run by the same company. (Such potential anti-competitive practices were why the Justice Department took a long hard look at Google’s acquisition of AdMeld several years ago.) Publishers using DFP – particularly ones with a lot of indirect-sold inventory – complained bitterly about not being able to turn off Dynamic Allocation.

Header bidding basically leveled the playing field for both demand sources and publishers by allowing certain parters to evaluate and submit bids on impressions before the ad call went out – before the page even loaded, as code was entered above the site header (hence the name header bidding). These bids corresponded with more precise tags within the ad server that could actually compete with AdX’s real-time bids. In effect, demand sources had a good sense of publishers’ overall supply of inventory while pubs had a better idea what a variety of advertisers would actually pay. 

So as programmatic consultant Steph Layser noted, it’s kind of amazing that a bunch of ad tech companies – many of them Google rivals – were able to force the mammoth tech company’s hand and open up Dynamic Allocation to third-party demand sources (notably other SSPs – The Rubicon Project and Index Exchange are the launch partners). But while headlines pondering “The End of Header Bidding?” are bound to get clicks, the future of the technology is far less clear.

What’s Actually Happening?

Ambiguity reigned following Google’s announcement, especially if you took a hard look at their own blogpost’s explaination of the development. Buried underneath notice that “First Look” service is now widely available to publishers, Director of Product Managment Jonathan Bellack wrote:

“Exchange bidding in Dynamic Allocation will allow publishers to invite trusted third-party exchanges and SSPs to submit real-time prices using industry-standard RTB calls. These prices will be considered along with bids from the DoubleClick Ad Exchange and the publisher’s reservation campaigns to pick the highest-paying ad.”

This is great news for publishers and advertisers alike (maybe not so much for all the intermediaries, but I’ll get to that). A chief concern on Google’s part was page-load latency caused by header bidding code sending and receiving data. Though many publishers had found work-arounds for this (e.g., timeout strategies), latency was keeping many publishers away from adopting header bidding. Because Dynamic Allocation uses a server-to-server integration, these concerns are lessened – they’re now the same as any auction happening after the ad call is sent out because partners will submit bids in parallel. A whole lot of publishers will be able to get far more accurate evaluations of their inventory.

However, at the moment the approved demand partner crew only includes The Rubicon Project and Index Exchange. This brings up the first of many questions around Google’s announcement (big kudos to some of my best pub resources on these):


  • Which demand partners will Google play with and which will they pooh-pooh? “They are going to be selective on who they all let into the program,” Layser argues. Rubicon and Index are a nice start, but there are a lot of demand sources publishers are working with. How long will it take for other partners to be certified and what kind of hurdles (e.g., technical criteria) will show up in the approval process? And are there some providers – looking your way, AppNexus – that Google will shun just to avoid being in bed with a multi-service competitor?
  • Some demand partners may not want to work with Google. For example, certain demand sources value price encryption to maintain privacy; it’s not clear whether Google would accept this.
  • Will Google leverage a fee per demand partner? Header bidding integrations typically don’t cost a thing. Will Google find a way to add in costs for third parties using its server-to-server integration? A source suggests publishers might be able to negotiate better revenue shares outside of Dynamic Allocation.
  • Will pubs be able to leverage third-party Deal IDs and private marketplaces through Dynamic Allocation? Signs point to no. This is a big deal because display PMPs received a great lift from the ability to insert accurate real-time bids.
  • Is Dynamic Allocation playing mediation platform? Will that tech run  a second-price auction or will original bids come through? As we’ve noted in other pieces, when header bidding wrappers serve as mediation platforms, they typically pick the highest bid to send onto the ad server. So will Dynamic Allocation simply push through the highest bid? This might not be advantageous because sometimes reasons other than price – e.g., fulfilling a PMP agreement – take priority on whom gets the impression.
  • Current header bidding setups enable demand sources to see all (i.e., 100%!) of publisher inventory – will Dynamic Allocation do the same? 
  • How will the Dynamic Allocation shift affect DFP setup as well as implementing floors in AdX? A source notes that header bidding proved highly effective for determining accurate floors in AdX –“If that’s removed (for pubs that choose to move to just this setup) how will they retain the line item granularity that’s allowed for that more effective flooring?”


What Happens Next?

There are plenty of reasons to keep your header bidding arrangements as they are – for the moment. Google’s announcement may slow down new or advanced adoption of the technology as publishers see what Dynamic Allocation will net them before messing with source code.

One source suggested that opening up Dynamic Allocation may simply be the first step in killing off header bidding. But there was always a question of whether header bidding was a means to a better solution – namely secure server-to-server connections between various demand partners and publishers. Many publishers with advanced header-bidding setups are already headed down this route.

As for other effects of this Dynamic Allocation move, look to the buy side. In many AdMonsters header bidding discussions, we’ve mulled over the possibility of advertisers bidding against themselves because they’re running campaigns through multiple buying platforms. Some participants have even suggested that may not be a bad thing – for example, an advertiser using a data set on a specific DSP may get a better idea of data value. 

However, where once the buy side had a great deal of control in driving discounts through programmatic, header bidding has already bumped up CPMs for many publishers and arguably given them added leverage in terms of pricing. Opening up Dynamic Allocation increases both publishers’ advantage and the potential for advertisers to bid against themselves.

“This makes things trickier on the buy side,” Layser said. “ Buyers are going to have to change their behavior, get smarter.”

And smarter buyer behavior will probably lead to a winnowing of the demand source herd, which is for the best. The winners will be the major independent SSPs and exchanges that Google will likely (begrudgingly) partner with. To stay alive in this brave new world, SSPs will need to demonstrate they have unique demand. Because a stable of publishers will desire this demand and header bidding may prove the best way to import accurate bidding, the technology may not go the way of the dinosaur… At all.