Myth: Increased Competition In Header Bidding Is The Biggest Contributor To Revenue Gains

The Sell Sider” is a column written for the sell side of the digital media community.

Today’s column is written by Michael Necheporenko, chief technology officer at Roxot.

Many still believe that header bidding increases pricing for publishers’ inventory. This should be one of the major reasons publishers enjoy a 20-30% revenue increase from implementing header bidding.

It is presumed that a demand partner in the header competes for every impression with five to eight other supply-side platforms (SSPs) in real time. It makes sense for an SSP to bid higher if it wants to win in this competitive environment. Otherwise, a demand partner will waste resources on processing billions of ad requests without making any money on the available inventory.

But it appears to be a myth that the increase in competition is what’s driving all revenue lift in header bidding. Based on a recent analysis of header bidding auctions I conducted, about half of the revenue lift experienced by publishers is the result of fewer unfilled impressions, with the other half coming from increased competition for inventory. Both matter equally.

Due to issues on the SSP side and incorrect or unoptimized setups, most of the publisher’s SSPs didn’t compete in every auction. A publisher with at least five bidders in the header can expect to fill 80% of ad requests, meaning that most header auctions will have at least one bid from these five bidders.

However, an average SSP bid only for 40-50% of available inventory. As a result, all bidders rarely competed in a header simultaneously. Often, only two to three bidders competed while others were timed out or did not provide a bid.

My analysis of the incremental value of publisher’s bidders underlined that the competition and bid density in the header is low. On average, 50% of bidder’s revenue is incremental. By incremental revenue I mean the sum of the differences between clearing prices and second-highest bids.

For example, if an SSP wins an auction with a $3 bid and the second-highest bid is $1.50, the winner’s total revenue is $3 but the incremental revenue is $1.50, or 50% of $3.

Thus, the publisher’s setup reminds me of an optimized version of the waterfall – the site fill rate is maximized but there’s no real competition for the inventory. SSPs still dictate the pricing for every impression.

However, this verdict does not have to be final. Publishers should pay more attention to their mix of SSPs and start getting picky, as Paul Bannister mentioned in a recent article.

Understanding the demand sourced from a publisher’s exchange is key to crafting a successful demand mix. Publishers should look for bidders that bring unique demand to the table, such as retargeters and SSPs that are not just listed open exchange resellers but bring additional demand, such as multipublisher private marketplace deals and other value-added demand.

Finally, publishers should work with their individual bidders directly to improve their performance. They should work closely with these bidders and share their data through anonymized bidder names. This will give bidders a clear understanding of how they perform in a stack and what metrics they are losing out on.

Follow Roxot (@roxot_team) and AdExchanger (@adexchanger) on Twitter.

Enjoying this content?

Sign up to be an AdExchanger Member today and get unlimited access to articles like this, plus proprietary data and research, conference discounts, on-demand access to event content, and more!

Join Today!


  1. Interesting article – there are some really smart nuggets of intel in here. I’m not sure ‘waterfall’ feels like the right term to me because it implies staggered access to inventory vs. parallel access to inventory, but quantifying the benefit of a partner specific to fill vs. pricing is a really smart analysis to do for publishers.

    I remember doing similar analysis years back when everyone was integrating with lots of different data providers and didn’t realize the high overlap some had with each other due to sourcing from the same underlying data asset. Kind of this problem in reverse, but one that benefits from the same analytical approach.

  2. Along the lines of Ben Kneen, I’m curious how much of this results from advertisers competing with themselves by engaging multiple DSPs and each DSP bidding on the same impression via multiple exchanges.
    Anyone know how much these problems persist these days?