A simple guide to prebid timeouts

Dive into the fast-paced world of programmatic advertising, where milliseconds influence your revenue. This guide simplifies the crucial concept of Prebid timeouts within Header Bidding using Prebid.js. Drawing from extensive research, we’ll explain their significance and offer strategies to balance between maximizing ad revenue and ensuring a smooth, enjoyable experience for your website visitors.

What are Prebid Timeouts?

Imagine, as a publisher, you’re hosting a programmatic auction where ads are the star of the show. When a visitor lands on your website, you kick off this auction. It’s not just any auction, though. It’s a speedy, high-stakes game where various SSPs have a limited window to place their bids for the opportunity to display their ads to your users. This is where the concept of “timeout” enters the stage.

A timeout in Prebid is essentially a countdown timer. It tells these SSPs how long they’ve got to throw their hat into the ring. If they don’t bid in time, they miss out, and the auction moves forward without them. The focal point of our discussion, the “Auction Timeout,” is critical because it determines how long this window stays open.

A Word on Our Research Methodology

Our study was a series of A/B tests of various Prebid Timeout settings conducted on four websites differing in structure (ranging from traditional news sites to specialized portals and web application-like platforms), various shares of Directly sold ad impressions(from 8% to 100%), and the share of Header Bidding (from 2% to 22%). This diversity provided a broad spectrum of insights into how different factors influence timeout optimization. It’s important to note we’re discussing client-side Prebid integrations here.

The Quest for the Perfect Timeout Setting

If you’re a publisher who has recently embraced Header Bidding technology through Prebid, you might still be operating with the default timeout settings – typically, an Auction Timeout of 1000 milliseconds (ms) and a Failsafe Timeout of 3000 ms. These settings are a solid starting point and might even hit the mark perfectly right off the bat for some publishers. Our extensive experience working with more than 1000 publishers has taught us that 1000 ms often serves as the minimum threshold for an optimal timeout setting. It’s rare to see the optimal setting to dip below this mark. 

The sweet spot for Prebid timeouts usually ranges between 1000 ms and 2500 ms. However, it’s not uncommon to encounter scenarios where a longer timeout is present, though such decisions should always be backed by extensive testing. 

An exciting aspect of Header Bidding auctions is that the actual wait time may not always reach the maximum set, such as 2500 ms. For instance, if the last SSP in the pool sends back a response and the auction concludes at 1500 ms, there’s no need to wait longer. Thus, the optimal timeout balances two critical factors: the potential for increased revenue from Header Bidding, thanks to SSPs having ample time to respond, and the risk of losing ad requests due to prolonged waiting periods.

Why Timing is Everything

Finding the right timeout setting is crucial for maintaining a fast-loading website while ensuring maximum ad revenue. By carefully increasing the Auction Timeout, you give SSPs a better chance to respond with their bids, potentially driving up your ad revenue by securing higher bid amounts. Yet, there’s a tipping point. Increase the timeout too much, and you might start seeing a drop in user engagement. Visitors are likely to abandon a site that takes too long to load, leading to fewer ad impressions and, consequently, lower ad revenue. The goal is to find a timeout duration that allows for competitive bidding among SSPs without negatively impacting the user experience.

Setting a longer timeout means you might lose some ad requests. This is because making users wait longer can lead to some of them leaving your site before ads finish loading.

This situation leads to a decrease in the number of ads available for display across all advertising channels, including Google AdX, AdSense, and direct ads. It’s important to note that for sites with a tiny share of revenue from Header Bidding (2-3%), the potential uplift in Prebid based revenue is very unlikely to offset the losses from missed ad requests on the entire ad stack. In such cases, if, despite increasing timeouts Header Bidding share remains marginal, it is crucial to investigate the reasons behind the low Prebid revenue share further and look deeper into such matters as limited SSP participation, low demand, inappropriate ad formats, or incorrect settings in Google Ad Manager (GAM), and attempt to improve these aspects. If losses still outweigh gains despite these efforts, discontinuing Header Bidding might be the most logical course of action.

