- Header Bidding
PubMonkey: Our Own Header Bidding Setup Tool
- App Monetization
- Header Bidding
- iOS Ad Mediation
Header bidding technology is an essential tool in the publisher’s arsenal. Over 67% of publishers are using it for ad monetization in 2021. However, most ad auctions are full of inefficiencies that impact user experience and bidders’ performance. Thankfully, you can easily resolve these problems with header bidding analytics.
But how do you make sense of the various measurements to improve your ad revenue? First, you need to get to grips with some key definitions.
Analytics for header bidding allows publishers to track, assess, and evaluate the performance of the bidders (ad sources) on ad auctions.
Your main tools for analysis are third-party tools and reports from your ad partners. Publishers rely on various metrics to optimize inventory, improve content strategies, enhance user experience (UX), and prioritize the best ad sources. Here are some of the key indicators you need to make sense of the reports.
Analyzing header bidding performance requires understanding the field’s terms and metrics. So let’s go through the most critical indicators, and how they may help you identify problems with your setup.
When someone visits a page with available ad space, your header bidding system creates an auction with an impression opportunity for bidders. A bid response is generated by an ad source that makes a bid on the impression opportunity. The wrapper or server then receives and processes bids that participate in the auction.
The bid rate is the ratio of bid responses to bid requests. In other words, this indicator shows how often the advertiser responds to requests in your wrapper.
A bid rate depends on many factors, such as the ad source, bidding strategy, audience, and demand. However, a low rate can indicate configuration problems with your setup.
The win rate shows how often the ad source wins after responding to the bid request.
You shouldn’t rely solely on this metric to determine the bidder’s efficiency. For example, some ad sources get a higher win rate by ignoring most bid requests, impacting their overall performance.
The latency represents the time needed for an ad source to respond to a bid request. Bidders with high latency can slow down the entire auction and, respectively, page load time.
The timeout rate shows how often ad sources fail to respond to an auction bid within a required time limit.
As a rule of thumb, the most competitive bidders have low latency. Publishers tend to set the time limit between 1000 and 2000ms, while most ad sources return bids in under 600ms.
Even one partner can slow down the entire auction. Still, you shouldn’t cut off potential profits by setting your time limit too low.
The fill rate is the number of answered ad requests divided by the requests you sent.
This indicator allows you to see the percentage of sold ad displays. A low fill rate can mean a lack of demand for your inventory, incorrect settings (such as floor price or response time limits), or technical issues.
The match rate refers to the percentage of users shared among partners in the media trading chain.
Demand-side platforms (DSPs) and supply-side platforms (SSPs) rely on third-party cookies to sync data about end-users. These cookies are sent down the media chain (from publishers to the SSP and from the SSP to the DSP). However, the number of shared cookies is reduced every time trading partners share data. In short, a lower match rate means less potential revenue for you.
CPM (also known as eCPM or effective cost-per-mille) means revenue generated for 1000 sold ad impressions.
This metric doesn’t necessarily mean that the partner is profitable. After all, most of the bidder’s inventory can be left unsold despite a high CPM.
Revenue per auction provides return value through specific SSPs. When used with other metrics, this indicator can help you determine the most profitable partners.
Discrepancy reports show differences between your and your bidder’s reports. These mismatches are often found in volume and CPM data. The main reason for mismatches could be different reporting tools (and formats), time zones, and metrics (gross metrics instead of net metrics).
Header bidding analytics tools should support formats that are compatible with your existing analytics stack. At the very least, you should be able to filter these metrics by source, geolocation, channel, website, device type, etc.
Let’s look at how you can use the metrics above to enhance the performance of your header bidding wrapper or server setup.
The latency is directly linked to conversions. According to Deloitte’s 2019 Milliseconds make Millions study, latency is the main reason consumers abandon websites. If a page takes more than three seconds to load, the probability of a bounce increases by up to 90%.
Publishers should implement an A/B test framework to experiment with their setup. For example, you can move the client-side ad sources to the server-side and then measure the page performance, timeouts, CPMs, and revenue.
Additionally, you can use an HTTP/2 protocol for server-side solutions. This protocol can download files from a web server asynchronously, drastically improving the page load time.
A tight upper limit time (under 500ms) may end the auction before most bidders respond to a request. At the same time, lower time limits (above 1600ms) can affect UX, reducing your potential revenue. In other words, you should find the balance between profit and timeout.
Try setting your initial timeout at 1300ms. Then, see how this time limit affects your page load speed, number of ad responses, and generated revenue. You can also experiment with the refresh time limit to allow more ad sources to garner higher bids.
High-performing publishers often group ads based on their page position to increase bid rates. You can allocate bidders with short response time to the above-the-fold inventory, while slower partners can be placed in below-the-fold ad slots.
Third-party ad sources are set to compete based on Price Priority in Google Ad Manager (GAM) by default. This means the unsold inventory is filled based on the highest-paying line item. However, you can improve your fill rates by changing header bidding line item priorities for certain inventories.
To do this, publishers should study the CPM threshold (the point where the line item’s fill rate stops growing proportionally to its price tier). You can find this data in the GAM report or request a density report from SSPs.
Based on your findings, you can set the GAM priority for high CPM header bids to Standard and Sponsorship. This can improve the fill rates because your header bidding lines will compete with directly sold campaigns.
Get actionable insights for your product
Thank you for reaching out,
Make sure to check for details.
As we mentioned, the number of shared users is reduced every time cookies are sent down the media chain. A typical match rate of shared users averages 70% for each partner, and the reduction is compounded at every step.
An alternative approach would be to sync user data directly with your DSPs. As a result, you’ll improve the match rate and potential revenue from your inventory.
Publishers should use a tool that monitors and compares performance indicators for their partners.
For example, one of your ad sources experiences a drop in CPM, which is equally compensated for by increased performance for other partners. However, if there’s not a proportional improvement, there may be technical issues with your setup.
A unified reporting tool also helps pinpoint discrepancies between partners. Searching for inconsistencies in reports is a crucial part of analytics for header bidding. A reliable tool can help you resolve these mismatches to improve your setup.
Ad partners often focus on selected locations with the highest user demand. Publishers can filter which ad sources provide the best CPM, win rate, and revenue for specific regions or user segments. Then, you can prioritize these partners for specific locations to maximize their performance.
You should also determine if any ad sources tend to timeout in specific locations. Afterward, you can adjust time limits or consider moving them to the server-side instead of the client-side in these regions.
Advanced analytics adapter plugins can track bidder performance in real time and notify you if the setup is facing issues.
You can get automated alerts about abnormal response time delays, poor fill rates, and traffic drops. With the right software, you can adjust your setup quickly to avoid losing profit.
Proper analytics are essential for your ad performance. Improving one setting can give you a little performance boost, but optimizing your entire setup yields greater and more consistent results. You should ask SSPs and DSPs for data and support, monitor all metrics, implement ongoing A/B testing programs, and experiment with your parameters.
If you’re interested in header bidding, you should check our post about the pros and cons of this technology. We can also help you pick an efficient third-party analytics tool for header bidding or even help you develop a custom solution—just drop us a line!