The eighth School of Programmatic event, hosted by OpenX, focused on the evolution of header bidding.
The event kicked off with a presentation from Tegdeep Kondal, product director for header bidding at OpenX, which gave a high-level analysis of header bidding, including how it came about and how it works, as well as understanding and dealing with header bidding’s inherent challenges, the value of header bidding containers, and a forward look into the innovations and breakthroughs. The presentation was followed by a panel discussion between Kondal, Barbara Agus, director of programmatic EMEA and LATAM, Time Inc., and Dave Reed, managing director EMEA, MediaMath, and was moderated by ExchangeWire. It explored header bidding further from the buy- and the sell-side.
Are buyers and sellers connecting through header bidding?
Barbara Agus from Time Inc., said header bidding is giving buyers the opportunity to access all the relevant users on a given property. This is a fantastic result; however, as a publisher, she is often asked whether she can see real changes in the way buyers are approaching inventory since starting to implement header bidding solutions. According to Agus, this is sometimes the case. At Time Inc., they are forecasting between a 10-20% revenue uplift. But publishers in the industry are often asking how buyers and sellers should be connected through header bidding, with concerns that a lack of understanding from the other side of the fence, for both buyers and sellers, could create confusion: HB might create a distorted and inflated vision of the volume of impressions available in any given auction. “We are sometimes accused of trying to make more money out of the same amount of inventory.”
Dave Reed from MediaMath responded to Agus’ concerns, saying header bidding is one of the most effective tools in bringing clients closer together. “It is often assumed that a higher price for the publisher means advertiser loss, and a lower price for the advertiser means publisher loss, but it doesn’t work that way. Advertisers will pay a lot for value, and that’s the world header bidding is starting to bring in.”
Is header bidding driving increased CPMs on the buy side and is that causing advertisers concern? According to Reed, no. He gave the example of a global client, where CPMs increased, but performance increased even more, which, as an advertiser, you might initially balk at. “It’s not a zero sum game”, said Reed, “as you can make the pie bigger for everyone, if you’re efficient.”
If header bidding is driving improved performance and yield for both advertisers and publishers, what’s the problem? Drawing from her previous experience using this tool, Agus points out that it’s the retargeters that seemed to be negatively affected. “We have seen buyer relationships improve significantly, but for retargeters looking for the cheapest impression to drive ROI, we have definitely seen a rapport decline. Before header bidding, those buyers had first access to inventory with no competition. With header bidding in the mix, they suddenly faced competition and, as a result, they wanted to stay out of the wrapper.” According to Agus, it is in dealing with premium advertisers that we can expect most improvements. However, this should not be regarded as a win/lose situation. According to Reed: “As marketers, we need to get beyond campaign-based buying. Consumers don’t live in a campaign world. Always-on marketing that allows programmatic tracking across devices and across platforms should increase CPMs, as buyers should be looking for quality.”
“Header bidding is bringing buyers closer to sellers”, said Reed, who feels it is helping promote a more transparent marketplace, with fewer walled gardens. Agus agreed, saying buyers have the opportunity to appreciate the value of their inventory, helping publishers understand how to price it better.
Are there concerns with header bidding?
Kondal said that, on the platform side, latency is the biggest topic: “I feel there’s a lot that can be done by publishers; but it’s really the job of the platform to optimise on behalf of the end user. At OpenX, we’re looking very seriously at how we can reduce the time it takes to return bids.”
Aside from latency, Kondal worried there isn’t much focus put on the DSP side; and explained how context should be provided to make them understand which bid requests are coming from header bidding and even ask them what type of auction they would like to participate in. “If you give them more context and control it would be a more balanced playing field on both sides.”
Speaking from the DSP side, Reed was, in general, very positive; but he did highlight that one downside is implementation: “If the publisher implements it, and the DSP doesn’t know the context, and is competing against an open auction, you haven’t achieved very much in grokking the waterfall and getting close to the first impression the user sees.” Reed admitted this isn’t a frequent occurrence, but it definitely happens.
HB is a fantastic piece of technology. When it fails, it is not because of the technology itself, but rather because it has not been fully understood.
How should performance be measured?
Kondal believes performance analysis and evaluation should focus on efficiencies, which encompasses many different metrics, noting the true metrics that matter should be fill and timeout rates. Kondal also noted that user experience is a challenge for everyone: “It’s time for the industry to have a metric defined to measure long-term value of the end user. Engagement is a good proxy for quantifying user experience.”
According to Reed, in an ideal world, ‘cost per’ metrics shouldn’t exist. He cited the FT as a publisher starting to play around with time of engagement metrics. Yet Agus worried about the direction of the industry, where the buyer is forcing publishers to become technology publishers, responsible for looking after audience data and losing focus on the value of their properties: “As publishers, content is, and should be, paramount. Targeting off property doesn’t deliver the same engagement and brand experience compared to consuming the same content in the intended environment.”
Choosing the right solution
In Agus’ experience, good technical support is a key priority, when it comes to evaluating header bidding solutions. “Being a global publisher, we need a tech partner with a unique source of demand, that is able to reach a global audience. Otherwise, we would require multiple partners, which is not feasible.” Header bidding is still very new in the industry; with many publishers still finding their feet and trying to gain understanding of how the technology works and how it can be of benefit to their business. Without strong technical support, many of these publishers would be lost.
Where does the future lie?
Aside from bringing more transparency to the market, which is already starting to happen, how will header bidding evolve? According to Kondal, it starts with the application of header bidding in different user cases and then trying to optimise it. He believed there will be a rise of data science and machine learning to allow optimisation at a user level.
Reed doesn’t believe header bidding is the end goal, but is on the path to get us there, thinking that the future will be led by publisher data. He sees a world with more data co-ops between publishers, where they are willing to give up a little bit of their data for the greater good of regaining publisher control. Beyond that, Reed predicted more publishers pairing up with SSPs to bring their own technology in house and completely removing the waterfall from the equation. Agus believes we will soon see the end of the traditional SSP, with more technology companies themselves becoming ad exchanges, but she, along with the rest of the industry, is still trying to figure out the current technology. Kondal agreed, arguing that the technology is still too complex. He believes header bidding is the next wave of programmatic and simplifying it will change the way the industry approaches it.