As 2023 progresses, the programmatic advertising industry is becoming increasingly competitive, complex and challenging. Publishers are under pressure to optimize their monetization strategies and reduce risks. At the same time, adopting new technologies and optimization in the prebid setup can be tricky: this is why testing any optimization to your ad stack before implementation is of the utmost importance. Pubstack launched new A/B testing capabilities in order to help publishers to take the good decisions and boost their revenues.
A/B testing (also known as bucket testing, split-run testing, or split testing) is a research methodology. A/B testing is an online experiment conducted on a website, mobile application or advertisement (and more) to test potential improvements over an original version. In another word, A/B testing is a randomized experiment with two variants: A and B (although the concept can be also extended to more than two variants). Most of the time, the test is divided between a large part of the population (~90%) as reference, and a smaller one (~10%) which will experiment with a new feature, user journey, design…
What about A/B testing for the Ad Tech industry and, particularly, for header bidding? Header bidding is the most effective solution publishers can use in 2023 to monetize their websites. Publishers are often willing to run A/B tests in order to acknowledge how valuable and reliable a change would be in their stacks, and to assess the holistic impact on revenue.
Let’s take the example of a publisher willing to run a new version of Prebid: before switching all their stacks on the new version and risking losing all their revenue because of technical issues, they may want to test it on a smaller part of their users first. To do so, the publisher will implement on his end a randomized traffic split on a specific scope (website, ad unit…), where population A will run their actual version of prebid, and population B the new version they wish to set up. Both populations will run on the same scope to make sure that the publisher will get precise results from the test.
Basically, almost any element of a programmatic stack can be A/B tested. Most common tests include:
On the topic of floor prices in particular, publishers often struggle in finding the optimal floor price value based on their business objectives. The best way to proceed to find the optimal value for floor pricing is to simply A/B test different floor prices and carefully monitor the relevant KPIs.
Nonetheless, A/B tests can be essential in order to find the best technologies and partners for your ad stack. Today we have many header bidding partners to choose from: loads of SSPs (Open Exchanges, specialized SSPs, bidders, etc.), multiple wrappers to connect with, and many ID providers, among others. A/B Testing against your best partners could be a good solution to understand the top partners you should be working with, you could potentially choose to pick partners giving the highest CPM or bringing the best website performances.
With the new A/B testing capabilities on Pubstack’s dashboard, publishers can deploy new Prebid features in a safe manner, and start by only pushing changes to a small percentage of their inventory. We have recently released an A/B test dedicated overview in Pubstack's “setting” that enables publishers to monitor their A/B tests with complete autonomy, by relying on real-time data. Basically, Pubstack will first guide the clients into implementing A/B tests on their websites; the publishers will then be able to monitor the performance directly on the Pubstack’s dashboard: publishers will therefore measure the impact on monetization performance of even the smallest change to their set-up.
Keep in mind that each test may require different timeframes" : i.e. 3 weeks for floor prices, few days for prebid version..
Furthermore, we added the A/B test dimension to the following dashboards’ breakdowns:
At the same time, for clients that have activated the brand new GAM Dashboard, they will also find an “A/B test” dimension (in the Group By).
Last but not least, Pubstack’s Data Analyst Team is able to generate custom reports to monitor and analyze customers’ A/B test: with this new feature, we empower publishers to directly control & evaluate their A/B tests, while supporting their monetization strategy by providing relevant insights and quick, reliable solutions.
We built a dedicated, self-service page for A/B test on the Pubstack Dashboard as we aim to empower publishers and give them the full control of their monetization strategies and monitoring. With Pubstack you can:
Publishers can now have a complete overview of their A/B tests, by assessing the overall impact of any change to their prebid setup. Therefore, Pubstack’s users are now able to test new solutions and technologies with full confidence and better understanding of the results.
The testing of the current setup is critical to the success of header bidding for publishers, and the changes that are done while testing may help you to better understand your monetization opportunities and get better results.
If you'd like to get some assistance on your programmatic project or would just like to discuss how you can improve your monetization performance, feel free to book a meeting with one of our programmatic experts by clicking the button below.
If you'd like to know even more about A/B testing and how it can take your monetization performance to the next level, we highly encourage you to read our dedicated whitepaper on "How to run a proper A/B test". This document will go over the different possible pitfalls behind an A/B test and will make sure that none of your data-based decisions are taken in biased way ever again.
Finally, check out our Customer Use Case with Le Monde to discover how the French publisher has been using Pubstack to obtain valuable insights on the effects of the slightest changes to their setup, and to implement a process of continuous improvement of their ad stack performance.
.png)
%20(2).png)
