Straight Talk About Header Bidding & Containers (part 1 of 3)

Adtech is in the middle of a radically disruptive period of change. This change has been driven by the advent of header bidding and the promise of open and increased competition for ad inventory. But, amidst the frenzy, there are clear warning signals that this radical change is unsustainable and even dangerous for header bidding participants, and our industry in general.

So I thought I’d step out of the raging rapids of change for a minute and try to provide some perspective about where we have come from, where we are now, and how best navigate the dangerous waters ahead.

Let’s first take a look back to assess how we got here. Most radical change starts with a large problem, and so it was with header bidding. Since the beginning of programmatic Google had a lock on the ad tech stack, which unnaturally limited competition by hard bundling Google’s exchange with their newly acquired ubiquitous ad server, DFP, back in 2009. This created an ‘Iron Curtain’ around the programmatic monetization of ad inventory, such that only Google’s ad exchange had exclusive real time bidding access to publisher inventory. To make matters worse, and further hindering competition, Google built their stack so that they retained a ‘last look’ on all publisher inventory, and operated as a ‘black box’ in terms of transparency with regard to revenue shares within their exchange.

This arrangement defied all basic marketplace design principles. At the most basic level, there was a conflict of interest because Google advertisers represent the majority of demand within AdX. As a point of reference, the NY Stock exchange certainly doesn’t operate as the largest buyer within its own exchange, and the idea of seeing the bids from all other participants before bidding (aka “last look”) is not a standard practice. It’s also not consistent with the foundational market principle of fairness and equal access to information for all marketplace participants. 

‘Last look’ was bad for maximizing publisher yield and it was bad the industry, but publishers had little choice or leverage and that’s the way things operated under the Google DFP “M” word for many years. Until a hack was developed to circumvent the lock that Google had on publisher ad inventory. That hack was header bidding.

Header bidding is actually pretty simple. It allows third parties, like ad exchanges and other real time bidders, to submit a bid into DFP via a header tag implementation. The DFP ad server recognizes the third party bid and activates a “line item” within the ad server and allows that line item to compete against all other line items including Google’s ad exchange, AdX (assuming dynamic allocation is activated). In this way, for the very first time, publishers could enable third party (non-Google) monetization partners to bid on their inventory in real time, driving competition with AdX. The results were immediate and dramatic.

Faced with external competition for the first time, Google had to pay 50% or more for publisher inventory, and even then often lost to third party exchanges, like OpenX, that were better able to extract full and fair market value for each publisher impression.

The proliferation of header bidders

Because publishers saw dramatically positive revenue impact from adding a header bidding partner, they made the next logical assumption – more header bidders on the page would result in more revenue. So publishers added more header bidding competition, which resulted in more header bidding companies entering the market. Today, there are more than 20 header bidding companies, and the number of header bidder solutions a publisher deploys can range between five and fifteen. 

This wide scale adoption of header bidding appeared logical from the publisher’s point of view and occurred with incredible speed—mostly over the course of 2016. This rapid change however, did not come about without some initial problems.

Unintended Consequences

As this rapid evolution to multiple header bidders developed in 2016, problems began to surface -- first for publishers and then other participants across the industry, specifically DSPs. Each header bidder a publisher added to their page introduced an additional level of technical risk, one of the most serious being latency, and has ultimately led to the creation of a new, and drastically unsustainable, dynamic for buying and selling ads programmatically.

The rise in redundant header bidders among publishers, many of which are ‘knock off’ ad technology companies offering no unique value, has facilitated the massive growth of redundant ad requests to DSPs and has put a massive strain on buy-side players. But, before diving into each of these consequences, and how they’re impacting publishers’ businesses, I want to take a closer look at the first industry solution that popped up to solve some of the emerging problems around managing multiple header bidders - containers.

Up next week, part two of straight talk about header bidding and containers -- Understanding Containers: what they promise, all the types and how they work.