Amid the cacophony of noise surrounding header bidding at the moment, it's hard to remember a time when ad tech panels would debate the pros and cons of a deterministic rather than probabilistic ID, and the merits of leveraging a 'Device Graph', writes ad tech consultant, Paul Gubbins (pictured below), exclusively for ExchangeWire. Today, we are more likely to be mooting client- versus server-side header bidding and all longing for the utopia of sell-side vendor interoperability.

While the innovations of header bidding are to be applauded, as they come at a pivotal time for most publishers, I can't but help think we have lost sight a little when it comes to ID management and how we are building buy- and sell-side products for a world post the cookie and pre the private, walled-garden, proprietary token/ID.

Throughout the header tag and wrapper arms race that has played out over the last 24 months, enterprise vendors, telcos, and those at the heart of ad decisioning have been picking up cross-device assets (Oracle and Crosswise, Telenor and TapAd, Flashtalking and Device 9, etc.) as they understand the ID management challenges that will surround businesses, that have historically had product architected around the 'cookie'.

The fundamentals of most marketing plans, like reach, frequency, and attribution, are all dependent on the ability to track a persistent signal, regardless of the line item from which it originated. However, in a world of fragmented IDs, this basic practice, that has underpinned the digital ad budgets of yesteryear, is going to become increasingly difficult to execute. We, as an industry, talk about AI driving smarter decisioning, and bidding algorithms becoming bespoke, based upon programmable instruction. However, in order to make these things happen, we need to feed event data to the machines. And this is going to become increasingly difficult to extract, as data portability becomes harder to action, due to the blocking or passing of non-uniformed IDs, dependent on the environment of their originating impression.

Paul Gubbins, Ad Tech Consultant

There has been much written recently about cookie synching and match rates dependent on a server- versus client-side implementation of header bidding and it feels like we are continuing to innovate around an ID that has a limited shelf life. The unification of demand and the practice of holistic yield management is, of course, a fantastic vehicle to drive both bid density and clearing prices for publishers. However, what happens when the impressions are originating from environments that don't support cookies? How will the buy side value? I appreciate some sell-side vendors are supporting in-app header bidding; but many solutions in play appear to be in their infancy and from the debates we see play out at conferences, narrative still seems to be centred around desktop headers.

Why then have we lost sight of the ID management discussion that was getting real traction before header bidding became the default topic of debate? Walled gardens are not getting any smaller; people are not spending less time on mobile devices; fridges, toasters, and cars are not getting any less smart due to the rapid rise of IoT, but it is a subject that, from the outside, can at times look labyrinthal in its complexity.

Considering how important the building and activation of holistic, screen-agnostic IDs will be in 2017, as evidenced by the recent announcement from GroupM re: their mID, maybe it's time we ask if standards are required for both the sell and buy sides to embrace them more confidently. Dependent on the white paper you are reading, best practices for the building of a PII-compliant, probabilistic ID are varied in their approach. The terms vendors use, such as 'precision' and 'recall', to describe the accuracy of their IDs can mean different things dependent on the publisher of said best practices.

We have a common protocol for programmatic in 'OpenRTB', one for video in 'VAST', and one for the building of mobile-rich creatives in 'MRAID'. Is it time we establish a protocol for cross-screen, probabilistic identifiers, considering they will underpin the digital plans of tomorrow?

Many are asking how those on the sell side differentiate, as the 'header tag' is starting to give parity of supply to all and the position in the waterfall is no longer a USP. Could supply-side technology, that can empower a publisher to capture the same yields, if not higher on mobile than desktop through probabilistic identification, get a head start in the differentiation race of 2017?

Header bidding is a great practice for publishers to capture incremental yield; it is also a great way for programmatic buyers to access inventory that has historically been reserved for publishers' direct-sold efforts. However, with more traffic originating from environments that are not supporting cookies, it's increasingly difficult for buyers to migrate from buying screens to true, people-based marketing.

It is going to be really interesting to hear the narrative of ad tech vendors at this year's Mobile World Congress, as the worlds of header bidding and ID management start to converge.