The Header Bidding Advantage

by Manny Balbin, VP of Operations
October 12th, 2015

There’s been a lot of buzz lately about the rise of header bidding that has publishers large and small alike trying to grasp the implications and benefits. The header bidding phenomena is generally touted as a programmatic efficiency play, but in reality it goes far beyond that in affecting current challenges publishers face today and their end to end strategies to generate their revenue throughput.

By no means is header bidding simple and to some it can even be daunting.  However, it is a powerful method that is best positioned to unlock the true value of publishers’ media. It provides a huge advantage to publishers and it can be used to address challenges around long term revenue optimization including tackling page speed issues. Maybe its greatest value is the potential to grant the fullest data sets, those that tie traffic acquisition, page speed optimization, and demand partner yield efficiency, the ability to help publishers maximize revenue generation.

What is header bidding? Header bidding is the advanced technique publishers can use to allow for multiple programmatic demand partners to evaluate inventory by placing javascript code on page and send their bids via key-value pairs along with the ad server requests. The bids are then placed in competition with the auction(s) within the publisher’s ad server. Publishers then manage these bids using waterfalls or flat auctions, with the goal of inducing higher bid competition against the publisher’s SSP and even their own direct sales efforts. Here’s how it works:

The Header Bidding Advantage

The expected result should produce an uplift in overall eCPMs and revenue for the publisher’s inventory. The key here is that bidders are no longer limited to work within the confines of an ad server and the integrated SSP itself (or its blackbox nature), but rely on direct relationships with the publishers in hopes of achieving higher yield efficiency. The creation of these direct relationships is key to the advantages of header bidding.

But the advantages are not limited to yield uplift. It is important to discuss header bidding “frameworks,” beyond just header bidding vendors and their revenue value proposition.  

Since header bidding is code intensive on the client side, it requires being embedded deeply into publishers’ existing frameworks. A good header bidding framework can provide long-term publisher gains, and can be flexible to include solutions around traffic and bid analytics, as well as features that help minimize page latency: solutions that all publishers need to address difficult challenges they face today.

Publishers Grapple with Finding the True Value of their Inventory

Publishers want to provide the best content for consumption and user experience for their visitors across all screens. Quality content and quality user experience drive quality audiences, all of which equate to the quality of the publisher’s product. Ultimately, this quality is what translates into brand equity and value for advertisers.

Publishers today grapple with the increasing challenges that reduce the overall value of their media: Pressures from advertisers to provide higher viewability numbers, the recent rise in ad blocking, the tectonic shift to mobile web that limits audience targeting, to name a few. These challenges force publishers to ensure that their site/content designs adhere to specifications of responsive design, the limits of number of ads on any given page, or even the length of time an ad is served to be viewable.  

Publishers’ product and operations teams manage these client-side executions and as you may be able to surmise, concerns around latency, data security, and code bloat require considerable resources to execute seamlessly. More than ever, the code of web publishing requires intensive development resources and that has put it into the forefront of publisher success as much as, if not more than, the content and traffic being generated.

Revenue generation and yield management is no longer limited to the traditional publisher teams responsible. Now more than ever, it is crucial that publishers have their ad/revenue ops teams be in sync with their content, product, technology/development, and sales teams (for those who have them), and even some of their choice partners around a holistic approach to ensure that their sites’ page load speed is optimized to produce the best user experience and ad rendering. It’s even more important now for publishers to align their productization and operations efforts with a partner whose total value proposition extends beyond that of bringing revenue from the indirect channels.

It is worth highlighting that at the heart of all of this is the importance of page performance or, in other words, speed.

Risk-Reward Balancing Act

Speed is of the utmost importance for publishers with regard to their overall content production and revenue generation efforts. Today, publishers find themselves tightly balancing the risks associated with adding header bidding partners into their technology stack against the rewards of higher yield output.  

Large and well-established publisher brands have spent years and considerable resources to develop their homegrown content and advertising frameworks. This means more code on top of more code on top of more code. One thing is for certain: more code on page means more latency. Page-load latency affects ad viewability and performance.  If publishers want to maximize their advertising dollars they must optimize their code bloat on page. Let’s not forget that there exist much longer term gains with respect to user retention or return visitors from the high quality user experience of  fast page load speeds. I tip my hat to how  Condé Nast recently cut their page load time on GQ by 80 percent.  Note the acknowledgment here that page latency is not limited to 3rd-party code that publishers add to their stack. It also includes the code for all content and media on the page, especially large uncompressed images or auto-play videos with heavy payloads that could block all other code on page.

Two notorious culprits for page latency that publishers need to be aware of are the use of passback tags and the cookie-matching (audience data) that happens before or during the ad server selection process. These really highlight the speed of your demand or data partners’ matching and decisioning process before coming back to the page with a response, and is highly variable between the vendors or platforms themselves. If you have Ghostery installed on your browser as an add-on, you may see the plethora of ad tech providers that contribute heavily to the slow load times of the page.

