This article is intended for publishers who use Google Publisher Tag (GPT) tags alongside Sulvo ad units (or any other ad units) to manage ad inventory on their websites. It explains why avoiding Single Request Architecture (SRA) is recommended for a specific setup that relies on a race condition to function correctly.
What is Single Request Architecture (SRA)?
SRA is a GPT request mode that combines requests for multiple ad slots (even from multiple ad providers) on a webpage into a single request. However, there are common situations where SRA can cause unintended consequences, impression loss, and revenue loss. This is especially likely if you have ads coming from multiple ad providers (such as GAM with Sulvo or any other ad provider simultaneously).
Why You Should Avoid SRA in Specific Cases
When using SRA (Single Request Architecture), GPT's display() function triggers an important timing behavior. The first time display() is called, GPT makes requests for all ad slots created up to that point but won't request any slots created on the page by any ad unit afterward. This timing behavior can significantly impact Sulvo's (or any other company’s ad tags) functionality since any ad slots created after the initial display() call won't be included in GPT's request.
The Solution
To avoid this issue, we recommend that publishers not use googletag.pubads().enableSingleRequest() while using ads from Sulvo or any other ad provider. This will prevent SRA from being automatically enabled and allow all ads to function as intended, regardless of the ad provider. Otherwise, all ad providers called on page will experience impression losses, resulting in unintended revenue losses.
Important Note
All sites with SRA architecture will experience impression loss. The only question is what percentage of sold impressions will be lost. On average, sites that abandon SRA and use the Sulvo ad serving system see an average 10-30% impression recovery rate.
We understand that navigating ad technology can be complex. If you have any questions about this recommendation or require further assistance, please don't hesitate to contact our support team.