The Fine Art of Timeout Optimization

Finding the right Auction Timeout is not about guesswork but testing. Some website owners employ A/B testing, which involves comparing two different timeout settings to see which one generates the most revenue at the end of the day. This approach allows you to make data-driven decisions that can significantly impact your ad earnings and is generally speaking the most accurate one.

Moreover, diving into your website’s analytics can provide a wealth of insights. Observing how changes in timeout settings affect your site’s traffic and ad revenue over time can guide you to the most effective configuration. Advanced analytics tools can offer a deeper dive, revealing how quickly each SSP typically responds and how different timeout settings impact your overall ad strategy.

Publishers who cannot employ A/B testing may try to select periods with stable traffic and user session data, without significant fluctuations, for testing – utilizing tools like Google Analytics to identify these optimal testing windows. This approach is much less accurate than A/B testing and in case of many publishers does not allow to get conclusive results due to performance fluctuations that result from different things than changed Timouts. 

Regardless of whether you employ A/B testing or period do period comparison having access to Prebid’s detailed reports helps a lot with this. It lets you see how each Bidder (SSP) bidding on your ads is doing, like how quickly they respond and how often they win auctions. This makes it easier to figure out how to get the most out of your ad inventory.

Impact of changing the timeout on your ad stack’s revenue

When you adjust the timeout settings, you might see your earnings go up a bit, but only to a certain limit, especially when your initial settings were on the lower end. It’s important to find that sweet spot where waiting a bit longer for ads to load helps you earn more without going too far. Also, think about how changing the timeout affects your total revenue, including both the ads you sell directly and those sold through the open market. Remember, finding the right timeout setting is a balance – it can help increase your income from ads, but there’s a point where making users wait longer doesn’t pay off anymore.

Revenue vs. Timeout Charts: Differentiating Between Sites With and Without Direct Sale

We looked into what might happen if we changed timeout settings for each ad unit, thinking it could generate additional revenue. To do this, we looked at the revenue generated by the best timeout (as determined on domain level) vs the revenue generated by the best timeout setting for each ad unit independently.. The results, unfortunately, were not super positive. In theory, doing this could make us a little more money, from nothing extra up to 1.8% more. But that’s just in theory. In the real world, based on what we’ve seen, the actual boost in money is usually at least 25% less than what such estimations show
Furthermore, adjusting timeouts settings per device resulted in even lower improvement – around 0.5%.

This small increase is a bit of a letdown because for publishers diving deep into the granularity of adjusting timeouts for individual ad units, the results, though positive, may not always justify the effort (and an increase in the script size to “communicate” to each ad unit the independently set up timeout setting). While each incremental improvement in ad revenue is valuable, publishers must weigh the operational complexities and resource investments against the relatively modest gains achievable through this approach.

Constancy of Optimal Timeouts Over Time

The behavior of optimal Prebid timeouts over time seems steady, constant and predictable. Despite expecting rapid changes in the programmatic world, the ideal timeouts for these auctions don’t vary much as time passes by. We’ve seen it hover around two similar values, with just a few unusual days here and there. Still, generally, it stays the same. This shows that for most publishers, it is completely enough to set an a/b test, determine the optimal timeout and forget about this setting for at least the next couple of months.

How Adjusting Timeouts Affects Ad Visibility and CTR

When we change timeouts for ads to bid in Prebid, it also slightly changes how often people see the ads (Viewability) and how often they click on them (CTR). Making the wait time a bit longer can make a slight difference in making ad impressions more visible by about 1 to 2 percent. But, when it comes to clicks, especially on websites that use a lot of programmatic ads and Header Bidding, increasing the wait time doesn’t really help. In fact, the more we rely on these types of ads and extend the wait time, the less likely people are to click on the ads. This happens because in the world of Header Bidding, the focus isn’t so much on getting immediate clicks. 

Mastering prebid timeouts with Prebid Stack

Our journey through the nuances of Prebid timeouts highlights the importance of adaptability and precision, a process greatly simplified by Prebid Stack. As the programmatic landscape shifts, Prebid Stack provides the tools you need for optimizing and managing prebid, including detailed timeout settings and an advanced A/B testing panel. Embrace the cycle of testing, learning, and iterating with the support of Prebid Stack, and navigate the path to sustainable growth in programmatic advertising.