“The Sell Sider” is a column written for the sell side of the digital media community.
Today's column is written by Paul T. Ryan, chief technology officer at OpenX.
I am hearing more publishers bring up Prebid when discussing their supply-side strategies. At a time when publishers are facing pressure from multiple fronts – evolving privacy standards, changes to cookie policies and COVID-19 – Prebid can give them control over their tech stack in a way no other tool in ad tech can today. And adopting the open source technology can increase publisher revenue upward of 50%.
Realizing all these benefits, however, doesn't necessarily come easy, which is part of the reason some publishers hesitate to get on board.
Setup and launch alone require a level of sophistication that falls outside the skill set of most publishers’ ad ops teams. It is a significant amount of work to choose configuration options, get the optimal number of bidders into Prebid, constantly manage new product features and updates, and optimize both within Prebid and how it works with other demand sources, such as Google Open Bidding and Amazon’s Transparent Ad Marketplace.
Given that a Prebid implementation is directly tied to publisher revenue, it must be done correctly. Optimization also takes time, resources and expertise, but for any publisher looking to get value from Prebid, there are three areas to address first.
Find the right configuration for the user experience
Prebid has a suite of products to choose from and a growing number of configuration parameters, so getting started can be daunting. The first decision a publisher needs to make is which product is best for them. Whether using client-side, server-side, mobile or a hybrid approach, there are several factors that should guide this decision. These may include a publisher’s users, page experience, expected revenue lift, timeouts or win/fill rates.
A common scenario use case would be to ensure an optimal user experience, either by reducing latency or getting creative with optimizing the number of bidders. For example, publishers with quick user sessions on page, which may include utility apps, such as a weather app, might choose Prebid to keep latency to a minimum. If someone will only be on site for a minute or two, speed is of the essence.
Or a publisher might leverage a hybrid model, where they can deploy more bidders split between the client side and server side. This allows it to retain top performers on the client side while keeping an additional source of revenue through the server side for bidders that would otherwise not be accessed.
Build a plan to manage release control and new capabilities
Keeping up with the Prebid release schedule can be especially tough as most ad ops teams have many other responsibilities. And, when you add in a growing list of new capabilities to explore, it’s difficult to know if you’re still taking advantage of all that Prebid has to offer.
One of the biggest challenges is that once a publisher builds its stack, adding its own “secret sauce,” so to speak, there’s often hesitation to make adjustments when something new comes along. But Prebid adds capabilities almost daily. If publishers have a set-it-and-forget-it strategy, they won’t get maximum value.
At the end of the day, publishers receive the most benefit from using the latest version available. So, while at times Prebid can seem like a Swiss army knife that’s too wide to use, publishers must come up with a plan to help keep it up to date.
Take advantage of all potential data, then test, test, test
Publishers who don’t regularly check performance and analytics are basically flying blind. Prebid analytics should provide insight into the performance of a variety of setups and are essential for optimization. Prebid analytics should support varying ongoing test and reference configurations so that performance optimization can be rapidly achieved.
The most reliable way is by using a Prebid analytics adapter, which is available from dozens of vendors and integrate with a publisher’s own data, providing a view of performance in one area. Once publishers have a handle on what data is available, basic A/B testing can spark ideas for how to improve performance. They should start with tests on a small percentage of inventory, where it’s relatively low risk, and when they see growth, they can start to expand it to all inventory.
An example of a basic test could be comparing performance of one day vs. the next or across different demand segments or combinations of demand sources. Publishers should conduct tests that get them closer to a setup that makes them more money and also brings in unique demand.
It takes a fair amount of effort to master all that Prebid has to offer. But in this difficult market, open source may be key to helping publishers maximize revenue in the long term – if they get it right.