Publishers can utilize their header bidding framework to mitigate some of these latency sources. Header bidding providers can now make their matching decisions outside of the ad server during page load, as opposed to the traditional ad network/SSP partners that make their matching decisions in the publisher’s ad server. Publishers can then set all 3rd-party requests against a set timeout, which they can manage and monitor against measured overall page load latency. Over time, and if armed with latency reporting data cross-referenced against yield output per partner, publishers can adjust timeout thresholds or optimize code to give higher priority to the best performing partners. This technique significantly reduces impression waste from the expected norms of 10-20% (sometimes more) allowable discrepancies from 3rd-party ad serving/matching decisioning.

On top of this, the revenue risk requires some more due diligence from an analytics perspective. Accurate revenue and inventory forecasting is crucial to mitigating that.

Forecasting, What’s That?

Forecasting: it’s not an exact science. Some historical data can be relied on, but publishers’ traffic volume is always subject to overwhelming factors like Google updates or mobile shift trends. Publishers must be armed with the correct tools to monitor and evaluate the revenue impact of all monetization partners. Publisher’s BI teams are essential in being as accurate as possible in forecasting avails and allocating the correct volumes for all partnerships, balanced against direct sales when appropriate.

There are a select few publishers that have the technology and analytics to use bidding data from header bidders to understand the performance of their inventory and optimize to price effectively against it. Historical bid information is crucial data to allow publishers to balance net-pay-bidders against 2nd price auction-bidders, or average-eCPM bidders. Utilizing historical bid data (you may need a data warehouse) will allow publishers to maximize their earning potential in as real time as possible. The balancing act done here is one of volume versus eCPM to maximize gross dollars over time.

Publishers with no direct sales teams have the luxury of flattening out their yield management strategy and letting the best bidder win. Using passbacks, some publishers still use the waterfall tiered approach in an attempt to eke out higher yields. What may hurt them in the long run is that the latency of ad serving for those bidders will hurt the performance of those advertisers’ campaigns, putting the return percentage of those dollars at risk for the inventory.

Beyond bid analytics, publishers need to consider page analytics and tie in traffic acquisition spend with ROI from the yield throughput. Header bidding frameworks on the client side could allow this type of data to be sent to an analytics consumer (GA, data warehouse, etc), which publishers could then leverage to see dimensions that relate pageviews, impressions, and latency data and allow them to make the most informed decisions on optimizing their header bidding partners on their pages.

Do it, and Do it Well

The huge advantage I mentioned in the beginning can be summed up in one statement: header bidding done right allows publishers to take back control of their page speed and their revenue generation.

A publisher’s path to success in maximizing revenue generation with header bidding requires the following basics:

Test, test, test: In the realm of header bidding, not all demand partners were created equal. Publishers must evaluate the true value proposition of each header bidding partner. Short-term monetary gains are not conducive to the long-term play that publishers need to survive.  How many header bidding partners is too many? Your page load speeds combined with the revenue output should help determine the answer. Continual testing and monitoring is required to ensure that you are maximizing the revenue potential against the risk-rewards from header bidding partners’ added latency to the page.

Speed, speed, speed: Header bidding should allow publishers to eliminate partners that use passbacks or match decisioning in ad server, which results in some earned page speed. Setting timeout thresholds so as not to block your page load and ad render speeds is a no brainer for any publisher CTO. One neat thing about a feature-packed publisher header bidding framework is that it should allow to gate not just header bidders, but also any and all 3rd-party JS on the page to minimize the latency effects of code bloat.

Data-Driven Code Optimization: Tie in yield analytics with latency analytics, if available, to help make informed decisions on prioritizing partners’ code. It really does matter what order every header bidder’s code is placed in the “stack.” Don’t believe anyone who says all async JS should load at the same time. It is advantageous to understand the difference between blocking and non-blocking partners and how each contribute to the overall load times of the page, so correlating dollars and milliseconds is a best practice.  

Don’t forget to look at your own content: consider the varying content elements on the page that also suck up browser resources. These affect the header bidders who come back to the page with (quick?) responses but have to wait in the queue longer for browser resources to free up.  One caveat here is that perceived latency is browser latency and not just network latency (another discussion for another time). Here is where the publishers’ product and development teams come front and center to clearly identify the real culprits of the perceived latency, and provide the best path of success to put code (hops) and content (videos, fancy heavy stuff) on a diet and maximize revenue earned.

Publishers come in all shapes and sizes. Their respective business models will dictate whether or not utilizing header bidding is a tangible path forward. Most likely the current programmatic efficiencies achieved in the ad server + SSP can be complemented by additional programmatic efficiencies from header bidding, or it can be completely replaced by it.  

There are many questions to be had about header bidding in general. The discussions above only cover some of these questions. One thing is for sure though, header bidding allows much more control for publishers on the client side and allows for flexibility in utilizing more features to incorporate into the ad stack. Want to target beyond just audiences and layer in traffic and URL targeting? Correlate bidder latency against yield optimization? Do you want return users from optimized UX gained from page speed optimization? Then header bidding is definitely for you.

__

Manny Balbin is the VP of Operations at Yieldbot.  He is the foremost expert in Header Bidding having lead over 700 integrations. He leads Publisher Integrations and Account Management at Yieldbot, and is on the forefront of providing the best header bidding solutions to Yieldbot’s premium publisher partners.

Contact Us

Download Case Study