Ditching DFP: 3 tips for publishers using header bidding

Despite its increasingly wide adoption, header bidding remains a source of considerable puzzlement. Many publishers regard it warily, fearing that it is too complicated, too hacky and too freighted with risks including latency and data leakage.

But never fear!  Header bidding, while certainly not a piece of cake, is actually a reasonably clear-cut process with proven benefits to publishers.  

Here are three things that every publisher needs to know:

In theory, header bidding is a workaround solution most commonly used by publishers who rely on Google’s DoubleClick For Publishers (DFP) for ad management.  Unlike other ad servers, DFP does not enable publishers to draw RTB demand directly from multiple sources. It only allows bids from AdX to compete against deals and guaranteed inventory.

Google may argue that most demand sources already integrate with AdX,  but AdX is a classic black box. Publishers don’t know whether its auction logic privileges certain demand sources over others. Moreover, there is a cost associated with using AdX as an intermediary – that’s money that should go to the publisher.

The limitations of this closed system are obvious: you would never agree to sell a valuable painting through an auction house that didn’t allow you visibility into competing bids or how it selected the winner. Doing so could mean leaving a lot of money on the table.

Realistically, however, DFP controls overwhelming market share. Until recently, publishers had little recourse but to accept its closed auctions. Header bidding – java script that publishers insert directly onto a page, and which solicits demand from multiple sources before the ad server call – changes that equation. It’s a hack around DFP’s disadvantageous structure.

Recently, Google announced its own answer to header bidding – a product called First Look.  

At first glance, First Look seems a lot like header bidding. Except it’s not. First Look establishes a preferred auction with a price floor. While a useful innovation in some regards, First Look does not resolve the structural limitations of DFP. Publishers are still unable to access multiple demand channels directly, and they are still locked into the black box that is AdX. In effect, First Look leaves you right back where you started.

Some publishers legitimately fear that loading extra script onto their pages will result in latency.  

In fact, if you choose the right code and implement it efficiently, latency ceases to be an issue. Be particularly attentive to whether your header bidding solution initiates a synchronous or asynchronous auction, and determine whether you have managed your integrations correctly. If executed properly, you can increase monetization without degrading user experience.  

Some publishers also worry about data leakage. But header bidding is no riskier in this regard than any other exchange-driven auction.  If you know your partners and understand their practices and policies, you can safeguard your data.  

Because its very purpose is to sustain a transparent and open ecosystem for online advertising, header bidding should be open-source and free. Today, there are multiple public domain header-bidding wrappers available, including prebid.js, which AppNexus designed.

If someone tries to sell you proprietary header bidding code, beware. Some vendors may bias their own demand channels, thus obviating the entire purpose of header bidding.

Depending on their technical proficiency, some publishers may find it necessary to pay for implementation and service. The beauty of open-source code is that it can be supported by multiple actors. Those actors will serve as a check against attempts by any one service partner to tilt the advantage to its own demand sources.  

In the final analysis, header bidding may not be the most elegant solution, but in a world where DFP continues to maintain a closed and opaque auction system, it remains the best option for publishers seeking the fullest and fairest price for their inventory.