One of the major benefits of header bidding versus standard waterfall is the promise of reduced discrepancies. Here are some ways to ensure that outcome.
One of the major benefits of header bidding versus standard waterfall is the promise of reduced discrepancies. In certain cases, however, discrepancies within an Auction in the Header setup arise. Often, these are due to technical implementation errors, and can be avoided. The following guide reviews the common errors, and what to check for within your implementation when troubleshooting.
Unlike traditional waterfall discrepancies, for Auction in the Header, one should compare the impressions for the DFP header order with the impressions for header bidding tags in //meridian. Compare impressions in DFP with impressions in //meridian.
The above diagram shows the process of recording impressions. DFP records an impression when the DFP creative is sent to the page. Sovrn records an impression after the DFP creative lands on the page and renders the Sovrn creative. This delay causes discrepancies. Reasons include, but are not limited to:
- The user navigates away from the page after DFP sends the creative, but before the creative is delivered.
- Network loss occurs between the Google server and the user’s browser.
- An ad blocking software prevents the line item from being delivered correctly.
Abnormally high discrepancies, however, could be caused by other factors. Avoid preventable discrepancy causing factors by reviewing those included within this document.
To troubleshoot discrepancies, make sure to gather the correct DFP numbers. Collect the header bidding impression data by pulling impressions for all the orders involved in header bidding for a specific time range.
Step 1 – In DFP navigate to the Reports tab in the top menu. Then, navigate to the Reports menu page in the left-hand side menu.
Step 2 – In the DFP Reports setting page: select date range (e.g. Yesterday), select main filter criteria (Choose Order * if header bidding is set up per partner), and select dimension (Order). Select the following typical metrics: Ad server impressions, Ad server average CPM
Step 3 – At the bottom of the report set-up page, click Run Report.
Your output on the dashboard results should look like the example below. You can also export in MS Excel (xlsx, xls) format.
To collect Sovrn’s impression data, log into your //meridian account and navigate to your earnings section. Select a custom date range to match your DFP data. Click on combined traffic to include all impressions.
Please note that the time zone of the DFP report may not match the time zone of the //meridian report. You can request a custom report from your Sovrn representative which includes data from the same time zone as your DFP report.
Header Complete and Header Code both use prebid.js at their core. It's important to understand that header complete partners share line items and DFP creative code. So, when running multiple partners through header complete, one must compare impressions from all partners with impressions in DFP for the order.
To troubleshoot discrepancies for a particular partner within your header complete setup, consider running that partner with different key-values in DFP. Your Sovrn representative can show you how to implement this test.
Referencing a supported and maintained CDN, such as Sovrn’s prebid source or AppNexus’s prebid source, reduces discrepancies. If at all possible, avoid building prebid from the source code and installing on the page. Among other reasons to avoid this practice, we’ve seen that this can often result in discrepancies.
We’ve include the Sovrn prebid library below for your benefit:
Additionally, a Sovrn tag id can not be included more than once within a request. Check that each Sovrn tag ID is included only once. Requesting a tag multiple times will result in discrepancies.
Finally, prebid community members recommend this loading pattern for reducing discrepancies. If you are not using this loading pattern, a Sovrn support team member can help with the update.
Custom/API Header Bidding Discrepancies
When using a custom wrapper for header bidding, its important to understand the Sovrn response object.
Ensure that both the nurl (the tracking pixel) and the adm (the creative code) are fired at the exact same time.
We’ve included sample code below for your benefit:
(Note that the object ‘sovrnBid’ contains the Sovrn response object)
These must be fired at the same time for the impression to be tracked in Sovrn’s system.
Advanced Considerations for Header Bidding Discrepancies
If your set up uses advanced monetization techniques, such as holding bids, constant refreshes, or price floor optimization techniques, please let your Sovrn representative know. These techniques can likely be supported with a tailored Sovrn solution.
If your set up uses advanced targeting or customization within DFP, please let your Sovrn representative know. The DFP reporting techniques can likely be verified by your Sovrn representative.
Q: I am seeing impressions in DFP, but I am seeing no requests/impressions in //meridian.
A: If you are seeing no requests at all in //meridian, but seeing impressions in DFP, it is likely that the DFP creative code you are using is incorrect. Reach out to Sovrn support to verify both the DFP creative code and the DFP reporting method.
If you are seeing requests in //Meridian, but no impressions, verify the following:
- The domain is whitelisted with Sovrn
- The Sovrn tracking pixel is firing
- The creative code in DFP is correct
Q: I have pulled a report in DFP with Sovrn’s creative and key-value pair, but I am finding many fewer impressions in //meridian.
A: These discrepancies can commonly be attributed to:
- The reporting method used in DFP
- The reporting method used in //meridian
- Multiple partners sharing line items
- Errors with custom technical methods
- Errors with DFP setup
Of course, if you're still unsure about your reporting discrepancy, please reach out to your Sovrn Publisher Support.