Header bidding, when RTB (re)makes sense.


Header bidding, when RTB (re)makes sense

As a publisher, when you hear RTB (real time bidding) in online advertising, you probably expect that all advertisers will have an equal chance to bid on your inventory and that the highest bidder will win. This seems logical. But in practice, this is often not the case.

The cascade auction.

Typically, the publisher will first run the ads that are sold live, which are the ones that are supposed to generate the most revenue. Then, if there is still inventory available, the publisher will often rely on Google's DoubleClick for Publishers (DFP) ad server to cascade through a list of ad exchanges, starting with the largest buyers, who are always preferred even if they don't necessarily offer the best bid.

And that's without counting on a little additional trick from Google. DFP will indeed allow its own exchange (AdX) to outbid (even by a small amount) the other players, thus arrogating to itself the quasi guarantee of always overtaking them on the finish line. This phenomenon is facilitated by the sequential nature - in cascade - of the bidding procedure.

Header bidding, for a true RTB.

Header bidding is gradually becoming the solution to abuse of dominant position, by putting everyone on the same level, for a simultaneous and non-sequential auction. This means respecting the strict definition of "real time bidding". A return to the roots in short.

We talk about header bidding because the process is generally done via a piece of JavaScript code inserted in the header of the site. As soon as a page is loaded, the code calls the selected exchanges to invite them to bid, and this before the site's ad server eventually allocates space to direct sales. This makes it possible to offer all the inventory simultaneously to the competing advertisers (note: DFP has introduced a First Look option that now allows advertisers to have access to all the inventory first hand, without necessarily playing the header bidding card).

The header bidding allows to increase the yield of the inventory. Some publishers have seen their CPM increase by 75%, simply by applying the RTB principle to the letter. On mobile, according to Pubmatic's Quarterly Mobile Index, eCPM from header bidding grew by 55% between Q1 2016 and Q1 2017. If you're curious and want to find out who is participating in bidding on a given site, add this Chrome extension and visit the Daily Mail site, for example. Here's what you can observe.

The other side of the coin.

On the DSP side, the replacement of a sequential logic by a true simultaneous auction can pose infrastructure concerns for the smallest players in the market, who have seen the volume of impression offers to be analyzed in real time explode in recent months. We have seen some DSPs react by simply reducing the number of requests taken into consideration, to avoid being overwhelmed by the load, which is compounded by the fact that several requests may concern the same inventory, offered by different exchanges connected to different header bidding partners implemented by the same publisher... It's a real challenge! But where there's a challenge, there's an opportunity: startups are now positioning themselves on the market for analyzing the flow of requests, to separate the wheat from the chaff before the flow of requests overwhelms the DSPs.

An evolution from client-side to server-side.

Critics of header bidding say that the JavaScript code inserted in the header can impact site performance. That said, it is possible to move the call to exchanges from the client to the server by implementing a server-side header bidding formula, which ensures that optimal performance is maintained on the publisher's side. This is the solution used by Ividence to manage RTB on spaces offered in programmatic mode by its publisher partners within their newsletters.

Contact us to find out more.

They trust us