How did Adwords change bidding at the end of 2023? (performance analysis)
What is AdWords? Google AdWords, now known as Google Ads, is an online advertising platform by Google. It allows businesses…
After many years of struggle with the growing popularity of header-bidding solutions, Google rolled out EBDA (currently known as Open Bidding). Its own technology established in Ad Manager.
Open Bidding is intended to make it easier for publishers to increase competition and demand sources for their inventory inside their ad server. All this without the need to increase page latency (from additional scripts on the head sections of their websites).
In the beginning, when this product was still limited to users of Ad Manager 360, we at Yieldbird decided to use it on our publishers’ inventory. We wanted to find out if it could be as good as its regular ad stacks (as Google advertised). Our evaluation can be summarized in five key points related to exchange-bidding performance.
[mc4wp_form id=”407″]
This factor works in favor of Open Bidding in comparison to the classic client-side header bidding solutions. Adding it to the current setup requires little additional work on the client’s end in terms of web development. However, only in the case of publishers who already use Ad Manager as their ad server.
All publishers must do is create new IDs that will be connected to the platforms (which will be added to the yield groups) in the Ad Manager. Update their Ads.txt file with those IDs – and this can be almost completely done by Ad Ops. One unnecessary complication is that most platforms, like OpenX, will require you to open an additional account with them to run EBDA. Even if you already have one dedicated to regular header bidding or tag-level monetization.
While considering the variety of demand sources, Open Bidding pales in comparison to the old way of doing things. At first, one could think 14 platforms available for implementation should be more than enough. However, deeper analysis and tests in the European market showed that the relevant scale of demand can actually come from only four to five of those SSPs. The lack of such key players as AppNexus, Adform, or Smart makes us more reliant on maintaining the setup where exchange bidding runs alongside its older sibling, especially since the platforms being present on EBDA and client-side header bidding at the same time tends to keep them on the same level to maintain the best possible performance.
One of the biggest issues publishers faced regarding the use of client-side header bidding was the rather low bid and win rate of non-Google SSPs outside of North America. In the end, it resulted in small-scale revenue coming from outside of AdX. It could be related to the fact that Google’s platform was in a privileged position of knowing the rate it had to outbid to take the impression it was after. Even more likely, related to latency on the page level, or very short timeouts that limited the specific SSP’s potential.
Fortunately, exchange bidding removes all of those issues. There is a 60–80% increase in the non-AdX revenue share and a much healthier diversification of income sources. Simultaneously, we have noticed that the eCPMs on specific platforms in EBDA tend to be lower than when using them in prebid. The only solution here is the application of specific hard floors. However, some platforms do not give you this opportunity on the exchange-bidding level (such as Sovrn).
After the simple implementation process of exchange bidding, publishers and yield managers usually expect the same user-friendly experience with the daily management of platforms running this technology, reporting performance and key metrics, and acquiring final earnings. Unfortunately, not all of those actions are hassle-free.
The easiest case, which might sound a little unusual for those who have worked in the programmatic business, is remuneration. Google (for its small share) collects money from other SSPs and sends it to publishers in one combined money transfer. Reporting performance is not much more complicated, as the report is available in Ad Manager’s historical reports, with most of the metrics being the same as other parts of the setup (apart from some irregularities, like naming the requests to Open Bidding platforms).
The biggest trick comes with daily management, and specifically, flooring the platforms and blocking specific advertisers on their level. Both actions are not available on some of the compliant SSPs, which is inconvenient already, but even trickier is that advertiser blocking only works for the whole account on most platforms.
This could be problematic if you, for example, have many different websites and want to block the specific URL of the competition for only some of them. Regarding floor prices for EBDA, Google recently rolled out a Beta of its so-called Unified Pricing Mechanism, which will give some options to publishers that do not want to sell inventory for their buyers too cheap via EBDA. For now, granularity and differentiation of those floors per specific platform are basically nonexistent.
Taking into consideration all of the above factors, it is only right to end this summary with the evaluation of the most important factor. Impact on inventory-revenue performance.
Yieldbird analysts were testing Open Bidding for several months on the inventory of more than 30 publishers. What they established is that using Open Bidding along with other elements of the advertising setup positively impacts publishers on the total rCPM level of their inventory on the scale of 7% on average.
This is probably a little less than one might expect after Google’s emphatic announcements regarding this product’s impact. On the other hand, considering the simplicity of its implementation and benefits from reporting and remuneration with SSPs, we can easily say it is worth the effort. Especially if there will be further improvements in the mechanism and further additions on the list of Open Bidding compliant platforms.
Karol Jurga
Chief Revenue Officer
See it in action.