Header bidding has provided a new opportunity for publishers to take control over their programmatic advertising. While for brands and agencies, it promised to make a new whole range of quality inventory options available via real-time bidding, writes Moti Tal (pictured below), co-founder and CTO, Simplaex, in this piece for ExchangeWire.
For every extra lane we add to the RTB motorway, it seems there’s an accompanying pile up somewhere further down the road. So it proves with header bidding, where leaping traffic and bid-request volume have turned into a bottleneck risk to be constantly monitored. For publishers, that means a heavy barrier between the promise of the tech and the reality. You might well want to add every SSP under the sun to your wrapper, but to do so right now could have a far greater negative impact further down the road – one that overshadows any short-term upside on yield. Latency and the effect on mobile data usage of client-side integrations, and cost of server-side header bidding, are just a few such factors to consider.
In short, as the never-ending publisher search for ROI rolls on, our evolution is still far from complete.
As we know all too well by now, ad tech loves an acronym. Header bidding’s firehose volume problem already has its own in the form of SPO, or supply path optimisation. Whether it’s supply, or indeed demand path optimisation (DPO) we’re talking about, the approach is ultimately shortsighted. As we’ve seen time and again, tactics that favour one side of the auction over another are doomed to fail. For instance, in the case of bid caching, they may even prove detrimental to trust and the reputation of the sector as a whole. While solutions that benefit everyone may be harder to execute, that doesn’t mean they don’t exist.
Of course, at the birth of the whole header bidding phenomenon, it’s understandable there were rough edges to the tech. In time, we saw those smoothed out, not least with the launch of Prebid, which added open source, transparent standards for the first time. It also brought a welcome increase in control and sophistication for publishers optimising their yield. In short, all great progress. However, there are still more fundamental challenges around header bidding that we must address for it to keep on growing the market. Of course, header bidding was a big advance over its predecessor, the waterfall. As it turns out, allowing all revenue sources to compete on price is just the first chapter in this story. In actual fact, with the wealth of real-time data now available in the wrapper, many current header bidding integrations are starting to look like pulling out a rusty penknife at a gunfight.
Header Bidding 2.0?
The next stage in our development is not just about looking indiscriminately at price, on a bid-by-bid basis. Instead, we need to apply a far deeper level of understanding to that process. Especially considering the crippling technical cost, duplication, and inaccuracy that currently afflict both sellers and buyers in this process. Long story short: it’s time to move towards header bidding 2.0.
What might that look like? Consider a machine-learning-driven approach that takes into account the user's value relative to each specific buyer's campaign. This approach would result in something far closer to true market value for the publisher because it weighs the relative worth of the user, both within and outside of that specific auction. In addition to factoring in variables for auction type, or whether a DFP bid comes from outside of the main header bidding auction, the list of possible variables that determine the true value of the user is long – and is growing all of the time. Ultimately, we’re talking about finding multiple ways of becoming smarter with the data available in our own bid streams – being more selective in order to optimise the entire programmatic process.
From SPO & DPO to EPO
For all these reasons, and more, header bidding needs to move on from considering just a handful of dimensions – such as country, ad type, size, device etc. – towards true optimisation, based on multiple, real-time factors. That means user-level floor prices, and optimisation based around specific campaign goals and KPIs. Most of the current crop of header bidding integrations don’t account for the true user value to the buyer from the publisher end. Unless you understand whether a bidder is operating in a hyper-local, retargeting, or contextual campaign, you can’t maximise value for either publisher or brand.
What we’re talking about involves building an understanding of both the supply and demand paths, and mapping one to the other – if you like, it’s EPO, or entire path optimisation. By optimising yield for the seller and the buyer simultaneously, this creates a true virtuous circle. Truly an opportunity to be grasped in 2019.