Open Bidding (EBDA): Frequently asked questions

open bidding

One of the most important things that happened in the ad tech industry within the last decade was the invention of header bidding. By enabling publishers to auction off impressions one at a time to multiple bidders, header bidding helped improve auction efficiency and maximize yield.

Two things happened as header bidding gained widespread adoption.

First, since header bidding was originally a client-side technology, i.e., the auctions were conducted within the user’s browser—page latency took a hit as publishers crammed more and more bidders in their demand stack for the promise of driving bid competition and revenue lift.

Second, Google started losing its stronghold on the supply chain, as an increasing number of auctions were now moving to header bidding, where it couldn’t cherry-pick impressions and outbid competing demand partners using “last-look privilege” in Dynamic Allocation.

In response, Google announced the Exchange Bidding open beta, also known as Exchange Bidding in Dynamic Allocation (EBDA), and later renamed to simply Google Open Bidding, as a server-side alternative to header bidding. It even dropped its last-look privilege in EBDA beta to level the playing field and gain the trust of demand partners and publishers.

In this post, we’ll cover some questions that publishers frequently ask about Open Bidding, including how it works, its pros and cons, and how it’s different from header bidding.

What is Google Open Bidding?

Each instance of Open Bidding transaction between a publisher and its demand partners is handled by Google Ad Manager using what is called a server-to-server (“S2S”) integration. So, instead of auctions being conducted in the user’s browser as is the case with header bidding, they are now conducted on the servers of the S2S vendor—in this case, Google.

This approach has some distinct advantages, as well as drawbacks, which are covered next. Here’s how the Open Bidding auction works:

(1) An ad request is triggered and passed to Google Ad Manager.

(2) Ad Manager runs a unified auction to determine the best yield for the available inventory.

  • Ad Manager selects the best-trafficked line item to compete.
  • A bid request is sent out to all yield partners (AdX, third-party exchanges, and networks).
  • Yield partners run their own auction and return the most competitive bid to Ad Manager.
  • Ad Manager hosts a unified auction and selects a winner.

(3) In the final step, Ad Manager returns the request to the page and the winner’s ad is displayed on the publisher’s ad space.

What are the benefits of Open Bidding?

Reduced page latency

Open Bidding requires only the GAM script to run, whereas header bidding requires the GAM script in addition to the script for the header auction. Less JavaScript at runtime translates into faster page load and better UX even before the auction starts.

For the auction stage, header bidding relies on the computing power and network resources of the users’ browser (a potential bottleneck on mobile devices and/or slow internet connections), while Open Bidding uses Google’s infrastructure—which makes the auction faster.

According to Google, Open Bidding adds 60ms to Google Ad Manager’s process from start to finish, for a total auction time ~150-300ms. In comparison, according to research conducted by the University of Crete, the median page latency added by header bidding is 600ms.

Simple, unified payments

Most publishers who run their own header bidding stack also have to manage payments from all the demand partners that they are working with.

These demand partners often work on different payment schedules, such as Net-30, Net-45, or Net-90, and sometimes, there are discrepancies in the revenue estimated by the publisher and the demand partner, making the payment process somewhat long and tedious.

In comparison, Google aggregates the payments from multiple demand partners and pays publishers on its usual Net-30 payment schedule.

Less operational overhead

Since Google Ad Manager manages Open Bidding, there is nothing additional to be installed or maintained by the publishers on their website.

When used as an alternative to header bidding, Open Bidding eliminates the need to maintain complex header codes and the numerous GAM line items associated with header bidding.

Also, since Google handles the billing for Open Bidding, the demand partners are billed exactly what they bid and the bid discrepancy issue goes away for publishers.

What are the drawbacks of Open Bidding?

Lower cookie match rate

Cookie matching allows buyers to identify and bid more for users that are most valuable to them, based on past online behavior.

Since header bidding works on the client-side, it allows bidders the chance to match cookies each time a request is made. Cookies can only be read by the domain that sets them, which is why Open Bidding can’t do it the same way.

Instead, cookies have to be synced between Google’s server and participating exchanges. This makes the process less precise and valuable bid opportunities can be lost.

Higher barrier to entry

It’s much easier to get started with header bidding when compared to Open Bidding. Any publisher can download the wrapper code for Prebid, the most popular open source wrapper for header bidding, and start plugging in their demand partners.

Open Bidding on the other hand, is only available to publishers who have their own full AdX account and access to Google Ad Manager 360, the premium version of GAM.

According to Yieldbird, publishers need to be generating at least 90 million monthly impressions to get access to GAM 360. The other way to get access is to work through a Google authorized reseller, however, there is usually an additional cost involved with that.

Limited access to bid-level data

Publishers have a complete view of and control over header bidding bid data because the auctions are taking place on their own website.

In comparison, Open Bidding auction data is a black box. While Google does provide advanced analytics and reporting, publishers don’t have any way of knowing about the individual bids made by demand partners for every auction.

This raw data is extremely useful in making data-driven decisions about which demand partners to drop or retain, configuring optimal timeouts, and making geo-based bidder optimizations.

What are the key differences between Open Bidding and header bidding?

For the purpose of this comparison, we’re highlighting the differences between client-side header bidding and Open Bidding. It’s important to note that there are a lot of other S2S header bidding solutions in the market now as well, for example, Amazon’s Unified Ad Marketplace (UAM).

The only reason header bidding continues to be synonymous with client-side auctions is because that’s how it was first introduced to the market, but that’s certainly not the case anymore.

Open BiddingHeader bidding
Auction typeServer-sideClient-side
Entry requirementPublishers need GAM 360No entry requirements
Technical know-how neededMinimalAdvanced
PaymentsManaged by GoogleSelf/vendor managed
Page latencyLow (~150-300ms)Medium (~600ms)
Cookie match rateLow, reliant on cookie syncHigh, direct cookie matching

Can I run Open Bidding with header bidding?

You can. In fact, many publishers are already doing so. Some actually run client-side header bidding (via Prebid), Google Open Bidding, and Amazon UAM simultaneously. As long as you keep an eye on page latency, adding additional demand should drive better yield performance.

By sending your winning bid from the header bidding auction as a Price Priority line item, you can make GAM serve the highest value bid received from header bidding and Open Bidding auctions.

Although it sounds simple, configuring, maintaining, and monitoring such a set up will obviously be more complex than just running either one, as you are running both client-side and server-side auctions independently and then selecting the highest value bid at the final step.

Which one should I choose?

As you can see, both header bidding and Open Bidding are compelling offerings with their unique set of pros and cons. Compared to the older waterfall method of ad serving, both represent a significant improvement in auction efficiency and performance.

However, which one will work the best for you depends on your individual needs and context. Header bidding is available in self-hosted, limited support, and fully-managed service models, has a vast network of demand partners, and drives immediate revenue lift. But it suffers from page latency issues, especially when timeouts and partner selection is not optimized.

On the surface, Open Bidding does the same thing as header bidding—making sure that the publisher receives the highest possible bid for every impression. But by moving the auction to the server-side, Open Bidding solves the latency issue, simplifies payments, and is easier to set up and manage. But only the largest publishers will have easy direct access to it.

Our recommendation: A/B test different auction configurations on limited parts of your inventory and make a decision based on the performance that you record during the tests.

Comments

Leave a Reply

Your email address will not be published. Required fields are marked *