How to test a new bidder in prebid.js with Prebid Analytics by Roxot
Publishers implement header bidding to increase competition for their inventory and ultimately yield more with higher CPMs and fill rate. Theoretically, the more bidders you have in the header, the more revenue your site generates. In reality, however, technical limitations and difficulties with identifying sources of shady demand complicate the maintenance and optimization of your setup. Every additional bidder has to successfully pass numerous publisher's requirements and demonstrate strong performance to stay in the header. Prebid Analytics by Roxot is a must-have prebid.js tool that allows you to conduct transparent tests of new demand partners and make sure you have reliable data to make business decisions.
How to test a new bidder in prebid.js?
First things first. When implemented a new bidder on your site, make sure you set it up correctly. Otherwise, you would make decisions that affect your bottom line based on questionable data. Also, inform your account manager about completing the implementation. It's an important step as additional work on the partner's side might be required for an adapter to start working correctly.
With Prebid Analytics by Roxot, the process of testing a new bidder is straight-forward.
When the system detects events coming from a new bidder, it will appear in bidder widgets on all dashboards and in the main menu under the Bidders tab. Go to the Total Dashboard, scroll down to the Total Bidders Performance widget and check if the bidder appeared in the table.
Mind that the system refreshes every hour. It might take some time for a bidder to appear in the tool. If you can't find the new bidder neither on the Total Dashboard nor in the main menu after an hour, double check your set up.
When the bidder appeared in the system, quickly audit its performance in the Total Bidders Performance widget. Check Requests, Bid Rate, and Impressions. These metrics allow you to see if ad requests are properly sent to the bidder, how the bidder responds to these requests, and if any of ads it responded with got rendered.
Before starting testing new bidder's performance, decide on rules of the game:
  • How long will you test your new bidder for?
  • How often will you audit its performance? We recommend auditing your new partner and prebid.js setup daily. This way you will be able to avoid losing revenue in case your new bidder causes an increase in latency, has massive discrepancies, or crashes your prebid.js.
  • How will you measure bidder's success?
  • What should be your bidder order in prebid.js? We recommend setting up random bidder order to minimize biases toward certain bidders and to ensure the results of tests are clean.
  • Etc.
After you ensured the bidder is set up correctly and the rules for the test are determined, you are ready to start analyzing bidder's performance. Follow the steps below to analyze how your new demand partner responds to requests, competes with other demand sources, and how much additional revenue it generates for your company. How often you audit new bidder's performance is up to you - analyze the performance daily, weekly or even monthly.
Check how bidder's performance affects overall prebid.js results by comparing your current site's performance with the average performance before the implementation. Start with the Site Dashboard to check how Fill Rate and Revenue have changed for the site. A new partner ultimately should increase your overall revenue. It can be accomplished by either filling requests that other bidders can't fill or offering higher prices for users all partners bid on. If the Fill Rate and/or Revenue haven't grown in comparison with the previous period, your new bidder isn't contributing enough to your prebid.js performance. Dig deeper to understand what the issue might be.
Scroll down to the Bidders Performance widget and compare a new bidder with other demand partners. Make sure it gets the same amount of requests as others. If a bidder has significantly less requests, check your setup - you could forget to implement it on some of the ad units or make a mistake in the config.
Switch to the Bidder Dashboard and analyze your new demand partner in more details. Start with understanding how the bidder responds to ad requests. It would help you identify potential issues and the direction of further investigation if necessary. The Response Types Part-to-Whole widget is a great visual representation of all bidder's responses spread in time. Each bar represents 100% of all responses for a particular hour, so you can quickly compare how often a demand partner returns a bid, $0 bid (no bid), and/or is timed out.
If your new bidder's Timeout Rate is acceptable but Bid Rate is low and No Bid Rate is, in turn, high, the partner might not have demand for your inventory. Scroll down to the Performance by Geo and Performance by Device widgets to analyze if this is the case. Sort the Performance by Geo table by the number of Requests by clicking on the column name. Compare bid rates for the top countries your users are from. If the Bid Rate is low for them, the demand partner might not have ads for you main audiences.
If the Bid Rate is acceptable, scroll down to the Performance by Device widget and proceed with the same steps: sort by requests and compare bid rates for different devices. If you see your new bidder doesn't bid often enough for mobile traffic, consider removing it from your mobile ad units. It may improve your site's latency as a browser won't have to wait for a response that is unlikely to come.
When analyzing bidder's responses, you may face high Timeout Rates which mean that prebid.js often doesn't receive a response from the demand partner in time set in your prebid timeout settings. In this case, contact your account manager to work together on troubleshooting the problem.
After learning how your new bidder responds to ad requests, analyze how it competes with other bidders. Go to the Site Dashboard and scroll down to the Bidders Performance widget. The Win Rate column shows a percentage of auctions where a bidder had the highest bid from auctions a bidder participated in - that's how many times a demand partner's bid was the highest among other bids. Click on the column name to sort your bidders by Win Rate. If your new bidder is one of the least competitive partners, it has a low potential to significantly improve your header bidding performance. Although, it may be used to fill the unfilled by other bidders inventory if its bid rate is close to 100%.
You have analyzed how your new bidder responds to ad requests, how it competes in your header, and how much value it adds to the prebid.js performance. Follow these straight-forward steps with Prebid Analytics to daily analyze both new and exisiting bidders and make data-driven decisions on header bidding optimizations.
The first prebid analytics tool for more efficient ad operations
Made on